Etymological Dictionary of the Hittite Inherited Lexicon

Edited by: Alwin Kloekhorst

Hittite is the oldest attested Indo-European language and therefore of paramount importance for comparative Indo-European linguistics. Although in the last few decades our knowledge of the synchronic and historical linguistics of Hittite has profoundly increased, these new insights have not been systematically applied to the whole Hittite material. This book fills this gap by, for the first time, providing an etymological dictionary of the entire Hittite lexicon of Indo-European origin in which all words are treated in a coherent way. Furthermore, it provides a thorough description of the synchronic phonological system of Hittite as well as a comprehensive study of the Hittite historical morphology and phonology. The result is a monumental handbook that will form an indispensable tool for Indo-Europeanists and Hittitologists alike.

See for more information on print edition.

Purchase Access
General Introduction
Introduction to the Dictionary
Towards a Hittite Historical Grammar
Hittite Historical Phonology
Aspects of Historical Morphology
Alwin Kloekhorst Ph.D. (2007) in Comparative Indo-European Linguistics, Leiden University, is a collaborator of the Leiden Indo-European Etymological Dictionary Project. His research focuses on the synchronic and historical linguistics as well as philological aspects of Hittite and the other Anatolian languages.

Etymological Dictionary of the Hittite Inherited Lexicon

Introduction to the Dictionary

In this part I will treat the following Hittite words: (a) words that in my view have a good IE etymology; (b) words that formally look as if they could well be of IE origin but for which no IE cognates are known; (c) words for which an IE etymology has been proposed that in my view cannot be correct;[1] and (d) words of which I found it important to show that they must be of a foreign origin.[2]

Of each word I have cited all spellings of the forms as attested, giving attestation place if necessary and a dating. If a certain form is attested in multiple texts, I have only indicated the oldest dating. Whenever needed, I have ordered the forms and spellings chronologically, in order to describe the most original state of affairs. On the basis of this material, I have treated the etymology of each word. Apart from words, I have also included in this dictionary the nominal and verbal endings, as well as most of the verbal and some nominal suffixes. Of these I have especially concentrated on morphological changes, as well as on their etymology.

An etymological dictionary can only be written on the basis of good philological descriptions of the words in question. In the case of Hittite, such descriptions are not available for the entire lexicon. The only dictionary that comprises the whole Hittite vocabulary[3] is Friedrich’s Hethitisches Wörterbuch (HW) that dates from 1952-1954, to which three Ergänzungshefte (1957, 1961, 1966) were added. Although this dictionary must be regarded as a milestone in Hittitology, it is nowadays outdated in some respects: it does not give examples of contexts to illustrate a word’s meaning, it cites forms in bound transcription, often disregarding plene spellings, and it does not give attestation places to all forms cited. Moreover, many more Hittite texts have been published since it appeared, which means that the dictionary is not exhaustive. Fortunately, other dictionary projects have been started in more recent times that do meet up to the expectations of modern-day Hittitologists. Yet, these are all unfinished. FriedrichKammenhuber’s Hethitisches Wörterbuch. Zweite, völlig neubearbeitete Auflage auf der Grundlage der edierten hethitischen Texte (HW2), which was initiated in 1975, sofar comprises a- till Ÿaššu- and gives a complete overview of attestations and an extensive semantic treatment (but note that its dating of texts does not follow the communis opinio). In preparation to this work Kammenhuber has published Materialen zu einem hethitischen Thesaurus (1973 ‑ 1989) that treats the lemmata āk-i / akk- ‘to die’, šu (conjunction), ta (conjunction), ‑a- (encl. pers. pron.), eku-zi / aku- ‘to drink’, dā-i / d- ‘to take’, šākk-i / šakk- ‘to know’ and Ÿandae-zi ‘to arrange’. It contains many attestation places and a detailed semantic description. The Hittite Dictionary of the Oriental Institute of the University of Chicago (CHD), edited by Güterbock, Hoffner and Van den Hout, first appeared in 1989 and thus far treats lā-i / l- till =šše. It gives many attestation places and a full semantic treatment, too. Two other works, which are not primarily meant as synchronic dictionaries, do provide philological information as well. Puhvel’s Hittite Etymological Dictionary (HED) first appeared in 1984, and up to now has been finished for the letters A M. It gives many (often all) attestation places and examples of contexts to illustrate the semantics of a word. Tischler’s Hethitisches etymologisches Glossar (HEG) falls into two parts. The first part (1977-1983), dealing with the letters A K, is a mere bibliographical work giving references to etymological treatments of the words cited. The second part (1990‑), for which Tischler has received the help of Neumann and Neu, thus far treats the letters L, M, N, P, Ša and T, and gives more philological information (although still not extensive), including forms that are attested on unpublished tablets. A small contribution was Otten’s Materialen zum hethitischen Lexikon (1971b = StBoT 15), in which he extensively treats the words beginning with zu‑.

All in all, good up-to-date philological treatments exist of the following part of the Hittite lexicon: A, E, î, I, K, L, M, N, P, Ša, T (but not as extensive as desired) and Zu. This means that the words beginning with Še Šu, U, ^ and Za Zi often still lack an extensive synchronic description. In my etymological treatment of the Hittite inherited lexicon, this means that for the words of the first category I often only refer to the works cited above for the synchronic treatment (unless I disagree, of course), whereas for words of the second category I give much more synchronic philological information, including attestation places, contexts and semantics. Since I do not have a card-tray system at my disposal that covers all published Hittite texts, I cannot claim exhaustiveness for these treatments. Nevertheless, on the basis of many treatments of texts and words in the secondary literature and using a collection of computerized transliterations of some 3300 Hittite texts (containing ca 280.000 words)[4], which has greatly enhanced the search for forms, attestation places and contexts, I have tried to be as inclusive as possible.

Each lemma is accompanied by grammatical information (the classification of the verbal system is elaborately treated in chapter 2.3), a translation, its corresponding sumerogram and/or akkadogram (if applicable), all attested spellings known to me (which are dated when relevant: note that if a certain form is attested in texts from different periods, usually only the oldest dating is mentioned), inner-Hittite derivatives and cognates, cognates in the other Anatolian languages, a Proto-Anatolian reconstruction (if possible), cognates in the non-Anatolian Indo-European languages, a Proto-Indo-European reconstruction (if possible), and, finally, an elaborate philological and etymological discussion.

In the treatment of cognates from the other Anatolian languages, I have tried to include all attested forms, for which I have used the following sources: for Palaic, the vocabulary in Carruba 1970; for CLuwian Melchert’s Cuneiform Luvian Lexicon (1993a); for HLuwian I have cited words as transliterated in Hawkins’ Corpus of Hieroglyphic Luwian Inscriptions (2000); for Lycian I have used Melchert’s A Dictionary of the Lycian Language (2004a); and for Lydian Gusmani’s Lydisches Wörterbuch (1964).

Note that in alphabetization the sequence -u¾a- is regarded as -u- + -¾- + -a-, so e.g. du¾arni-zi / du¾arn- follows MUNUSduttari¤ata/i- and precedes tuzzi-.

[1] Although I certainly do not claim exhaustiveness for this category.

[2] I am aware that this latter category is quite arbitrary.

[3] Sturtevant’s A Hittite Glossary (1931, second edition 1936), Tischler’s Hethitisch-Deutsches Wörterverzeichnis (1982) and Tischler’s Hethitisches Handwörterbuch (2001) are all mere glossaries: they only cite the stem of a Hittite word with its translation without giving (much) linguistic information.

[4] Kindly provided to me by prof. Tischler, for which I am very grateful.

Etymological Dictionary of the Hittite Inherited Lexicon


Etymological Dictionary of the Hittite Inherited Lexicon

General Introduction


0.1 Aim of this work


Since Hittite is the oldest attested Indo-European language, it is of prime interest for anyone involved in comparative Indo-European linguistics. A thorough description of the historical phonology of Hittite is therefore of paramount importance. In my view, one cannot describe the historical phonology of a certain language without having etymologically treated the entire inherited vocabulary of that language nor without having a coherent view on the morphological changes that have occurred in it. Moreover, in the case of Hittite, it is essential to distinguish between the three chronological stages it displays (Old, Middle and Neo-Hittite, cf. § 0.3) and the changes that took place between these stages. In all recent handbooks dealing with the etymology of Hittite, this crucial combination is lacking: Melchert’s Anatolian Historical Phonology (1994a) and Kimball’s Hittite Historical Phonology (1999) both focused on the historical phonology only, not always taking into account all relevant material, whereas Puhvel’s Hittite Etymological Dictionary (HED) and Tischler’s Hethitisches etymologisches Glossar (HEG) have no coherent view on the historical phonology in their treatment of the lexicon. Furthermore, most scholars do not seem to differentiate between orthography, phonetics and phonology, which in my view is a decisive part of the understanding of the Hittite language. I therefore felt it my task to write a historical phonology of Hittite on the basis of an extensive treatment of the Hittite inherited lexicon,[1] in which not only phonological change, but also morphological change and inner-Hittite chronology are taken into account.

Besides this introduction, in which I will give general information on the Hittites and their language, the Anatolian language branch and the place that this branch occupies within the Indo-European language family, this book consists of two parts. The first part is called Towards a Hittite Historical Grammar and contains two chapters: chapter 1, Historical Phonology, contains a detailed discussion of the phonetic and phonological interpretation of Hittite orthography, as well as an overview of the sound laws that took place between the reconstructed PIE mother language and Hittite as it is attested; chapter 2, Aspects of Historical Morphology, contains an overview of the Hittite nominal system, a treatment of the prehistory of
the Hittite personal pronouns as well as an elaborate morphological interpretation of the Hittite verbal system. The second part is called Etymological Dictionary of the Hittite Inherited Lexicon and contains etymological treatments of all the relevant Hittite words.

These two parts cannot exist without each other. The sound laws and morphological interpretations described in Part One are illustrated by the material from Part Two, whereas the treatment of the words in Part Two rests heavily on the findings of Part One. The reader should be aware of this when consulting one of them.


0.2 The study of Hittite


The rediscovery of the Hittite language actually started in 1887 when in the Egyptian village Amarna 382 clay tablets inscribed in the cuneiform script were discovered containing letters from and to the Egyptian pharaoh Akhenaten (ca. 1352-1336 BC). Although most of these letters were written in Akkadian, a language that at that time was already well understood, two letters, addressed to the king of Arzawa, turned out to contain a hitherto unknown language. In 1902, the Dane J.A. Knudtzon elaborately treated these two letters and even claimed that their language belonged to the Indo-European language family. In absence of any positive reactions to this claim, he retracted his views in 1915 (Knudtzon 1915: 1074).

When in 1905 the imposing ruins of an ancient city near the little Turkish village Boğazköy (presently called Boğazkale) was first excavated, soon tens of thousands of (fragments of) clay tablets were unearthed. Many of these were written in Akkadian, which made clear that the tablets constituted the royal archive of the land îatti[2] and that the site in fact was its capital îattuša. The bulk of the tablets were written in a language identical to the language of the two Arzawa-letters, however, which now was coined ‘Hittite’. Just ten years later, the Czech assyriologist Bedřich Hrozný published a preliminary “Lösung des hethitischen Problems” (Hrozný 1915), followed by a full description of “die Sprache der Hethither” (Hrozný 1917), probably the most complete decipherment of a language ever written. The results were baffling: Hrozný showed beyond any doubt that Hittite belongs to the Indo-European language family. Therewith Hittite immediately became the oldest attested language within that family. Not only did Hrozný’s decipherment open up a new academic field, Hittitology, it also brought a fully new aspect to comparative Indo-European linguistics.

As is now known, the oldest evidence for the Hittite language can be found in Old Assyrian texts (ca. 1920-1850 BC) that stem from the Assyrian trade colony or kārum established in the town Neša / Kaniš (= modern-day Kültepe). In these texts we find many Hittite names and some loanwords that clearly show that Neša / Kaniš was a Hittite town during that period. It may well have been the most important Hittite city at that time. Not only does the oldest Hittite text, the so-called Anitta-text (Neu 1974a = StBoT 18), which must be regarded as an account of the foundation of the Hittite royal dynasty, tell how Anitta, son of PitŸana, the king of Kuššara, conquers Neša and from then onwards uses this city as its residence, the Hittites themselves also refer to their language as URUni-ši-li (IBoT 1.36 iii 64), na-a-ši-li (KBo 5.11 i 3), ne-eš‑[u]m!-ni‑[li] (VBoT 2, 24) and ka-ni-šu-um-ni-l[i] (KUB 41.14, 8), i.e. ‘in Nešite, in the language of the people of Neša / Kaniš’. Nevertheless, two centuries later, around 1650 BC, the Nešites apparently chose the city îattuša as their new capital (possibly because of its more strategic position) and the surrounding land îatti (URUîa-at-ti, the region enclosed by the Kızıl Irmak) as their heartland. From that moment onwards their rulers call themselves LUGAL KUR URUîa-at-ti ‘king of the land îatti’, which ultimately is the source of our term ‘Hittite’. The Hittite kingdom grew rapidly and in the 14th century it reached its peak, ruling over vast parts of Anatolia and northern Syria (then also called ‘Hittite Empire’). From ca 1250 BC onwards rivalry between two branches of the royal family both claiming to be the legitimate heirs to the throne caused it to descend into civil war and around ca 1175 BC the Hittite kingdom ceased to exist.

Throughout its existence, the Hittite kingdom used Hittite as its administrative language (although Akkadian was used as well, especially for international affairs). The bulk of the Hittite texts (some 30.000 pieces) therefore were found in the royal archive at îattuša, but important finds have been made in Ugarit / Ras Shamra (some 200 pieces), Maşat Höyük (116 pieces), Kuşaklı (48 pieces) and Ortaköy (allegedly some 3500 pieces, of which only a handful have been published up to now) as well. Although most of the Hittite texts deal with religious affairs (ritual texts, hymns, prayers, festival descriptions, omens, oracles, mythological texts), we also find historical texts (annals), political texts (treaties with vassal kings, letters), administrative texts (instructions for functionaries) and legal texts.


0.3 Dating of texts


The Hittite texts span the whole period of the Hittite kingdom (ca. 1650-1175 BC). Already in the 1930’s it had become clear that texts that had to be attributed to kings from the beginning of this period showed linguistic features that were different from texts that had to be attributed to kings from the end of this period: the language was changing throughout the Hittite period (like any living language is changing through time). It therefore nowadays has become general practice to divide the Hittite language into three successive linguistic stages: Old Hittite (OH, ca 1650-1450), Middle Hittite (MH, ca 1450-1380) and Neo-Hittite (NH, ca 1380-1175). From the 1960’s onwards it has become clear that also the cuneiform script in which the Hittite tablets were written underwent changes: the shape of some signs has been altered drastically within the Hittite period. We therefore nowadays also distinguish between three palaeographic stages: Old Script (OS), Middle Script (MS) and Neo-Script (NS).[3] Since it is well known that within the Hittite archival system older texts were often copied in younger periods and that the scribes who carried out the copying did not always refrain from modernizing the language of the older original according to their own standards, any scholar who wants to be seriously involved in Hittite linguistics must date a given text according to these two criteria: composition and script. For the dating of compositions I have mainly used the lists of Melchert (1977: 45-131), Oettinger (1979a: 573-580) and Weitenberg (1984: 13-21) as well as the datings used in CHD. In this book I have adopted the practice of e.g. CHD to use the following abbreviations: OH/NS = a Neo-Hittite copy of an Old Hittite composition; OH/MS = a Middle Hittite copy of an Old Hittite composition; etc.[4] It has to be borne in mind that a given form from an OH/NS text cannot be immediately identified as a linguistically old form: it is quite possible that the text has been ‘polluted’ by the NH scribe and that the form in question in fact is a modernization.


0.4 Methods of transcription


In this book, I use four different layers of representing Hittite words: (a) a one-to-one transliteration of the cuneiform signs,[5] e.g. ne-e-pí-iš; (b) a bound transcription,[6] e.g. nēpiš; (c) a phonological transcription, e.g. /nébis/; and (d) a phonetic transcription, e.g. [népis].


0.5 Anatolian language family


Linguistically, Hittite belongs to the Anatolian language family, which further consists of the following languages:

Palaic was the language of the land Palā (URUPa-la-a),[7] probably situated in north-west Anatolia between the rivers Sakarya Nehri / Sangarios and the Kızıl Irmak / Halys (possibly identical to classical Paphlagonia or one of its regions, Blaene). Reference to this land is found in the OH version of the Hittite Laws only, which indicates that in MH times it had ceased to exist as such. The language, which the Hittites call URUpa-la-um-ni-li, i.e. “in the language of the Palaic people”, is only known from a few Boğazköy-tablets that deal with the cult of the god Zapar¾a. These texts were composed in the OH period, and therefore it is generally agreed upon that the Palaic language had died out by MH times. Its corpus is very small, and therefore many basic matters regarding grammar and lexicon are unclear. For texts, grammar, vocabulary and historical phonology, see e.g. Carruba 1970, Carruba 1972, Kammenhuber 1969, Melchert 1994a: 190-228.

Cuneiform Luwian is only known from passages of incantation and cult songs cited in Hittite texts dealing with rituals and festivals with a Luwian background. In these texts, which date from the 16th-15th century BC, the language is referred to as lu-ú-i-li, i.e. “in Luwian”. The language is closely connected with Hieroglyphic Luwian (see below). In Hittite texts from the NH period we find many words, often preceded by the gloss-wedges É and µ, which have to be regarded as Luwian borrowings (although it is not always clear whether these words derive from CLuwian or HLuwian). Although in the OH versions of the Hittite Laws the land Lu¾i¤a or Lū¤a is attested (URULu¾i¤a), which has to be equated with the land Arza¾a as attested in younger texts and therefore must have been situated in west Anatolia between the rivers Gediz Nehri / Hermus and Büyük Menderes Nehri / Maeander, it is not necessarily the case that the CLuwian texts derive from that area. According to Melchert (2003: 174) “[i]n the few cases where a determination can be made, the Luwian rituals found in Hattusa are imported from the southern region of Kizzuwatna”. For texts, grammar, vocabulary and historical phonology, see e.g. Starke 1985, Starke 1990, Melchert 2003c, Laroche 1959, Melchert 1993a, Melchert 1994a: 229-281.

Hieroglyphic Luwian is a language closely related to (but nevertheless clearly distinct from) Cuneiform Luwian (see above). It is written in an indigenous hieroglyphic script that seems to have been especially designed for this language. Although seals containing names written in these hieroglyphs can be dated back to the OH period (ca. 1600 BC), the oldest real HLuwian text (the Ankara Silver Bowl, cf. Hawkins 1997) may be dated around 1400 BC. By far most of the ca 260 known HLuwian texts are rock inscriptions. Some thirty of these were set up during the Hittite Empire period (13th century BC), but the bulk (some 230) date from the post-Empire period (1100-700 BC). The Empire-period inscriptions are found all over the area of the Hittite Empire, whereas the post-Empire-period inscriptions are found in south-east Anatolia only, the region of the so-called Neo-Hittite city states. For texts, grammar, vocabulary, script and historical phonology, see e.g. Hawkins 2000, Melchert 2003c, Plöchl 2003, Payne 2004, Meriggi 1962, Marazzi (ed.) 1998, Melchert 1994a: 229-281.

Lycian is the language of the Lycian region, situated on the south-west Anatolian coast between the modern-day cities Fethiye and Antalya. The Lycians called themselves Tr"mili, which must be identical to the name Termilai used by Herodote. The Lycian language is known from some 150 coin legends and 170 inscriptions on stone, dating from the 5th and 4th century BC, using a native alphabet related to Greek. Most inscriptions are funereal and show little variation, but a few are edicts, the most important one of which is the trilingual of Letoon (with Greek and Aramaic translations). On two inscriptions, including another important edict, the stele from Xanthos, (part of) the text is written in a dialect distinct from ‘normal’ Lycian and that is either called Lycian B (with “Lycian A” referring to normal Lycian) or Milyan. Linguistically, Lycian is closely related to CLuwian and HLuwian, and it is probable that they formed a sub-group within the Anatolian family. Nevertheless, the old view that Lycian is a younger variant of Luwian cannot be upheld: for instance, the Lycian vowel system (a, e, i, u) cannot be derived from the simpler vowel system of Luwian (a, i, u). For texts, grammar, vocabulary and historical phonology, see e.g. Kalinka 1901, Neumann 1979a, Houwink ten Cate 1961, Neumann 1969, Hajnal 1995, Melchert 2004a, Melchert 1994a: 282-328.

Lydian is the language of classical Lydia, situated in central western Anatolia, in the modern-day provinces of İzmir and Manisa. It is attested on ca one hundred stone inscriptions in a native alphabet related to Greek, dating from the 8th-3rd century BC, with a peak around the 5th and 4th century. Most inscriptions stem from Sardis, the capital of Lydia. Although some inscriptions are fairly lengthy, the absence of a large bilingual text (the four bilingual texts, two Lydian-Greek and two Lydian-Aramaic, are too short to be of much help) makes Lydian difficult to understand. The little knowledge we do have shows that it stands quite apart from the other Anatolian languages. For texts, grammar, vocabulary and historical phonology, see e.g. Gusmani 1964, Gérard 2005, Melchert 1994a: 329-383.

Carian, Sidetic and Pisidian are so poorly attested or badly understood that I will disregard them in this work. Carian is the language from classical Caria (south-western Anatolia, between Lydia and Lycia), but most Carian texts (some 150) have been found in Egypt where Carian mercenaries were working. They date from the 6th-5th century BC. The twenty texts from Caria itself seem to date from the 4th century BC. Although the language is written in a script seemingly related to Greek, it still has not been fully deciphered. Only recently (in the 1990’s) reliable sound values have been established for some signs on the basis of (short) Carian-Egyptian bilinguals. In 1996 a rather large Carian-Greek bilingual inscription was found in Kaunos, which hopefully will elucidate our knowledge of the Carian language in the future. Sidetic is known from 8 inscriptions from the city Side in the region Pamphylia, written in a native alphabet related to Greek. They date from the 3rd century BC. Three of them have a Greek version, which enables us to identify a few words that show that the language must be Anatolian. Nevertheless, with such a small corpus not much is known about Sidetic. Pisidian is the language known from some thirty tomb inscriptions from Pisidia, located between the lakes Eğridir Gölü and Beyşehir Gölü. They are written in the regular Greek alphabet and can be dated to the 1st-2nd century AD. Until now only names have been attested, but the fact that a genitive ending ‑s is used (= Luw. ‑ašša/i‑) in combination with the observation that some names clearly are Anatolian (e.g. Μουσητα = Luw. Mu¾aziti), points to an Anatolian language. For an introduction to these languages, see e.g. Van den Hout 1998.


Within the Anatolian branch, Hittite is by far the best attested and best-known language. Therefore, Indo-Europeanists understandably often use data only from this language when working on the reconstruction of Proto-Indo-European. Nevertheless, I agree with Melchert (1994a: foreword) who states that “only [an Anatolian] perspective can fully illuminate the history of Hittite”. Although within the research project that enabled me to write this work there was no time to fully describe the prehistory of the other Anatolian languages, I have taken them into account whenever necessary and sometimes digressed on certain aspects of their historical phonology.


0.6 The place of the

Anatolian branch within PIE


A mere six years after the decipherment of Hittite, Forrer (1921: 26) writes: “Man wird [...] nicht umhin können, das Kanisische [= Hittite, AK] als Schwestersprache des aus den indogermanischen Sprachen erschlossenen Urindogermanischen zu bezeichnen”. This idea was soon taken over by Sturtevant (e.g. 1926, 1929) who even introduces the name ‘Indo-Hittite’ for the proto-language that would underly Anatolian on the one hand and Indo-European on the other. The ‘Indo-Hittite hypothesis’ has been much discussed over the years, even resulting in a monograph (Zeilfelder 2001). Although at first scholars were sceptical, in the last decade it seems as if a concensus is being reached that the Anatolian branch indeed was the first one to split off from the Proto-Indo-European language community. Nevertheless, these opinions are often based on the archaicity of some phenomena in Hittite (compare e.g. Zeilfelder’s book that is entirely devoted to the archaic features of Hittite), whereas already Pedersen (1938: 12) rightly remarks that “[d]as unmittelbar abweichende Aussehen des Hittitischen hat natürlich keine Bedeutung; Neuerungen des Hittitischen oder Verlust des Alten entweder im Hittitischen oder in den zehn Sprachzweigen haben keine Beweiskraft”. He states “dass nur gemeinsame Neuerungen der zehn lebendigen Sprachzweige sie dem Hittitischen gegenüber als eine Einheit charakterisieren können”.[8] So, no matter how archaic some features of Hittite or the other Anatolian languages are and no matter how many of them have been preserved, we can only decide for a special position of Anatolian within the IE family if we could show that the other IE languages share a common innovation that is not present in Anatolian. It is my intention to present a few of these cases here (for full etymological treatment cf. their respective lemmata).


(1) The Hitt. verb mer-zi / mar- ‘to disappear’ is generally considered cognate with the PIE root *mer- that means ‘to die’ in the other IE languages (Skt. mar- ‘to die’, Av. mar- ‘to die’, Gr. ἄμβροτος ‘immortal’, (Hes.) ἔμορτεν ‘has died’, Lat. morior ‘to die’, OCS mrěti ‘to die’, Lith. miSti ‘to die’, Goth. maurþr ‘murder’, Arm. me²anim ‘to die’). Since typologically it is improbable that an original meaning ‘to die’ would develop into ‘to disappear’, whereas a development of ‘to disappear’ to ‘to die’ is very common,[9] we must assume that the original meaning of the root *mer- is ‘to disappear’, as is still attested in Anatolian, and that the semantic development to ‘to die’ must be regarded as a common innovation of the other Indo-European languages.


(2) The words for ‘you (sg.)’ in the Anatolian languages (e.g. Hitt. zīk / tu‑) must go back to the PAnat. pair *tiH, obl. tu- (cf. § 2.2.3). All other IE languages point to a pair nom. *tuH, obl. *tu‑, however. If we assume that the pair *tuH, *tu- is original, it is in my view impossible to explain how the Anatolian pair *tiH, tu- has come into being. If, however, we assume that the pair *tiH, *tu- is original, we only need to assume a trivial analogical change of *tiH to *tuH on the basis of the oblique stem *tu-,[10] in order to explain the pair *tuH, tu‑. I therefore conclude that the Anatolain pair *tiH, *tu- is original and that the pair *tuH, *tu- as reflected in the other IE languages is the result of a common innovation, namely introduction of the oblique stem *tu- into the nominative.


(3) The Hittite verb šāŸ-i means ‘to fill up, to plug, to stuff’ and in my view is cognate to the PIE root *seh2- that is usually translated ‘to satiate’ (Gr. μεναι ‘to satiate oneself’, Skt. ásinvant- ‘unsatiable’, TochB soy- ‘to be satisfied’). Since it is understandable how a meaning ‘to fill up, to stuff’ would develop into ‘to satiate’, but not how ‘to satiate’ would be able to turn into ‘to fill up, to stuff’, it is likely that the original meaning of the root *seh2- was ‘to fill up, to stuff’ as attested in Hittite, and that the meaning ‘to satiate’ as visible in the other IE languages is a common innovation.


(4) As I argue s.v. MUNUSduttari¤ata/i‑, the HLuwian word tuwatra/i- ‘daughter’ and Lyc. kbatra- ‘daughter’ point to a PLuw. form *duegtr‑, whereas MUNUSduttari¤ata/i- can only be explained from PLuw. *dugtr‑. So the Anatolian material points to an original inflection *dhuégh2tr, *dhugh2térm, *dhugh2trós ‘daughter’. In all other IE languages, we find forms that point to an inflection *dhugh2tËr, *dhugh2térm, *dhugh2trós, however (e.g. Skt. duhitÄ, duhitáram, duhitú¡; Gr. θυγάτηρ, θυγατέρα, θυγατρός; Lith. duktG, dùkterį, dukteSs; OCS dъšti; Osc. futír; Arm. dowstr; ModHG Tochter; Gaul. duχtir). Since I do not see how an original inflection *dhugh2tËr, *dhugh2térm, *dhugh2trós could ever be replaced by *dhuégh2tr, *dhugh2térm, *dhugh2trós, whereas the other way around it is a trivial development (introduction of the acc. form in the nominative combined with analogy to *ph2tËr ‘father’), I assume that the situation as reflected in Anatolian, *dhuégh2tr, *dhugh2térm, *dhugh2trós, represents the original state of affairs and that the morphological change to the paradigm *dhugh2tËr, *dhugh2térm, *dhugh2trós as reflected in the other IE languages is a common innovation.


(5) The PIE root *h2erh3-, which denotes ‘to plough’ in all non-Anatolian IE languages (Gr. ἀρόω ‘to plough’, Lat. arō ‘to plough’, OIr. ‑air ‘to plough’, OHG erien ‘to plough’, Lith. árti ‘to plough’, OCS orati ‘to plough’), is reflected in Hittite in the verbs Ÿarra-i / Ÿarr- ‘to grind, to crush’ and Ÿārš-i ‘to harrow, to till the soil’ (with an s-extension). Although the latter term indeed has agricultural connotations (but note that its usage in the pair Ÿārš-i ... terepp-zi ‘to harrow and plough’ clearly shows that it itself did not mean ‘plough’, but rather originally ‘to crush (the ground)’), the basic meaning of *h2erh3- seems to have been ‘to crush’, which could be used having the ground as an object as well. The semantic development of ‘to crush; to harrow (the ground)’ to the specific technical term ‘to plough (with a plough)’ as visible in the non-Anatolian languages must be regarded as a common innovation of them and shows that the Anatolian branch split off before the introduction of the plough.


(6) The verbal root *meh1- ‘to refuse, to reject’ is attested as a fully living verb in Hittite, mimma-i / mimm- < *mi-moh1- / *mi-mh1, whereas in the other IE languages it only survives in the 2sg.imp.act. form *meh1 ‘don’t!’ that has been grammaticalized as a prohibitive particle (Skt. mÄ, Arm. mi, Gr. μη, TochAB ). This grammaticalization must be a common innovation of these languages.


(7) As I will show s.v. *ekku-, the Anatolian words for ‘horse’ (Hitt. ANŠE.KUR.RA-u-, CLuw. ANŠE.KUR.RA-u-, HLuw. EQUUSá-sù-, Lyc. esb-) all reflect an u-stem noun and therefore point to a preform *h1e§u-, whereas the words for ‘horse’ in the other IE languages (Skt. áśva-, Av. aspa-, Gr. ἵππος, Myc. i-qo, Lat. equus, Ven. ekvo-, OIr. ech, OE eoh, TochB yakwe, TochA yuk ‘horse’, Arm. ēš ‘donkey’, Lith. ašvà ‘mare’, OPr. aswinan ‘mare’s milk’) all point to an o-stem noun *h1e§uo-. There is no known phonological development through which PIE *h1e§uo- could yield PAnat. *h1e§u- and in view of the productivity of the o-stem inflection in Anatolian it is unlikely that PIE *h1e§uo- would have yielded PAnat. *h1e§u- through secondary developments. We therefore must conclude that the PAnat. u-stem *h1e§u- reflects the original state of affairs and that the thematicization as visible in the non-Anatolian IE languages (which is a trivial development) must be regarded as a common innovation of them.


Although these examples may be of uneven value, we must realize that each and every example on its own already shows that the non-Anatolian IE languages have commonly undergone an innovation where Anatolian has preserved the original situation. This can only lead to one conclusion, namely that the non-Anatolian IE languages still formed one language community (at least close enough for innovations to reach all speakers) at the moment that the Anatolian branch split off. In other words, each of these examples is conclusive evidence that the Anatolian branch was the first one to split off from the mother language. Whether we then call this mother language Proto-Indo-European, Proto-Indo-Hittite or something else is only a matter of terminology. I think that the term Proto-Indo-European is still adequate as long as we keep in mind that the Anatolian branch may have preserved an original situation that has undergone innovations or losses in the other IE languages (but likewise the Anatolian branch may have innovated or lost an original situation that is still present in the other IE languages, of course). So, the times of a solely Graeco-Indic reconstruction of PIE are definitely over: we should always take the Anatolian material into account and keep in mind the possibility that the non-Anatolian IE languages have commonly undergone an innovation where Anatolian preserves the original, PIE situation.


[1] With ‘inherited lexicon’ I mean those words that are build up of morphemes that can be reconstructed for the Proto-Indo-European mother language (i.e. have cognates in the other IE languages). It should be noted that I therefore do not claim that each reconstructed form that in this book has been glossed as ‘PIE’ did exist as such in the PIE mother language. These reconstructions should rather be seen as explications of the morphology of the Hittite words in the light of their PIE origin.

[2] The existence of the land îatti had already been known from documents from Mesopotamia as well as from the Amarna-letters, some of which were written (in Akkadian) by Šuppiluliuma, king of the land îatti.

[3] Some scholars further divide the last period into two stages, namely Early Neo-Script (ENS) and Late Neo-Script (LNS = German spätjunghethitisch, sjh.). Since this is not yet common practice and since as a historical linguist I am mainly interested in the oldest stage of the language, I have decided to only use the term Neo-Script (NS) in this book (although I must admit that in hindsight for instance the many morphological processes in the Hittite verbal system that take place in the NH period perhaps could have been described more precisely if I had used this more precise system of dating).

[4] Since any text in Old Script by definition must contain an Old Hittite composition, I only use the abbreviation OS in these cases (and not OH/OS). Similarly in the case of Neo-Hittite compositions which by definition can only be written in Neo-Script: I use the abbreviation NH (and not NH/NS).

[5] In this book I have transliterated the cuneiform signs according to Rüster & Neu’s Hethitisches Zeichenlexikon (= HZL). Note however that the signs GIR, GAD, NIR, UD and ŠIR, for which HZL cites the phonetic value kir, kid/t9, nir, pir and šir, respectively, sometimes have to be read ker, ket9, ner, per and šer as well.

[6] Because of the ambiguity of this way of transcribing (e.g. the transcription ¾alaŸzi does not tell us whether the underlying form was ¾a-al-aŸ-zi or ¾a-la-aŸ-zi), I have used this transcription only when the transliteration of a certain form has been already given.

[7] Possibly pronounced /pla?a/, which can be deduced from the difference between palaumnili ‘in the language of the Palaic people’ and nešumnili ‘in the language of the Nešite people’: just as neš-umnili is derived from Neš-a, with morphological replacement of the final ‑a, pala-umnili must be derived from a form Pala-a = /pla?a/.

[8] Note that Pedersen concludes (1938: 190-1): “Wir haben weder in der Grammatik noch in der Lautlehre einen Anhalt dafür gefunden, dass das Hittitische in einem Gegensatze zu den anderen indoeuropäischen Sprachen stünde. Es stellt sich als elfter Sprachzweig neben die zehn altbekannten Sprachzweige”.

[9] Consider euphemisms like ModEng. to pass away, to be gone and, even more clearly, French disparaître.

[10] In personal pronouns it is a common phenomenon that nominatives are altered on the basis of oblique forms, cf. for instance the NH use of ammuk ‘I’ instead of original ūk on the basis of the oblique forms ammuk ‘me’.

Etymological Dictionary of the Hittite Inherited Lexicon

Towards a Hittite Historical Grammar


This part consists of two chapters. In the first chapter, called Historical Phonology, I will first give an overview of the phonological systems that I reconstruct for Proto-Indo-European and Proto-Anatolian. Then I will treat in detail the arguments on the basis of which a phonetic and phonological analysis of the cuneiform script in which Hittite is written can be made, which results in the establishment of the Hittite phoneme inventory. The last step is that the phonological changes that took place between Proto-Indo-European and Hittite as attested are described in detail.

The second chapter, Aspects of Historical Morphology, deals with the historical morphology of the Hittite nominal system, the prehistory of the Hittite pronominal system and the morphological and historical interpretation of the Hittite verbal system. It should be noted that the paragraphs on the nominal system are rather rudimentary: in Part Two each noun has received an extensive etymological treatment, including a detailed analysis of its morphological prehistory. Furthermore, each nominal ending is etymologically treated under its own lemma.

Etymological Dictionary of the Hittite Inherited Lexicon

Hittite Historical Phonology

1.1 Proto-Indo-European phoneme inventory

In the present book I have worked with the following reconstruction of the Proto-Indo-European phonological system (based on Beekes 1995: 124):

                stops                      p             t                           k              kw

                                             b             d             œ             g             gw

                                             bh            dh            œh            gh            gwh

                fricative                s

                laryngeals             h1            h2            h3

                liquids                    l              r

                nasals                    m             n

                semivowels           i              u

                vowels                   e              o

                                             ē              ō[1]

It should be noted that despite the fact that I have used the traditional symbols for the reconstructed stops, I follow Kortlandt (2003: 259) who argues that the traditional ‘voiceless’ series (*p, *t, *, *k and *kw) in fact were plain fortis stops [p:, t:, kj:, k:, kw:], the traditional ‘voiced’ series (*b, *d, *œ, *g, *gw) were lenis (pre‑)glottalized stops [pˀ, tˀ, kˀj, kˀ, kˀw] and the traditional ‘aspirated voiced’ stops (*bh, *dh, *œh, *gh, *gwh) were plain lenis stops [p, t, kj, k, kw]. Note that the stops therefore have “neither voicedness nor aspiration as distinctive features” and that “[t]he phonetic distinction between fortes *T: and lenes *T was probably a matter of consonantal length” (ibid.).

1.2 Proto-Anatolian phoneme inventory

Although in this book it was not my aim to provide a historical treatment of the Anatolian family as a whole, it is in some cases convenient to use Proto-Anatolian reconstructions, especially when a word can be reconstructed for the Proto-Anatolian stage, but not for Proto-Indo-European. I work with the following phoneme inventory.

            stops               fortis           p          t                    k          kw

                                    lenis            b          d          œ         g          gw

            fricative[2]                           s         

            affricate[3]                           ts

            ‘laryngeals’                       ʔ          H         Hw

            liquids                                l           r

            nasals                                 m         n

            vowels                        i, ī                           u, ū

                                                e, ē                          o, ō


                                                                a, ā


The reconstruction of only two rows of stops is based on the fact that in none of the Anatolian languages evidence can be found for a distinction between the PIE ‘voiced’ and ‘aspirated’ series, which makes it likely that these merged in the pre-PAnatolian period already. The PIE palatovelars and normal velars have different reflexes in Luwian and Lycian (e.g. * > Luw. z, Lyc. s vs. *k > Luw. k, Lyc. k), and therefore must have been distinct in PAnatolian as well.

In Kloekhorst 2006b I have treated the reflexes of the PIE laryngeals in initial position in the Anatolian languages. There I have suggested that for PAnatolian there is only evidence for two ‘laryngeals’ word-initially, namely */ʔ/ and */H/, which is valid for word-internal position as well. Moreover I have argued that since the Hittite phoneme /Hw/ < *h2u corresponds to the Lycian phoneme q = [kw] < *h2u it is likely that this phoneme, /Hw/, was PAnatolian already.

Because the old PIE laryngeal system collapses (in the positions *#He- and *CRHV PIE *h3 merges with *h2 into PAnat. */H/; in all other positions *h3 merges with *h1 into PAnat. */ʔ/), the allophonic colouring of pre-PAnat. *e due to adjacent *h2 and *h3 becomes phonemicized, yielding the PAnat. phonemes */a/ and */o/ (the latter ultimately merging with the reflex of PIE *o). Note that Lycian shows different reflexes of *a (namely a) and *o (namely e), which proves that at the PAnatolian level the vowels /a/ and /o/ were distinct.

1.3 Hittite phoneme inventory

1.3.1 Cuneiform script

The history of the cuneiform scripts starts with the Sumerians’ desire to keep track of business transactions: around 3200 BC the first economic records and inventories were made on lumps of clay by drawing pictures of specific objects together with strokes and cones to represent numbers. Although these pictographs were initially used only as a one-to-one representation of the objects they depicted, in the course of time they not only received a broader semantic notion (e.g. the sign ‘mouth’ could be used for ‘to speak’ and ‘word’ as well), but also could be used more or less phonetically (e.g. the sign ‘mouth’ was pronounced ka, and could be used for writing words with a similar phonetic shape). Together with the fact that the pictographs became more and more stylized and in the end were not well recognizable as the original object anymore, a breeding ground was laid for this system’s development into a phonetic script. Around 2350 BC the Sumerian script was adopted by the Akkadians, who reshaped it into a writing system in which the phonetic representation of the language served as the basis, although logograms, i.e. signs that represent a certain notion without referring to it phonetically (the abstracted descendants of the Sumerian pictographs), were still used on a large scale.[4]

The cuneiform script that is used by the Hittite scribes is derived from an Old-Babylonian cursive type that is known from Northern Syria (e.g. Alalaḫ). How exactly the practice of writing found its way from there to îattuša is not fully clear.[5] Just as in Akkadian, the writing system is basically phonetic.[6] Nevertheless, a word can be written logographically with so-called sumerograms (i.e. the logograms that are derived from the Sumerian script,[7] e.g. DINGIR ‘god’) or with akkadograms (i.e. as if in Akkadian,[8] e.g. Ú-UL ‘not’). It is likely that in both cases the Hittites read these logographically written words as their Hittite counterparts, as can be seen by the use of phonetic complements (i.e. the addition of phonetic signs to spell part of the word underlying the logographic writing, usually to indicate the proper ending, e.g. DINGIR-uš = šīuš ‘god’). To complicate matters, sumerograms sometimes could be extended by an Akkadian phonetic complement, e.g. DINGIRLUM,[9] which functions as a sort of sumerographic writing of Akk. ilum ‘god’, which itself must be regarded as an akkadographic writing of the Hittite word šīuš ‘god’. Moreover, the cuneiform writing system makes use of so-called determinatives, i.e. logograms that indicate a certain semantic sphere of the word next to which they are placed.[10] For instance, GIŠ ‘wood’ can be used with words that denote objects that are made of wood (e.g. GIŠninial- ‘cradle’), É ‘building’ can be used with words that denote buildings (e.g. Éištā, Éištī ‘mausoleum(?)’). Although usually placed in front of a word, some determinatives can be placed at the end of a word (e.g. MUŠEN ‘bird’ as in āran-MUŠEN ‘eagle’).

For the linguist interested in the Hittite language this complicated system has some disadvantages: certain words are only attested with a sumerographical spelling and never with phonetic signs, which means that we do not know the Hittite rendering of these words. This is not only the case with some rare words, but also with certain words that belong to the basic vocabulary. For instance, ‘son’ is attested with the sumerogram DUMU only; ‘daughter’ is only spelled DUMU.MUNUS[11]; the Hittite reading of the sumerogram MUNUS ‘woman’ is disputed[12]; we do not know the Hittite words for îUR.SAG ‘mountain’, GUŠKIN ‘gold’, KÙ.BABBAR ‘silver’ or numerals like ‘five’, ‘six’, ‘eight’, etc. Nevertheless, we must not forget that exactly the usage of these sumerograms has played a key-role in deciphering the Hittite language and that even nowadays the best evidence for the meaning of a rarely attested word is when a parallel text or copy is found with this word duplicated by a sumerogram.

Despite the wide use of logograms, the Hittite writing system is basically a phonetic one. The phonetic signs are all syllabic, which means that they possess a value V, CV, VC and CVC only (in which V = vowel and C = consonant). Herewith, the script was not very well equipped for writing Hittite. As an Indo-European language, Hittite possesses many words with sometimes large consonant clusters, which are difficult to render with a syllabic script: if one wants to write word-initial or word-final consonant clusters or internal clusters of three or more consonants with syllabic signs, one cannot avoid to write vowels that are neither phonetically nor phonologically real. For instance, the word /parHtsi/ ‘he chases’ is spelled pár-a-zi as well as pár-a-zi. In this case, the alternation between pár-a-zi and pár‑a‑zi proves that these a’s are “empty”. In other cases, determining whether a vowel grapheme is phonetically and/or phonologically real can be quite difficult, however.

In the following sections I will discuss in detail the peculiarities of the cuneiform script as used by the Hittites in order to determine the Hittite phonological system. I will first look at consonants and then move on to the vowels.

1.3.2 Stops


The Old-Babylonian cuneiform syllabary that functioned as the source of the syllabary used in Boğazköy originally had distinct signs for voiced and voiceless stops, e.g. BA vs. PA, DA vs. TA, GI vs. KI, etc.[13] Nevertheless, the Akkadian texts from Boğazköy do not use these contrasting pairs to express a distinction between voiced and voiceless stops. For instance, the sign PA is used as pa as well as , whereas BA is used as ba as well as . Similarly, TA is used as ta as well as ; DA as da as well as ; TI as ti as well as ; DI as di as well as ti4, etc.

In the Hittite texts, the contrasting pairs are not used for voice distinctions either. They are largely interchangeable instead: e.g. ba-i-iš = pa-iš = /páis/ ‘he gave’; da‑it-ti = ta-it-ti = /táiti/ ‘you place’; gi-nu-uz-zi = ki-nu-uz-zi = /kinÜtsi/ ‘he opens up’.[14] It must be admitted that certain words show an almost consistent spelling with e.g. DA whereas others are spelled exclusively with TA (e.g. dāi ‘he puts’ is consistently spelled with the sign DA; the sentence initial conjunction ta is consistently spelled with TA), but all attempts to interpret these cases as pointing to a phonemic opposition in voice,[15] have failed.[16] We rather have to interpret these cases as spelling conventions.

Nevertheless, it cannot be denied that the Hittite scribes did distinguish between two series of stops which were expressed by single (V-C1V) vs. geminate spelling (VC1-C1V). Sturtevant (1932a) was the first to describe this phenomenon and to show that from an etymological point of view the single spelled stops correspond to the PIE ‘voiced’ and ‘voiced aspirated’ series *D and *Dh, whereas the geminate spelled stops etymologically correspond to the PIE ‘voiceless’ series *T (‘Sturtevant’s Law’). The exact phonetic interpretation of the single spelling (which is often termed ‘lenis’) vs. the geminate spelling (often termed ‘fortis’) is difficult, however.

In Hurrian, we find a similar system, namely a distinction between stops that are spelled V-C1V and stops that are spelled VC1-C1V. Yet, on the basis of Hurrian texts from Ugarit that are written in an alphabetic script, we are much better able to interpret these spellings phonetically. According to Wegner (2000: 40), Hurrian shows a phonemic distinction between short (= single spelled) and long (= geminate spelled) stops, which are both voiceless. The short stops became phonetically voiced in some environments (namely intervocalically and after resonant), but these should be regarded as mere allophones.

Kimball (1999: 54) assumes that the Hittites took over the cuneiform script from the Hurrians and states that “[s]cribes adapting the syllabary for Hittite, if they were native speakers of Akkadian, which had phonemic voicing, or native speakers of Hittite, which probably had phonemic voicing, would have tended to hear and spell Hurrian single intervocalic stops as voiced and to hear and spell double stops as voiceless, and, unless they themselves were acquainted with the Old Babylonian values, they would have spelled Hittite voiceless stops with double stops and voiced stops with single stops”. Apart from the fact that this reasoning is rather circular (using the assumption that Hittite probably had phonemic voicing in an argumentation to show that the Hittite spelling reflects phonemic voicing), it would predict that Boğazköy Akkadian would use the same spelling convention to distinguish between voiced and voiceless stops. This is not the case, however: “[t]here seems to be no trace of this orthography [i.e. a system of distinction between stops spelled VC-CV (voiceless(?)) and those spelled V-CV (voiced(?))] in Bo[ğazköy] Akk[adian]” (Durham 1976: 371). Moreover, there are spelling conventions in Hurrian that are not used in Hittite, e.g. the use of the sign GE/I as having the e-vowel only (/ke/) vs. the use of the sign KE/I as having the i-vowel only (/ki/) (Wegner 2000: 37-8). This shows that the Hittites cannot have adopted the cuneiform script directly from the Hurrians.

Melchert (1994a: 20) interprets the Hittite ‘fortis’ stops as long and voiceless (TT‑), whereas the ‘lenis’ stops are short and voiced (D‑). Furthermore, Melchert assumes that secondarily a third series arose, namely stops that are long as well as voiced (DD‑) (the result of e.g. *‑Dh2). The existence of this last series is highly improbable, however: there is not a shred of evidence for a distinction in spelling between “-TT-” and “-DD-”, and therefore a phonetic and phonological distinction between the two cannot be proven. Moreover, Melchert does not give any evidence for the view that the long stops were voiceless and the short ones voiced.

In my view, voice cannot have been a distinctive feature between the geminate spelled and the single spelled stops. If voice really was a phonological feature of one of these series, why did the Hittite scribes not use the voice-distinction available in the Akkadian syllabary? Even in writing Akkadian, of which we know that it had phonemic voicing, a distinction in voice is not expressed in spelling, which suggests that the Hittite scribes just were unable to distinguish voiced from voiceless stops. Moreover, as we saw above, the fact that in Boğazköy Akkadian the system of single vs. geminate spelling is not used, shows that the ‘fortis/lenis’-distinction cannot be compared phonetically to the distinction in voice known from Akkadian.

The fact that the Hittite scribes used the orthographically awkward distinction between geminate vs. single spelling in writing Hittite can only mean that the phonetic distinction between the two series of stops was length. This is supported by the following observations.

First, in certain phonological developments where it is significant whether a syllable is closed or open, a geminate spelled stop counts as a closing factor. For instance, the form kitta ‘he lies’ < *tta < *éito shows the ‘shortening’ of *i in a closed syllable,[17] which shows that ‑tt- closes the syllable and therefore must be regarded as phonetically long [t:].

Second, if voice was a distinctive feature, we would expect to find voice assimilation. So, if a word like e-ku-ud-du ‘he must drink’ really contained a cluster [‑gwt‑] with a voiced stop [gw] before a voiceless stop [t], I do not see why neither the [gw] was devoiced yielding **[‑kwt‑] (spelled **e-ek-ku-ud-du), nor the [t] was voiced yielding **[gwd] (spelled **e-ku-du). Since neither of these assimilations took place, we are bound to conclude that voicedness is neither a phonemic nor a phonetic feature of the Hittite stops.

I therefore assume that the ‘fortis’ consonants (spelled with a geminate) were phonetically long and the ‘lenis’ consonants (spelled single) were short and that there was no distinction in voice. So VppV = [p:] vs. VpV = [p]; VttV and VddV = [t:] vs. VtV and VdV = [t]; etc. Nevertheless, I have chosen to adopt the following phonemic spelling throughout the book:

            Fortis                     /p/           /t/            /k/           /kw/        

            Lenis                      /b/           /d/           /g/           /gw/        

The choice of these symbols for the phonological representation of the stops is a matter of convenience. It does not indicate that I consider voicedness a phonemic feature at any point in the history of Hittite.

It should be noted that the phonetic change of a fortis stop into a lenis stop or vice versa (which can happen in certain phonetic environments) should consequently not be called ‘voicing’ or ‘devoicing’, but rather ‘lenition’ and ‘fortition’.[18] For instance, the fact that impf. ak-ku-uš-ke/a- ‘to drink’ shows a fortis /kw/ whereas the basic verb has /gw/ (eku-zi / aku‑) is due to fortition of /gw/ to /kw/ in front of /‑ske/a‑/, and not due to devoicing.[19]

Since in word-initial position no orthographic distinction between geminate and single stop could be made, it is unclear whether the two series are distinct in this position or have merged. Since there is not a single spelling practice in Hittite (nor in Palaic and CLuwian, for that matter) that even attempts to indicate a distinction between initial *T and *D / *Dh, I cannot but assume that in initial position this distinction has been lost. Nevertheless, the distinction must have been present in Proto-Anatolian, as is indicated by the fact that initial *ti- yielded Hitt. z- and *di- > Hitt. š‑, whereas they merged in Luwian as ti‑. So, if the two series have merged in Hittite in initial position, this must be a post-Proto-Anatolian development.[20] On the basis of reduplicated forms like kikkiš-tta(ri), the imperfective of kīš-a(ri) / kiš- ‘to happen, to become’ < *œeis‑, it has been assumed that in Hittite the initial stops merged in the fortis series /p, t, k and kw/.[21] Since the moment of the creation of this reduplicated form is unknown, it does not shed much light on the situation in Hittite, however.[22] On the contrary, the stem atuk- ‘terrible’, which probably reflects *h2tug‑, shows lenition of PIE *t to Hitt. /d/ in the initial cluster *h2t-[23] and therefore could be used as an argument for the opposite view, namely that all initial stops merged into the lenis series. This example is again non-probative, however, because the fact that /d/ is a lenis stop does not prove anything regarding the status of initial . All in all, the matter cannot be decided. Since merger equals absence of a phonemic distinction, the matter may not be very interesting from a phonological point of view. In this book I will cite initial stops with their fortis variant in phonological interpretations, so /p‑/, /t‑/, /k‑/ and /kw‑/.

We could assume that in word-final position a similar merger has taken place, and Melchert (1994a: 85) states that “[v]oiced stops ha[ve] been generalized in word-final position”, giving “pa-i-ta-aš = /páyd-as/ ‘went he’” as an example. This example is non-probative, however, since the enclitic personal pronoun =a- may have had a leniting effect on the preceding consonant (just as the enclitic particle =(m)a ‘but’ had, in contrast with the fortiting enclitic particle =()a ‘and’). It is moreover contradicted by the words takku /takw/ and nekku /nekw/ that show a fortis /kw/ in word-final position. When compared with 2sg.imp.act. e-ku /ʔégw/ ‘drink!’, which unmistakingly has a lenis stop in word-final position, we must conclude that the fortis and lenis stops remained distinct word-finally.

For the phonemicity of the labiovelars, compare the spellings e-ku-zi, e-uk-zi ‘he drinks’ and tar-ku-zi, tar-uk-zi ‘he dances’ that point to a monophonemic /gw/ and /kw/, respectively, and not to /gu/ and /ku/. Moreover, a-ku-e-ni ‘we drink’ contrasts with ar-nu-me-ni ‘we transport’ which shows that the former is /ʔgwuéni/ < *h1gwhéni, whereas the latter is /ʔrnuméni/ < *h3r-nu-éni, where ‑u- yielded ‑um‑. A third argument is that e-ku-ut-ta ‘he drank’ shows the postconsonantal allomorph ‑tta of the 3sg.pret.act. ending (cf. e.g. e-ep-ta ‘he took’), whereas e.g. ar‑nu-ut shows the postvocalic variant ‑t. Compare also the fact that 1sg.pret.act. ekun ‘I drank’ shows the postconsonantal ending ‑un which contrasts which the postvocalic variant ‑nun as attested in e.g. ar-nu-nu-un ‘I settled’.

Summarizing, with regard to the stops, the Hittite phonological system nicely matches the Proto-Indo-European phonological system. If we compare the two systems, we see that between PIE and Hittite only three major developments took place. First, the loss of glottalization in the glottalized lenis series (the traditional ‘voiced’ series) caused this series to merge with the plain lenis series (the traditional ‘voiced aspirated’ series). Note that there is no indication that anywhere in the development between PIE and Hittite voice or aspiration has been a phonological or even phonetic feature. Secondly, the PIE palatovelars and the plain velars (which were still separate phonemes at the Proto-Anatolian stage) merged into Hitt. /k/ and /g/. Thirdly, word-initially the lenis and fortis series seem to have merged.

1.3.3 Glottal stop


In Kloekhorst 2006b, I have argued that in word-initial position Hittite possesses a phonemic glottal stop /ʔ/. This is apparent e.g. in the spelling difference between ú‑a-a-tar ‘inspection’ and a-a-tar ‘water’, where the former reflects *Hu-ótr and the latter *uódr. This means that ú-a-a-tar represents /ʔuā́dr/[24] and a-a-tar stands for /uā́dr/. A word-initial glottal stop also clarifies the symmetry between ša-ša-an-zi ‘they sleep’ /ssántsi/ < *ssénti and a-ša-an-zi ‘they are’ /ʔsántsi/ < *h1sénti.[25]

OS spellings like ne-e-a ‘turns’ < *néih1/3-o and é-e-a-u-e-eš ‘rains’ < *h2éih3‑eu- show that in the oldest period the glottal stop was still present in intervocalic position: /néʔa/ and /Héʔaues/. Younger spellings like ne-e-a (MH/MS) and é-e-a-u-e-š=a (OS), which must represent /néa/ and /Héaues/, respectively, show that intervocalic glottal stop was lost in the late OH period.

In the position *CRh1V, the glottal stop was retained as such throughout Hittite as can be seen by spellings like pa-ri-pa-ra-a-i ‘he blows’, which must represent /priprʔā́i/ < *pri-prh1-ói-ei.[26] Note that if *h1 would have been lost in this position, we would expect a spelling **pa-ri-ip-ra-a-i = **/priprā́i/.

1.3.4 Affricate

It is generally assumed that the consonant ‑z-[27] must be phonetically interpreted as an affricate [ts],[28] which for instance follows from the fact that the outcome of *‑ent-s is spelled ‑an-za. To which extent this affricate [ts] must be regarded as a single phoneme instead of a sequence of the phonemes /t/ and /s/ is less clear. A major source for ‑z- is the assibilation of *‑t- in front of *‑i‑. Nevertheless, the outcome of *‑ti- is not identical to the outcome of *‑Tsi‑, as we can tell from the fact that 2sg.pres.act. *h1édsi ‘you eat’ yields a form spelled e‑ez‑ši, whereas the 3sg.pres.act. ending of e/a- and ‑ške/a-verbs, *‑e‑ti, yields a form spelled ‑ez-zi or ‑Ce-zi, but never **-ez-ši. This shows that the former form, e‑ez-ši, represents /ʔédSi/,[29] whereas the latter forms represent /‑etsi/, with a monophonemic sound that I have rendered with the symbol /ts/ throughout this book. Yet, I do not interpret every spelling of ‑z- without a following ‑š- as a spelling of the phoneme /ts/. In cases where synchronically an analysis of t + s or d + s is obvious, I just write /ts/ or /ds/.[30] Note that I also interpret the outcome of *‑tt- or *‑dt- as /‑tst‑/ and /‑dst‑/. This is indicated by spellings like az-za-aš-te-ni /ʔdsténi/ ‘you eat’ < *h1d-th1é and e-ez-za-aš-ta /ʔédsta/ ‘he ate’ < *h1éd-t(o). This also makes it unnecessary to assume a variant /ds/ besides /ts/.

Yoshida’s attempt (2001) to show that in the oldest texts there was an opposition between geminate spelled ‑zz- and single spelled ‑z- that reflects PAnat. *‑ti- vs. *‑di- and therefore must be interpreted as an opposition between fortis /ts/ and lenis /ds/ is not convincing.[31]

1.3.5 Fricatives

I assume the following phonemic fricatives:

                Fortis                     /H/          /Hw/        /S/


                Lenis                      /h/           /hw/         /s/

The difference between fortis and lenis is expressed by geminate vs. single spelling. In initial position, we cannot decide whether we are dealing with the fortis or the lenis variant, and I therefore write /H‑/, /Hw‑/ and /s‑/ initially. For the phonemicity of the labialized laryngeals /Hw/ and /hw/, see Kloekhorst 2006b, where I argue that a spelling variation like tar-u-uz-zi, ta-ru-u-zi and tar-u-zi ‘he conquers’ points to a phonological form /tárHwtsi/.[32]


1.3.6 Resonants


The following resonants are phonemic:

Fortis                     /R/          /L/          /N/          /M/

Lenis                      /r/            /l/            /n/           /m/

Again, the difference between fortis and lenis is expressed by geminate vs. single spelling. Since this difference is not discernible in word-initial position, I arbitrarily write /l‑/, /n‑/ and /m‑/ here. Note that /r/ does not occur word-initially, which is a direct result of the PIE constraint that no word could start with an *r-.[33]

1.3.7 Syllabic resonants


Although the fact that a PIE sequence *CRC yields the Hittite spelling CaRC is well-established, the exact phonetic and phonological interpretation of this spelling is not fully clear. Usually, the spelling CaRC is phonologically interpreted as /CaRC/, having a real vowel /a/.[34] This cannot be correct, as can be seen from the verb ārš-zi / arš- ‘to flow’. Here we find a distribution between the strong stem that is spelled a-ar-aš- and the weak stem that is spelled ar-aš- or ar-šº. As I argue s.v., we expect the strong stem to reflect *h1ers‑, which suggests that the spelling a-ar-ašº phonologically must be interpreted as /ʔarS‑/, containing the vowel /a/.[35] This means at the same time, that the weak stem arš‑, which must reflect *h1rs‑, cannot contain the vowel /a/, since we then would have expected the same spelling for strong and weak stem. This forces us to look for another solution. There are two options. We can assume that in *CRC an anaptyctic vowel emerged that, although it did resemble /a/, was not identical to it. We could think of [ə] or [6] or similar, which was spelled with ‑a‑. This would mean that we would have to assume a phonemic vowel /ə/: so *CrC > Hitt. /CərC/, spelled CarC.

Alternatively, we could also envisage that these ‘vocalic’ resonants in fact were still identical to their consonantal counterparts, /r/, /l/, /m/ and /n/, and that their syllabicity was a pure phonetic feature that is predictable on the basis of the phonetic environment. This would mean that PIE *CrC yields Hitt. /CrC/, phonetically realized as [CərC] or [C6rC], spelled CarC.

The Hittite texts seem to offer arguments for both options. For instance, the verb appat(a)rie/a-zi ‘to confiscate’, which is a derivative in e/a- of the noun appātar ‘seizing’, is spelled ap-pa-at-ri-ez-zi (OS), ap-pa-ta-ri-ez-zi (OH/MS?), as well as ap-pát-ri-a-az-zi (MH/NS). The first and last attestation seem to point to phonetic [ʔp:atrié/á‑], whereas the second points to phonetic [ʔp:atrjé/á‑] or [ʔp:atərjé/á‑]. Phonemically, this verb must be interpreted as /ʔpadrié/á‑/, which subsequently shows that the noun appātar must represent /ʔpā́dr/, without a phonemic vowel /ə/.

In §, I argue that the i-verbs that show a synchronic ā/ɨ-ablaut, e.g. ga‑ra-a-pí / ka-ri-pa-an-zi = /krā́bi / krɨbántsi/, must ultimately reflect the normal *o/Ø‑ablaut, in this case *œhróbh1-ei / *œhrbh1-énti. Since the phonetically regular outcome of these verbs, Hitt. CRāCi / **CaRCanzi < *CróC-ei / *CRC-énti, shows a synchronic Schwebe-ablaut CRāC- / CaRC‑, the weak stem form was adapted by inserting the epenthetic vowel /ɨ/ on the place of the strong stem vowel: CRɨC- in analogy to *CRāC‑. This scenario implies, however, that the vowel of **CaRC- < *CRC- was at least phonetically real. Moreover, we would be inclined to think that this vowel must have been phonemically real as well in order to trigger a replacement by the secondary stem CRɨC-.

In word-initial position we encounter forms like *nsós > an-za-a-aš ‘us’ vs. *lghént- > la-ga-an-t- ‘felled’. Here it is quite clear that the outcome la-ga-an-t- cannot be regular: we should expect **al-ga-an-t‑, just as *nsós yielded an-za-a-aš. Quite obviously, the form la-ga-an-t- has been influenced by full grade forms like *lóghei > la-a-ki. This indicates that here we should assume a phonemic vowel /ə/, and subsequently interpret la-ga-an-t- as /ləgánt‑/. A similar concept explains a-al-a-an-zi, a-al-a-a-an-zi ‘they hit’ < *ulh3-énti. In analogy to the strong stem *uélh3-ti > Hitt. /uálHtsi/, a-al-a-zi ‘he hits’, the weak stem, which should have regularly yielded /ulHántsi/, was changed to /uəlHántsi/.

The vowel /ə/ is also necessary for the interpretation of ku-a-aš-ke/a‑, the imperfective of kuen-zi / kun- ‘to kill, to slay’. As I have argued in Kloekhorst 2007, a sequence *CuRCC or *KwRCC yields Hitt. CuaRCC (whereas *CuRCV or *KwRCV yields CuRCV). This means that ku-a-aš-ke/a- reflects /kwəske/a‑/ < */kw­ske/a‑/, the regular outcome of *gwhn-sé/ó‑.

Consequently, I will in principle treat the ‘syllabic’ resonants phonemically as their consonantal counterparts and assume that any phonetic realization with an epenthetic vowel is automatically determined by the environment. So the pair āršzi / aršanzi in my view represents phonological /ʔarS- / ʔrS‑/. Nevertheless, some words where the vocalization of a resonant is analogically altered or where the buccal part of the vocalized resonant has been lost, can only be analysed as containing a phonemic vowel /ə/ (e.g. la-ga-an-t- /ləgánt‑/ << *lghént‑, ma-ak-nu- /məgnu‑/ << *mœ-nu‑, a-al-a-an-zi /uəlHántsi/ << *ulh3énti and ku-a-aš-ke/a- /kwəske/a‑/ < *gwhnsé/ó‑). The vowel /ə/ is rather marginal, however.



1.3.8 Semi-vowels


It is usually assumed that Hittite possessed two semi-vowels or glides, namely /y/ and /w/. This implies that these are phonologically different from the vowels /i/ and /u/. Let us look at several phonetic environments to see if this is really the case.

In the case of *TiT and *TuT (in which T = any stop), it is quite clear that in Hittite there is no phonological difference between /TiT/ and /TuT/ on the one hand and /TyT/ and /TwT/ on the other. In the case of *ViV and *VuV, it is also clear that in Hittite there is no phonological distinction between /ViV/ and /VuV/ and /VyV/ and /VwV/. So in these environments it is not useful to distinguish between /i/ and /y/ and between /u/ and /w/. The question becomes more interesting when dealing with cases like *CuV / *CiV and *CuRC and *CiRC.

Let us first look at *CuV and *CiV. We may wonder if a form like la-ak-nu-an-zi, la-ak-nu-a-an-zi ‘they fell’ < *lgh-nu-énti is phonologically to be interpreted as /ləgnuántsi/, as /ləgnwántsi/ or even as /ləgnuwántsi/. The last option is impossible, since Hittite has a synchronic sound law that ‑u¾V- yields ‑umV-,[36] so we must choose between /ləgnuántsi/ and /ləgnwántsi/. It is clear that the latter option is impossible as well, since we then would have expected a phonetic realization [ləgnántsi] or [ləgənwántsi], spelled **la-ga-nu-a-an-zi (vocalization of ‑n- in between consonants). So we must conclude that /ləgnuántsi/ is the only correct phonological interpretation. It is likely, however, that the sequence /CuaC/ was phonetically realized with a glide [u̯], so [CuaC], but we must keep in mind that this glide did not have a phonemic status.

The case of ap-pa-at-ri-ez-zi is similar: should we analyze this as /ʔpadriétsi/, /ʔpadryétsi/ or /ʔpadriyétsi/? Although in principle the last option cannot be discarded (there are no indications that a sequence ‑i¤V- would undergo a phonetic change), it is inevitable that here as well we should choose for the analysis /ʔpadriétsi/. We can therefore conclude that in the case of *CuV and *CiV, the outcomes must be phonologically interpreted as /CuV/ and /CiV/ and not as **/CwV/ and **/CyV/.

The case of *CuRC, including *#urC and *Cur#, is very interesting. For instance, the suffix ar, which forms verbal nouns, always has the form ar, no matter if preceded by a consonant or a vowel. We will see s.v. that ar reflects *‑ur, however. The idea is that on the basis of postvocalic positions, e.g. *‑é-r or *‑sé‑r, the variant *‑r was generalized, also when following a consonant, e.g. inkuar. The question now is, does this form synchronically represent /Hínkwr/, or should we analyse it as /Hínkuər/? The latter form would show the position /CuV/ of which we have seen that there is no distinction between /CuV/ and /CwV/.

A similar question can be asked with regard to al-zi ‘to hit’. As we will see s.v., this verb must have undergone some levelling. The PIE paradigm *uélh3-ti, *ulh3‑énti should regularly have yielded **ualzi, *ullanzi, which is quite different from the attested forms: a-al-a-zi, a-al-a-an-zi. In order to explain these forms, we should assume the following scenario: (1) prevocalic *u is phonemicized as /w/: *uélh3ti > *élh3ti; (2) * spreads over the paradigm, replacing *ulh3énti by *¾©h3énti; (3) at the moment that interconsonantal laryngeals drop, *h3 is analogically restored in *élh3ti because of *¾©h3énti where it was retained; (4) *élh3ti, *¾©h3énti yields Hitt. a-al-a-zi, a-al-a-an-zi. As we see, in the prehistory of Hittite it is of crucial importance to assume a phonological difference between /w/ and /u/. The question is whether in synchronic Hittite this is the case as well. If 3pl. a-al-a-an-zi < *¾©h3énti is to be phonologically interpreted as /wlHántsi/, we should certainly assume a separate phoneme /w/, because /ulHántsi/ would have been spelled **ul-a-an-zi.[37] If however, a-al-a-an-zi is to be phonologically interpreted as /uəlHántsi/, as was suggested above (§ 1.3.7), we are dealing with a sequence *#uV, of which it is likely that it does not show a distinction between /#uV/ and /#wV/ (in analogy to *CuV).

Compare also the example of ú-ra-a-ni ‘burns’. As we will see s.v., this form reflects *urh1-óri, and I therefore phonologically interpret ú-ra-a-ni as /urʔā́ni/. From MH times onwards, this form is spelled a-ra-a-ni, however. Does this form represent /wrʔā́ni/, with a real /w/, or should we assume /uərʔā́ni/, with initial /uV/?

As we see, in cases where *u is adjacent to a syllabic resonant, the phonological interpretation is a matter of taste. If one wants, one could assume a phoneme /w/ in these positions, but I would rather analyse these cases as /uəR/, in which no distinction between /u/ and /w/ has to be made.

Summing up, I do not think that it is necessary to assume a phonological distinction between the semi-vowels /y/ and /w/ on the one hand and the real vowels /i/ and /u/ on the other.[38] I will therefore only use the vowels /i/ and /u/ in my phonological system (and consequently write /ViV/ and /VuV/ as well).

Note that with the elimination of phonemic /w/, the rule */uw/ > /um/ and */wu/ > /mu/ should be reformulated as */uuV/ > /umV/ and */VuuC/ > /VmuC/. For instance: /ʔau‑/ + /‑ueni/ > */ʔáuueni/ > /ʔáumeni/.[39]



1.3.9 Vowels


Because of the deficiency of the cuneiform script, the reconstruction of the Hittite vowel system is not easy.

As I stated above (§ 1.3.1), the fact that the script only contains signs with the value V, CV, VC and CVC makes it impossible to write word-initial or word-final consonant clusters or internal clusters of three or more consonants without writing vowels that are neither phonetically nor phonologically real,[40] e.g. /parHtsi/ ‘he chases’ which is spelled pár-a-zi as well as pár-a-zi in which the underlined a’s must be ‘empty’. Unfortunately, it is not always clear when a written vowel is real or empty or if we have to reckon with a difference between a phonetically real and a phonologically real vowel (cf. for instance the status of the spelling of ‑a- in reflexes of *CRC as discussed in § 1.3.7 above). It therefore can be informative to look at spellings of Hittite words in other languages. For instance, in the Old Assyrian texts from Kültepe (Neša / Kaniš),[41] we find the Hittite word išparuzzi- ‘rafter, roof batten’ attested as išpuruzzinnum, which points to a pronunciation [isprutsi‑], just as we would expect on the basis of its etymology, *spr-uti-; the (hypothetical) Hittite word *lauzzi- ‘vessel for pouring’ is attested as luuzzinnum, a vessel, pointing to [lhutsi] < *lh2u-uti-; the Hittite word aluka- ‘message’ is attested as ulugannum / ilugannum, pointing to [hluga‑] < *h2l(e/o)ugho‑. Although the OAss. words are attested in texts predating the Hittite texts with a few centuries, I do not see why these forms would not have been pronounced with initial clusters in synchronic Hittite as well. I would therefore interpret išparuzzi- as /ɨsprutsi‑/ and aluka- as /Hluga‑/. Plene spelling

A second problem we encounter is the practice of ‘plene spelling’, i.e. the extra writing of the vowel of a CV or VC-sign by its own separate sign, e.g. la-a-u-i, a‑aš-šu, ma-a-ar-ka-a-i. The function of plene spelling has been and still is a hotly debated topic in Hittitology. For an excellent overview of the views on plene spelling throughout the history of Hittitology, I refer to Kimball 1999: 54-68. It is very important to bear in mind that “[p]lene writing was never used with absolute consistency in texts of any period” and that “[a]s a general rule, plene writing is more frequent in early texts (texts in OH ductus and many MH texts) than it is in original compositions of the NH period” (Kimball 1999: 55).

In my view, plene spelling can have several functions. The most common function is to denote phonetic length of a vowel, e.g. ne-e-pí-iš in which the plene ‑e- denotes a long ē, which is the phonetically regular reflex of an underlying accented /e/ in open syllable. So ne-e-pí-iš denotes phonetic [né:pis] = phonological /nébis/.[42]

Although a long vowel is usually the result of accentuation, a plene spelled vowel cannot automatically be regarded as accented.[43] For instance, a word like la-a-u-a-a-i cannot have had two accents. In my view, it represents /lāhwā́i/, a secondary adaptation of original la-a-u-i = /lā́hwi/ into the productive tarn(a)-class.

In word-initial position, a plene vowel can denote an initial glottal stop, and does not necessarily indicate vowel length: e.g. a-ar-aš-zi = ’a-ar-aš-zi = /ʔárStsi/; e-eš-zi = ’e-eš-zi = /ʔéstsi/; a-a-an-ši = ’a-a-an-ši = /ʔā́nsi/, etc. In the case of ‑e- and ‑i‑, a plene vowel can also be used to disambiguate an ambiguous sign (see below).

It should be noted that the sequence îU-U- occurs so often in MS and NS texts in contexts where a long vowel would be unexpected that this plene spelling must be interpreted otherwise. Kimball (1983: 566-7) remarks that the signs îU and U in these texts are written close together as a ligature (BÉ), which would support Rosenkranz’ idea (1959: 420, 42610) that the writing of U is used to more clearly distinguish the sign îU (B) from the closely resembling sign RI (J). Since such a disambiguation could have been achieved by writing îU-Ú- (BS) as well, which is virtually never attested, there must have been additional reasons to write îU-U‑. Below it will be argued that this sequence denotes /Ho/, and that the sign U indicates the phoneme /o/ here.

It is important to realize, however, that no theory about plene spelling will be able to explain every single instance of plene spelling as attested in the Hittite texts. For instance, in my text files, the word ta-ga-a-an ‘on the earth’ occurs spelled thus 30 times (of which 5 times in OS texts), as ta-ga-an 3 times (once in an OS text), as ta‑ka-a-an once, as da-ga-a-an 21 times, and as da-ga-an 7 times. These spellings can safely be phonologically interpreted as /tgā́n/, the phonetically regular reflex of an endingless *dhœh-Öm. Nevertheless, in NH texts, we find three aberrant spellings, namely da-a-ga-an (KUB 43.17, 6 (NH)), ta-a-ga-an (KUB 34.120, 7 (NH)) and da-a-ga-a-an (KUB 40.46, 9 (NH)), all with a plene vowel ‑a- where we would not expect it. Especially the third spelling, da-a-ga-a-an, is remarkable because of its two plene spellings. One could offer several ad hoc solutions in order to explain these spellings,[44] but the fact is that aberrant spellings exist and one must accept that they are not always explicable in an orthographic or phonetic sense. E/I-Ambiguity

Many signs are ambiguous regarding their vocalic value: they can be read with either ‑e- or ‑i-.[45] The only unambiguous signs are E, I, TE, TI, îÉ (but îI can be read îE as well), ME, MI (which in principle can be read MÉ as well), NE, NI (which in principle can be read NÉ as well), ŠE, ŠI, ZÉ (but ZI can be read ZE as well), EL, IL, EN, IN, EŠ, IŠ, MEŠ and MIŠ.

When an ambiguous sign is used together with an unambiguous sign, we can read the vowel of the unambiguous sign (e.g. KE/I-eš-šar = ke-eš-šar = /kéSr/ ‘hand’), but this is not always the case (e.g. ar-KE/I-E/IR can in principle be read ar-ki-ir, ar-ke-er, ar-ki-er and ar-ke-ir). Fortunately, sometimes we are offered a helping hand by plene spellings that indicate the appropriate vowel (in this case, the spelling ar-KE/I-e-E/IR, which must be read as ar-ke-e-er, shows that ar-KI/E-E/IR must be read ar-ke-er /Hárger/ ‘they perished’).

Because of the complicated situation regarding the spelling of the vowels e and i, it is not always easy to distinguish between these vowels on a phonological level either. This has led some scholars to the idea that within the Hittite period the vowels e and i are merging. For instance, CHD L: xvi states that “[i]t is well-known that the vowels e and i often interchange in the spelling of Hittite words. It is quite likely that the two vowels, still kept distinct in Typical Old Script, began to merge in later Old Hittite, and certainly had completed their merger by the Empire period”.[46] Melchert (1984a: 78-156) has carefully examined the spelling and phonemic status of e and i throughout the Hittite period and arrives at a different conclusion, however, namely that “[t]he vowels /e/ and /i/ are phonemically distinct at all stages of Hittite. Any mergers or free variation between the two are conditioned”.

Nevertheless, Kimball (1999: 78-9) states that despite Melchert’s statements “[t]he evidence is consistent with a phonemic distinction between /ē/ and /ī/ in the earliest language that was lost through merger by the NH period”. She even goes as far as claiming that “[e]ven the limited variation in OH texts may indicate the beginning of merger; or it may point to the existence of a scribal tradition predating the OH texts of Boğazköy, suggesting that Hittite was first committed to writing at a time somewhat before the date of the earliest texts that have been recovered when the language did distinguish high and mid front vowels, but that even by the time the Boğazköy texts in typical old ductus were written that distinction was on its way to oblivion”. She bases her view on spellings like i-eš-zi ‘he is’ (KUB 34.115 iii 5 (OS)) instead of normal e-eš-zi, which she calls “[c]ompelling evidence for merger”. In my view, however, taking this attestation[47] as more significant than the more than 1400 examples in my text files (ranging from OS to NH texts) of attestations where the verb ‘to be’ is consistently spelled with an initial e‑, is undesirable.[48]

In this book I therefore have made a phonological distinction between /e/ and /i/ for all periods of Hittite. It should be noted, however, that several environments can be identified in which OH /i/ is regularly lowered to /e/ from the MH period onwards, cf. Moreover, there are several instances where indeed a spelling -e- alternates with -i-, but these cases are to be regarded as showing the epenthetic vowel /ɨ/, for which see § Plene spelling of E and I

Since the vowel signs E and I can be used to disambiguate an ambiguous sign, it is not always clear whether their use can be interpreted as indicating length. For instance, the spelling ar-ke-e-er, as we saw above, hardly reflects /Hárgēr/, but rather /Hárger/ < *h3érg-ēr, which means that its plene E is used to disambiguate the signs KE/I and E/IR; pí-i-ú-e-ni ‘we give’ cannot denote /pīuéni/, but must stand for /piuéni/ < *h1p-i-éni, which shows that the plene I is used to disambiguate the sign PÉ/Í.

Nevertheless, there remain some forms in which the plene E or I can hardly have been used for disambiguation. For instance, in še-e-er ‘above’, the unambiguous sign ŠE would have been enough to disambiguate the ambiguous sign E/IR (and the spelling še-er therefore does occur as well), so the plene E in that sense is superfluous. Similar, and more clear, are the cases of te-e-eš ‘you said’, where both TE and EŠ are unambiguous signs, še-e-eš ‘sleep!’, where ŠE and EŠ are unambiguous signs and ne-e-pí-iš ‘heaven’, where NE is unambiguous. As we will see below under the treatment of the reflexes of PIE *e, *ē, *ei and *eh1 (§, §, in accented position these vowels all yield Hitt. /é/ which is spelled plene in open syllables and in monosyllabic words and therefore probably was phonetically long in these positions.

Plene spellings of the type Ci-i-iC are quite rare, but do occur: u-ur-ki-i-il ‘perversity’, li-i-ik ‘swear!’, na-ak-ki-i-iš ‘important’, ni-i-ik ‘quench!’, zi-i-ik ‘you’. Although some of these cases seem to show an underlying short *i that is accented and therefore lengthened,[49] some seem to show a real accented long /ī/.[50] The signs U and Ú

Hittite uses two phonetic signs that are traditionally transliterated with the vowel u, namely É = U and S = Ú.[51] From the beginning of Hittitology, it has been noticed that in many words these two signs are kept distinct. For instance, lūli- ‘pond’, when spelled with a plene vowel, is consistently spelled lu-ú-li- and never **lu-u-li-; kūša- ‘daughter-in-law’ and its derivative kūšāta- ‘bride-price’ are always spelled ku-ú-šº and never **ku-u-šº; ūmant- ‘all, every’ is consistently spelled u-u-ma-an-t- and never **u-ú-ma-an-t‑, etc. It therefore has been proposed that these two signs represent phonologically distinct sounds. Already Weidner (1917: 2-13) suggested that the sign U indicates the sound [o] and the sign Ú the sound [u]. Such a distinction is not unparalleled in cuneiform traditions: it is known from Hurrian (cf. Wegner 2000: 37), but also from e.g. some Old Babylonian lexical lists from Nippur (Westenholz 1991). Despite some claims in favour of this interpretation,[52] it has never gained a broad acceptance.[53] Most recently, Rieken (2005) has attempted to revive this theory, however. According to her, the sign U denotes a vowel /o/ that is the result of lowering of an older u in certain phonetic environments.[54] She assumes that the vowels /u/ and /o/ originally were allophones, but were marginally phonemicized in Hittite. Although the bulk of Rieken’s observations seem correct to me, I do not agree with all details.[55]

An important clue regarding the idea that U and Ú could reflect different sounds is the fact that the preverb u- ‘hither’ (the antonym of pe- ‘thither’) is spelled with both U and Ú, but that the choice for one of these signs is always consistent within the attestations of each verb. We come accross the following spellings: u-uC-C° (in ūnna-i / ūnni- ‘to drive (here)’), uC-C° (in uppa-i / uppi- ‘to send (here)’), ú-uC-C° (in ūššie/a-zi ‘to draw open (curtains)’) and ú-CV° (in uda-i / ud- ‘to bring (here)’). At first sight, we seem to be dealing with three different spellings, namely u-uC-C°, uC-C° and ú-uC-C° (assuming that ú-C° is equivalent to ú-uC-C°). It must be noted, however, that the only verb that is spelled ú-uC-C°, namely ūššie/a-zi, occurs as uš‑ši- as well. Moreover, the spelling ú-uš-ši- occurs in OS texts only, whereas the spelling uš-ši- is attested in MS and NS texts. Since the only verb that is consistently spelled uC-C°, uppa-i / uppi‑, is not attested in OS texts, but only in MS and NS texts, it is in my view quite likely that this verb must be compared to ūššie/a-zi, and that we are allowed to assume that in OS texts this verb would have been spelled **ú-up-p°.[56]

So in fact we are dealing with two different spellings, namely ú-uC-C° (OS) = uC‑C° (MS and NS) = ú-C° versus u-uC-C°. Since these spellings must eventually go back to the same etymon, namely *h2ou‑, I agree with Rieken that some phonetically conditioned split must have taken place. Apparently, *h2ou- developed into two different forms, one spelled with the sign U and the other with Ú.

For a phonetic interpretation of the difference between U and Ú, we should look at the paradigm of au-i / u- ‘to see’ in comparison to the dāi/tianzi-class verbs, exemplified here by pai-i / pi- ‘to give’:

            1sg.     u-u-i          < *Hóu-h2ei              pé-e-e-i     < *h1pói-h2ei

            2sg.     a-ut-ti             < *Hóu-th2ei             pa-it-ti           < *h1pói-th2ei

            3sg.     (a-uš-zi)                                             pa-a-i             < *h1pói-ei

            1pl.     ú-me-e-ni       < *Hu-uéni                pí-ú-e-ni        < *h1pi-uéni

            2pl.     uš-t[e-e]‑ni[57] < *Hu-sténi               pí-iš-te-ni      < *h1pi-sténi

            3pl.     ú-a-an-zi     < *Hu-énti                 pí-a-an-zi    < *h1pi-énti

We clearly see that the spelling with U corresponds to e- in the paradigm of pai‑i / pi‑, whereas Ú corresponds to i‑. On the basis of this comparison alone, it is attractive to assume that U stands for /o/, whereas Ú stands for /u/. The fact that this outcome perfectly matches the Hurrian practice to spell /o/ with U and /u/ with Ú strengthens this interpretation even more. I therefore phonologically interpret the above forms as follows:

            u-u-i          = /ʔóHi/,                  cf.         pé-e-e-i     = /péHi/

            a-ut-ti             = /ʔáuti/                                 pa-it-ti           = /páiti/

            (a-uš-zi)                                                         pa-a-i            

            ú-me-e-ni       = /ʔuméni/ < */ʔuuéni/       pí-ú-e-ni        = /piuéni/

            uš-t[e-e]‑ni    = /ʔusténi/                             pí-iš-te-ni      = /pisténi/

            ú-a-an-zi     = /ʔuántsi/                             pí-a-an-zi    = /piántsi/

This means that the u-preverbed verbs as mentioned above must be phonologically interpreted as follows: ūnna-i / ūnni- ‘to drive (here)’, spelled u‑un‑n°, = /ʔoNa/i‑/, ūššie/a-zi ‘to draw open (curtains)’, spelled ú-uš-ši- and uš-ši‑, = /ʔuSié/á‑/, uppa-i / uppi- ‘to send (here)’, spelled up-p°, = /ʔupa/i‑/, and uda-i / ud- ‘to bring (here)’, spelled ú-d°, = /ʔud(a)-/.

In the following sections I will study the use of the signs U and Ú in specific phonetic environments, in order to determine (1) if a complementary distribution between U and Ú can be established for this environment, and if so, (2) how we should interpret this distribution phonetically and historically. Word-initially before vowels

_aC : Here we basically find only the spellings a- and ú-a‑. The spellings ú-a° and u-a° are extremely rare,[58] whereas the spelling u-a- occurs in the middle paradigm of au-i / u- ‘to see’ only.[59] As I stated in § 1.3.3 as well, I believe that the spelling a- reflects phonological /ua-/ (e.g. a-a-tar /uā́dr/ ‘water’ < *uódr);[60] the spelling ú-a- = /ʔua-/ (e.g. ú-a-a-tar /ʔuā́dr/ ‘inspection’ < *Huótr;[61] the spelling u-a- represents /ʔoa-/ (e.g. u-a-a-a-at /ʔoaHat/ ‘I have become visible’, cf. au-i / u- for treatment).


_eC : Here we only find the spelling ú-e‑, which denotes /ue-/ (e.g. ú-e-ek-zi /uéktsi/ ‘wishes’ < *ti).


_ɨC : Here we find the spellings ú-e°, ú-i° and ú-i5, which can stand for both /uɨ‑/ as well as /ʔuɨ-/. For instance, ú-i-te-e-ni, ú-e-te-ni ‘to the water’ = /uɨdéni/ < *déni << *udéni and ú-i5-te-na-aš ‘of the water’ = /uɨdénas/ < *dénos << *udéns, whereas the spellings ú-iš-ke/a‑, ú-i-iš-ke/a- and ú-e-iš-ke/a‑, imperfectives of e-zi / ua- ‘to come’, must stand for /ʔuɨské/á-/ (a synchronic derivation of the stem /ʔué/á-/).


_iC : Here I only know of the spellings ú-i° and i5- that always stand for /uiC‑/: ú‑i‑it-t° and ú-it-t° ‘year’ stand for /uit‑/ < *uet-; ú-i-t° and i5-t° stand for /uid‑/ ‘water’ < *uedo‑.

Summarizing, in absolute word-initial position before vowels (note that ú-a- = /ʔua‑/ and u-a- = /ʔoa‑/ in fact belong to word-internal position), there is no distinction to be found between /u/ and /o/. Word-initially before consonants

There are only a few examples here.[62] The verb ur-āri, which reflects *urh1óri, is in OS texts consistently spelled ú-ra-a-ni, pointing to /urʔā́ni/.[63] The verb ušnie/a-zi is always spelled uš-(ša‑)ni‑. I see no reason not to interpret this verb as /uSnie/a-/ < *usn-e/o‑.

As we see, there is no trace of a distinction between /u/ and /o/ in absolute word-initial position before consonants. Word-internally between consonant and vowel

C_a : The spelling Cu-a° is especially attested in older texts (e.g. ar-nu-an-da-an = /ʔrnuántan/). The spelling Cu-a° is the most common spelling, especially in younger texts (e.g. ar-nu-a-an-zi = /ʔrnuántsi/). The spelling Cu‑u‑a° is quite rare, but does not seem to stand for the same phonetic value as Cu-a° and Cu-a- (e.g. ar-ku-u-ar = ar-ku-ar = ar-ku-a-ar = /ʔrkwuər/). This spelling occurs quite often when the sign îU precedes, the reason for which we will see below,[64] e.g. u-u-ap- = u‑a-ap‑. The spelling Cu-ú-a° only occurs in ka-ru-ú-a-ri-a-ar, which is a secondary form (see s.v. kareariar). The spelling Cu-u-a- is rather uncommon and is also predominantly found with a preceding sign îU (e.g. u-u-a-ap‑, u-u-a-an-za, iš-u-u-a-i, etc.). Other examples are: kap-pu-u-a-an-zi (= kap-pu-a-an-zi), kar-šu-u-a-aš (= kar-šu-a-aš), ka-ru-uš-ši-a-nu-u-a-an-zi (= ka-ru-uš-ši-a-nu-a-an-zi), etc. The spelling Cu-ú-a- is rare: it is attested in a-ru-ú-a-iz-zi (= a-ru-a-iz-zi and a-ru-u-a-iz-zi), ka-ru-ú-a-ri-a-ar (which is a secondary form, see s.v. kareariar), ša-ak-ru-ú-a-an-zi (= ša-ak-ru-a- and ša-ak-ru-u-a‑), šu-ú-a-i° ‘to spy’ (= šu-a-i°), šu-ú-a-ru- ‘heavy’ (= šu-a-ru and šu-u-a-ru) and šu‑ú-a- ‘to push’ (= šu-a- and šu-u-a‑).

I conclude that the spellings Cu-a°, Cu-a‑, Cu-u-a° and Cu-u-a- are equivalent and denote phonological /Cua/. The spelling Cu-ú-a° as found in ka-ru-ú-a-ri-a-ar is unique and is probably orthographically influenced by ka-ru-ú. The interpretation of the spelling Cu-ú-a- is less clear since it is quite rare and the etymological interpretation of the words in which it occurs is controversial. Nevertheless, on the basis of the fact that a-ru-ú-a-iz-zi is also spelled a-ru-a-iz-zi and a-ru-u-a-iz-zi, I conclude that in the sequence C_a the difference between the sign U and Ú does not denote a distinction between /o/ and /u/.

Note however, that as we have seen in § above, there is a distinction in the sequence #ʔ_a, namely ú-a- = /ʔua‑/, e.g. ú-a-a-tar /ʔuā́dr/ ‘inspection’ < *Huótr, and u-a- = /ʔoa‑/, only attested in the middle paradigm of au-i / u- ‘to see’, e.g. u-a-a-a-at /ʔoaHat/ ‘I have become visible’. Since these middle forms are recently created (see s.v. au-i / u‑), the phonemic difference between /ʔua-/ and /ʔoa-/ must be a recent innovation as well.


C_e : The spellings Cu-e°, Cu-u-e° and Cu-ú-e° are all used in equal environments, which shows that they should be regarded phonologically equal as well: e.g. ak-ku-e-ni = ak-ku-u-e-ni = /ʔkuéni/ ‘we die’, [a]p-pu-ú-e-ni = e-ep-pu-e-ni = e-ep-pu-u-e-ni = /ʔpuéni, ʔépueni/ ‘we grab’, a-aš-šu-ú-e-ni = a-aš-šu-e-ni = /HSuéni/ or /HəSuéni/ ‘we open’, še-ek-ku-e-ni = še-ek-ku-u-e-ni = še-ek-ku-ú-e-ni = /sékueni/ ‘we know’, etc. Again, there is no indication that the signs U and Ú denote a difference between /o/ and /u/ in this environment.


C_i : The spellings Cu-i°, Cu-u-i°, Cu-ú-i°, Cu-i5- and Cu-u-i5- are used in equal surroundings, e.g. a-ap-pu-i = a-ap-pu-u-i = a-ap-pu-ú-i /Hapui/, pár-ku-iš = pár-ku-u-iš = pár-ku-ú-iš! = /prkwis/; pa-aš-šu-i = pa-aš-šu-u-i = pa-aš-šu-ú-i = pa‑aš-šu-i5 = /paSui/; u-i-ša-u° = u-u-i-ša-u° = u-u-i5-ša-u° = /Hwisau‑/, which shows that in this position the signs U and Ú do not represent distinct phonemes. The spelling Cu-u-i° occurs especially often when the sign îU precedes (see below). The form ka-ru-ú-i-li- ‘former’ is a synchronic derivative in ili- of the adverb ka-ru-ú ‘early, formerly’ and therefore probably represents /krÜili/. The words šūil ‘thread’ and mūil ‘spade(?)’ are treated under C_ʔ.


C_u : The only word that seems to belong here, viz. šūu- / šūa- ‘full’, in fact reflects *souH-u- and therefore will be treated under C_ʔ. Word-internally between vowels

a_a : We find the spellings °a-u-a°, °a-a‑, °a-ú-a- and °a-u-a- in equal positions: e.g. a-ra-u-aš = a-ra-u-a-aš = a-ra-a-aš = /ʔarauas/; ar-na-a-u-aš = ar-na-a-a-aš = ar-na-ú-a-aš = ar-na-a-u-a-aš = /Hrnā́uas/. It must be admitted, however, that the spelling °a-ú-a- is quite rare, and seems to have a special function in the paradigm of auri- ‘look-out’. Here we find a-ú-ri-a-aš besides a-ú-a-ri-a-aš for phonological /ʔaurias/ < *Hou-ri-os, which could either be phonetically realized as [ʔaurijas] spelled a-ú-ri-a-aš, or as [ʔawrjas] spelled a-ú-a-ri-a-aš. The sequence °a-ú-a° is only attested in KBo 30.51 iv? (1) [...(‑)g]aʔ‑ú‑a[n(‑)...], if this is the correct reading.


a_e : In this position we predominantly find the spelling °a-u-e°. The spelling °a‑ú‑e° is rare, but when attested, it is identical to °a-u-e°: a-aš-ša-ú-e-et = a-aš-ša-u-e-et; al-zi-a-ú-en = al-zi-a-u-en; compare [z]i-in-na-ú-e-ni to e.g. a-ri-a-u-e-ni.[65]


a_i : Although the spellings °a-ú-i° and °a-u-i° are occasionally interchangeable (e.g. ar-na-(a‑)ú-i (often) = ar-na-u[‑i] (1x); i-ta-a-la-ú-i (1x) = i-da-a-la-u-i (often)), some words are consistently spelled °a-ú-i°: e.g. i ‘not yet’ is spelled na-a-ú-i (OS), na-ú-i (OH/NS), na-a-i5 (OH?/NS, MH/NS), na-i5 (NH) and na-u-i5 (OH?/NS), but never **na-a-u-i; šāidišt- ‘yearling’ is spelled ša-a-ú-i° and ša‑ú-i°, but never **ša-a-u-i° or **ša-u-i°; SIšāītra- ‘horn’ is spelled ša-a-ú-i° and ša-ú-i° but never **ša-a-u-i° or **ša-u-i°. It is remarkable that Ú almost consistently occurs when a long /ā/ preceeds, whereas U is used after a short /a/. So Ca-a-ú-i° = Ca-ú-i° = /Cāui°/, whereas Ca-u-i° = /Caui/ (or /Caoi/?). The reason for this distribution is unclear to me.


e_a : Here we predominantly find the spellings °e-a‑, °e-u-a° and °e-u-a‑, which are interchangeable: me-mi-iš-ke-a-an = me-mi-iš-ke-u-an = me-mi-iš-ke-u-a-an = /memiskéuən/; e-a-an = e-u-a-an = /ʔéuan/; ne-e-a-an = ne-e-u-a-an = /néuan/. The spelling °e-ú-a- only occurs in ka-re-ú-a-ri-a-ar, which is spelled ka-re-a-ri-a-ar as well, and in [k]u-re-ú-a-nu-uš, which is spelled ku-re-e-a-n° and ku‑re-a-n° as well. The spelling °e-ú-a° is not attested at all.


e_e : In this position we only find the spelling °e-u-e°: ku-e-u-e-en ‘we killed’, da‑aš-ke-e-u-e-ni ‘we are taking’, é-e-u-e-eš ‘rains’. The spelling °e-ú-e° to my knowledge does not occur.


e_i : The only cases known to me are ne-e-u-it (instr.) ‘new’ and ú-e-u-iš-ke-u-an (KBo 24.5 ii 10) ‘crying’. The spelling °e-ú-i° does not occur.


i_a : The normal spelling is °i-a‑, e.g. i-a-ar, a-ši-a-an-t‑, mi-iš-ri-a-an-t‑, ka-ru-ú-a-ri-a-ar, etc. Very rarely, we find the spellings °i-u-a- (a-aŸ-li-u-a-an-za, na-an-ni-u-a-an-zi, ti-u-a-an-zi, ka-ru-ú-a-a-ri-u-a-ar), °i-ú-a- (me-mi-ú-a-an-zi, ši-ú-a-at-ti = ši-a-at-ti, ar-ki-ú-a-az = ar-ki-i-ú-az), °i-ú-a° (ar-ki-i-ú-az = ar-ki-ú-a-az) and °i-u-a° (zi-in-ni-u-an-zi, ka-ru-ú-a-ri-u-ar). There is no indication against the assumption that all these spellings denote phonological /Ciua/.


i_e : We find both the spelling °i-ú-e° and °i-u-e° in the same environments, although °i-ú-e° seems to occur more often than °i-u-e°: me-mi-u-e-ni, mi-iš-ri-u-e-eš-zi vs. al-zi-ú-en, [u-et‑]ti-a-an-ni-ú-e-ni, mi-im-mi-ú-en, pí-i-ú-e-ni, pí-ú-e-ni, pé-en-ni-ú-e-ni, etc. Once we even find °i-ú-u-e°, viz. in pa-i-ú-u-en (KBo 3.60 iii 1). It is clear that all spellings denote /Ciue/.


i_i : This position is not well attested in native Hittite words. We find a spelling °i‑ú‑i° in ú-i-ú-i-iš-ke/a‑, the imperfective of ia-i / ii- ‘to scream’ and in i5-ú-i-da-a-i (KBo 5.4 rev. 29) = ú-i-i5-ta‑[a-i] (KBo 5.4 rev. 36). Other cases are the city name URUKu-li-ú-iš-na (also spelled URUKu-li-i5-iš-na), Éar-ki-ú-i ‘vestibule(?)’ and a-az-zi-ú-i ‘ritual’ (also spelled a-az-zi-i5). A spelling °i-u-i° is found in [É]a-le-ti-u-i and [É]a-le-en-ti-u-i, of Éalent(i)u- ‘palace’ only.[66] It is quite possible that in these words the spelling °i-ú-i° and °i-u-i° represent phonetically different forms. Since the stem of Éalent(i)u- is consistently spelled a-le-en-ti-u, it probably was /Halentio/. This makes it likely that the spelling a-le-en-ti-u-i stands for /Halentioi/. It must be noted that Éalentiu- is not a native Hittite word, and that we have no evidence for other instances of a sequence /ioi/. Word-internally between vowel and consonant

a_C : First, we should distinguish between a_CV and a_C# / a_CCV: the former must be spelled °a-U-CV or °a-Ú-CV, whereas the latter can be spelled °a-uC(-CV).

In the case of a_CV, we find many words that show a consistent spelling °a-ú-CV, e.g. a-ú-me-ni ‘we see’, a-ú-me-en ‘we saw’ (never **a-u-me‑), a-ú-ri- ‘lookout’ (never **a-u-ri‑), a-ú-li‑, a certain organ (never **a-u-li‑), an-na-ú-li- ‘of equal rank’ (never **an-na-u-li‑), ša-ú-di-iš-t°, ša-a-ú-ti-iš-t° ‘weanling’ (never **ša(‑a)‑u-Ti‑). In some other words, we do find both U and Ú, however, e.g. pár‑ta-ú-na-aš = pár-ta-u-na-aš, a-ša-ú-ni = a-ša-u-ni, e.a. It is remarkable that this situation occurs in front of n- only, and that there seems to be a chronological distribution between the forms: in OS texts we only find °a-ú-n°,[67] in MS texts predominantly °a-ú-n° and occasionally °a-u-n°,[68] in NS texts predominantly °a‑u‑n°.[69] This seems to point to a change of OH °a-ú-n° to NH °a-u-n°, which then must be phonologically interpreted as OH /Caun/ > NH /Caon/.

In the case of a_C# / a_CCV, the situation is less clear, mainly because the number of plene u-spellings is so low. It is perhaps best to look at the cases one by one. The spelling °a-u-uC(-CV) is found in the following forms:

pa-a-u-un ‘I went’: this spelling is found several times, but only in NS texts, and contrasts with the spelling pa-a-ú-un that is found in MS texts. The neutral spelling pa-a-un, without a plene u-vowel, is attested in OS, MS and NS texts. In my view, the spelling change of pa-a-ú-un > pa-a-u-un again points to the change of OH and MH /pā́un/ to NH /pā́on/ (cf. above).

i-a-u-un ‘I did’ (KBo 4.10 obv. 50) can hardly be correct and must probably be emended to i-a-nu!-un. [al‑]lu-ú-a-u-uš (KBo 3.8 iii 4 (OH/NS)), al-lu-a-u-uš (KBo 26.135, 2 (OH/NS)) and pár-ga-u-uš (KBo 3.8 iii 22 (OH/NS)) are to be regarded as grammatically incorrect forms showing ‑a instead of correct ‑amuš as attested in e.g. al-lu-a-mu-uš (KBo 12.86 obv. 19, KUB 17.10 i 26, etc.) and pár-ga-mu-uš (KUB 17.10 i 24, KUB 12.63 i 30, etc.). Since the sign U is used ‘intervocalically’ here, the forms are irrelevant for our discussion.

ta-a-a-ta-u-uš-ša-aš (KBo 25.112 ii 7 (OS)) is a hapax of non-IE origin (cf. the single ) and therefore irrelevant here.

da-ra-a-u-ur (KBo 22.186 v 2 (OH/NS)) ‘handful(?)’ stands in development between ta-ra-a-ur (KBo 17.74+ i 53 (OH/MS)) and ta-ra-a-u-a-ar (KUB 44.64 i 5, 10 (NS)).[70] The first two spellings point to a phonological interpretation /trā́or/, whereas ta-ra-a-u-a-ar = /trā́uər/. In my view, the word /trā́or/, for which I know no convincing etymology, and which may be of a foreign origin, has been reinterpreted as an abstract noun in ar, and secondarily substituted by /trā́uər/ in younger times.

ti-e-ra-u-ur-ta-an (KBo 3.2 lower edge 2, KBo 3.5+ ii 37) and ti-e-ru-u-ur-ta-an-na (KBo 3.5+ iii 17) ‘for three laps’ is also spelled ti-e-ra-a-ar-ta-an-na (KBo 3.2 obv. 65) and ti-e‹-ra›-a-ar-ta-an-na (KUB 1.11+ iv 35). Since it is of foreign origin (< Indo-Aryan *tri-vartana-), it is irrelevant here.

The spelling °a-ú-uC(-CV) is found in the following forms:

a-ú-um-me-ni ‘we see’ and a-ú-um-me-en ‘we saw’ are clearly NH adaptations of older a-ú-me-ni and a-ú-me-en.

a-ú-uš-ta (KBo 3.60 i 8 (undat.)) ‘he saw’ is a combination of the normal spelling a-uš-ta and other forms of the verb au-i / u- that are spelled a-ú- (like a-ú-me-ni and a-ú-me-en above).

ar-na-a-ú-uš (KUB 9.22 ii 40) is a mistake for ar-na-iš ‘sap’, and therefore irrelevant.[71]

ar-na-ú-un (ABoT 17 ii 9 (MH/NS)) seems to denote /Hrnā́un/. Although this is not impossible in a NS text (especially since it is a copy of a MH text), we would have rather expected /Hrnā́on/, spelled **ar-na-u-un. NINDAar-ša-ú-uš (KBo 17.4 ii 17 (fr.) (OS), KUB 7.8+ ii 11 (NS)) ‘thick-bread’ is equivalent to NINDAar-ša-uš and NINDAar-ša-a-uš and must represent /HárSāus/ < */HárSaius/.

iš-u-na-ú-uš (KBo 32.14 ii 49 (MH/MS)) ‘upper arm’ is equivalent to iš-u-na-a-uš (KBo 32.14 rev. 44, l.edge 1 (MH/MS)), iš-u-na-uš! (text: ‑aš, KUB 9.34 ii 25 (MH/NS)) and must represent /ɨsHonā́us/ < *sh2u-nÖus.

la-a-ú-un (KUB 7.1 iii 20 passim) is a mistake for 3sg.imp.act. la-a-ú[72] and therefore irrelevant here.

pa-a-ú-un (KBo 16.59 rev. 5 (MS), KBo 16.42 obv. 24 (MS), KUB 34.45 + KBo 16.63 obv. 13 (MS)) represents /pā́un/, which in younger times phonetically changed to /pā́on/, spelled pa-a-u-un (cf. above).

ta-lu-ga-ú-uš (KBo 17.22 iii 6 (OS)) is equivalent to the spellings da-lu-ga-uš and ta-lu-ga-uš and represents /talugāus/ < */talugaius/.

So we can conclude that the diphthong /au/ is lowered to /ao/ before /n/ from MH times onwards, but is preserved as such in other positions.[73]



First I will treat the words that show a spelling °i-ú-CV or °i-ú-uC:

a-ni-ú-úr and a-ni-ú-ri are occasional spellings for normal a-ni-u-ur and a-ni-u-ri. See s.v. aniūr below.

aš-a-i-ú-ul (KUB 24.10 iii 18, KUB 24.11 iii 17) // aš-a-i-ú-úr (KBo 21.8 iii 6) // a-aš-a-ú[‑...] (KBo 12.126 rev. 14) is of unclear meaning. Since this word can hardly be of native origin,[74] it is irrelevant here.

é-i-ú-un (KBo 3.7 ii 25 (OH/NS)) is a hapax spelling for normal ēun ‘rain’ and therefore will be treated below under e_C.

imiūl (n.) ‘grain mix, horse feed’ is consistently spelled with Ú: i‑mi‑ú‑l=a-a=š-ma-aš (KUB 29.41, 8 (MH/MS)), i-mi-ú-ul (KBo 12.126 i 29 (OH/NS)), im-mi-ú-ul (KBo 4.2 ii 33 (OH/NS), KUB 7.54 ii 17 (fr.) (NS)), im-mi-i-ú-ul (KBo 10.37 ii 15 (OH/NS)). These spellings point to /imiÜl/ < *im-ié-ul.

iūl (n.) ‘binding, treaty’ and its derivative iulaŸŸ-i ‘to bind by treaty’ are consistently spelled with Ú: iš-i-ú-ul, iš-i-ú-la-aš, iš-i-ú-li, 3pl.pres.act. iš-i-ú-la-a-a-an-zi, part. iš-i-ú-la-aŸ-Ÿa-an-t‑. These spellings point to /ɨsHiÜl/ < *sh2-ié-ul.

iūk, iuka- (n.) ‘yoke, pair’ and its derivatives iuga- ‘yearling’, iugašša- ‘yearling’ and uga- ‘two-year-old’ are always spelled with Ú: i-ú-uk (KBo 25.72 r.col. 11 (OS)), i-ú-kán (KBo 12.22 i 11 (OH/NS), KBo 12.131 r.col. 5 (OH/NS), KUB 31.4 + KBo 3.41 obv. 7 (OH/NS)), i-ú-ga-an (KBo 13.78 obv. 2 (OH/NS), KUB 7.8 ii 8 (MH/NS)), i-ú-ki (KUB 13.5 ii 21 (OH/NS)); i-ú-ga-aš (OS), i!!-ga-an (text: ú-i-ga-an, KBo 17.65 rev. 53 (MS)), i-ú-ga-aš, i-ú-ga-aš; gen. pl. i-ú-g[a-aš]‑ša-a[n] (OS), i-ú-ga-aš-ša-aš (OH/NS)); ta-a-i-ú-ga-aš (OS), ta-a-ú-ga-aš (OH/NS), ta-a-i-ú-ga-aš (OS), ta-a-i-ú-ga-aš. All these spellings point to /iug‑/ < *iug‑. kap-pí-ú-uš (KBo 34.47 ii 8 (MH/MS)) of kappi- / kappai- ‘small’ is a younger adaptation of original kap-pa-uš (KUB 12.63 obv. 31 (OH/MS)) < *kappa. So kap-pí-ú-uš must stand for /kapius/.

TÚGka-ri-ú-ul-li ‘hood’, also spelled ka-ri-ul-li, is a derivative in ‑ulli- of karie/a-zi ‘to cover’ and represents /kriÜLi/ < *kr-ié-ul+. ku-i-ú-uš (HKM 23 obv. 9 (MH/MS), KBo 18.57a + 57 obv. 2, rev. 42 (MH/MS)) of the interrog. / indef. pronoun kui- / kue- / kua- is usually spelled ku-i-uš and stands for /kwius/. ma-ši-ú-u[š]? (KBo 9.109 rev. 4) of maši- ‘how many’ represents /masius/.

mīu- / a- (adj.) ‘soft, mild’ and its derivatives mīumar ‘gentleness’ and NINDAmīumīu(t)- “soft bread” are always spelled with Ú:[75] mi-i-uš = mi‑i-ú-uš = mi-ú-uš = /mÏus/ < *mih1-u-s, mi-i-ú-un = /mÏun/ < *mih1‑u‑m, mi-i-ú = mi-ú = /mÏu/ < *mih1-u; mi-i-ú-mar = mi-ú-mar = mi-ú-um-mar = /mÏumr/ < */mÏuur/, instr. mi-ú-um-ni-it = /mÏumnit/ < */mÏuunit/; mi-ú-mi-ú (MH/NS), mi-i-ú-mi-u=š-ša-an, mi-i-ú-mi-i-ú = /mīumīu/, etc.

pár-ši-ú-ul-li ‘crumb’ is derived from paršie/a-ari ‘to break’ and represents /prSiÜLi/ < *bhrs-ié-ul+. pu-u-ri-ú-uš of pūri- ‘lip, ridge’ equals pu-u-ri-uš and pu-ri-uš and represents /pórius/.

šīu- (c.) ‘god’ and its derivatives šīuna- ‘god’, šiunal(a/i)- ‘divine one(?)’ and šiuniaŸŸ-tta(ri) ‘to be hit by a disease’ are always spelled with the sign Ú: ši-i-ú-uš (OS), ši-ú-uš, ši-uš=mi-iš (OS) = /sÏus/, ši-ú(n)=šum-m[i-in] (OS), ši‑ú‑n=a-an = /sÏun/, ši-ú-na-aš (OS) = /sÏunas/, ši-ú-ni (OS), ši‑i‑ú-ni (OH/MS) = /sÏuni/, abl. ši-ú-na-az (OH/NS), instr. ši-ú-ni-it (OH/NS), ši-ú-na-an, ši-ú-na-aš (OS), ši-i-ú-na-aš (OH/NS); ši-ú-na-li-eš; 3sg.pres.midd. [š]i-ú-ni-a-ta, ši-ú-ni-a-a-ta, ši-e-ú-ni-a-ta, 3sg.pret.midd. ši-ú-ni-a-a-a-ti.

NINDAzu-ri-ú-un is a hapax (KBo 22.186 v 7) of probably foreign origin (cf. zu‑) and therefore of little value here.

The following words show the spelling °i-u-CV or °i-u-uC(-CV):

aniūr (n.) ‘ritual’ is predominantly spelled with U: a-ni-u-ur (KBo 15.19 i 18 (NS), KBo 15.29 obv. 6 (NS), KBo 19.144 i 25 (NS), KBo 20.87 i 7 (NS), KUB 9.15 iii 20 (NS), KUB 12.58 ii 31 (NS), KUB 22.40 iii 29 (NS), KUB 29.4 i 7, 15 (NH), KUB 32.123 ii 33, 47, iii 11 (NS)), a-ni-ur (KUB 46.38 ii 6 (NS), KUB 46.42 ii 12 (NS)), a-ni-u-ra-aš (KUB 35.18 i 9 (MS), KBo 21.1 iv 3 (MH/NS)), a-ni-ur-aš (KBo 12.126+ ii 19 (NS)), a-ni-u-ri (KUB 35.54 iii 45 (MS)), a-ni-u-ra-an-za (KUB 41.9 iv 38 (OH/MS)). Nevertheless, I know of four instances where we find a spelling with Ú, namely a‑ni-ú-úr (KBo 19.92, 4 (OH/NS), KUB 5.6 ii 52, 59 (NS)) and a-ni-ú-ri (KUB 5.6 iii 30 (NS)). Since three of these occur on the same tablet (KUB 5.6), we are actually talking of two instances. Since I am unable to explain these spellings with Ú in comparison with those with U in phonological or chronological terms, I assume that the spellings with Ú are mere mistakes and that the spellings with U are the correct ones. This would mean that a-ni-u-ur represents /ʔniór/ < *h3n‑ié‑ur.

instr. a-aš-ši-u-ni-it (KUB 33.62 ii 20) from āššiauar ‘love’ is probably a scribal error for a-aš-ši-‹a-›u-ni-it as is attested on the same tablet: a-aš-ši-a-u-ni-it (ibid. 20). This form therefore is irrelevant here. mi-u-ri-šẹ[‑?] (KBo 17.17 iv 4 (OS)) and mi-u-ra-aš (KUB 43.53 i 14 (OH/NS)) denote a certain body part, but details are unclear.

ši-iš-ši-u-ri-iš-ke/a- (KUB 31.84 iii 54, 55) ‘to irrigate’ apparently represents /siSiorɨské/á‑/, derived from a noun */siSiór/ < *h1si-h1s-ié-ur.

It is remarkable that the sign U only occurs in words where the consonant ‑r- follows, whereas in all other cases we find the sign Ú. This points to a lowering of */iur/ to /ior/, which has happened in pre-Hittite already (cf. OS. mi-u-ri‑). Note that /iun/ remains unchanged and does not show a lowering comparable to */aun/ > /aon/.



Apart from the one spelling e-ú-uk-zi ‘he drinks’, which is equivalent to e-uk-zi and e-ku-zi and therefore must represent /ʔégwtsi/, a spelling °e-U/Ú-uC° only occurs in ēu- / ēa- ‘rain’ and me()u- / mea- ‘four’.

The of ēu- is spelled é-e-ú-uš, é-e-uš as well as é-uš, which points to Hitt. /Héus/ < /Héʔus/ < *h2éih3-u-s. The is usually spelled é-e-un (attested in OS texts already), but occurs as é-ú-un and é-i-ú-un in some OH/NS texts and as é-e-u-un in an MH/NS text. This seems to point to a phonetic change within Hittite, namely OH /Héʔun/,[76] spelled é-ú-un, develops through /Héun/ into younger /Héon/, spelled é-e-u-un.[77] For this lowering, compare the lowering of /au/ to /ao/ in front of /n/ in § The forms é-e-u-uš (KUB 7.5 i 17 (MH/NS)) and é-e-u[‑u]š (KUB 19.50 iv 27 (NH)), the form é-u-uš (KBo 3.7 ii 22 (OH/NS)) and é-e-ú-uš (KUB 16.37 iv 6 (NH), KUB 28.4 obv. 19 (NS)) as well as é-e-u-uš (KBo 13.245 rev. 7 (NS)) are all, in my view, formally forms that should be interpreted as ē, an incorrect secondary formation instead of correct ēmuš as attested in e.g. é-e-mu-uš (KUB 24.1 iv 15), é-mu-uš (KUB 51.50 obv. 14) and e-e-mu-ú-uš (KBo 43.137 l.col. 7) = /Hémus/ < */Héuus/ < */Héʔuus/.

In the paradigm of meu- / mea, the only relevant form is mi-e-ú-uš (KUB 31.127 i 52), which must be read as me, an incorrect formation instead of expected **memuš, or even better **meamuš < *mea. Word-internally between consonants

If there is a phonological distinction between /o/ and /u/ in interconsonantal position, we would expect that each word that shows a plene spelling with one of the u-signs is consistent in its spelling: either it is spelled with U or it is spelled with Ú. This is not always the case, however: we do find words of which some forms are spelled with U and others with Ú. Let us look at these cases:

apūn ‘that (one)’ ( is consistently spelled a-pu-u-un (more than 150x in my text files),[78] but once we find the spelling a-pu-ú-un (KBo 6.2 ii 32 (OS)). In my view, this last spelling must be a mistake, which is strengthened by the fact that on the same tablet we find the aberrant form u-ú-ni-ik-zi, which is usually spelled u-u-ni-ik-zi (see below).

apūš ‘those (ones)’ ( is almost always spelled a-pu-u-uš (more than 210x in my files),[79] but once we find a spelling a-pu-ú-uš (KUB 14.14 obv. 21 (NH)). In my view, this spelling must be a mistake, just as the form ku-u-ú-uš (ibid. rev. 31) instead of normal ku-u-uš (see below).

arša(r)šur- (n.) ‘flowing, stream’ is attested several times with the sign U: ar-ša-aš-šu-u-ur (KBo 23.9 i 12 (OH/NS)), ar-šar-šu-u-ra (KUB 33.13 ii 14 (OH/NS)), ar-ša-ar-šu-u-ri-i=š-ši-it (KUB 36.55 ii 26 (OH/MS)), ar-šar-šu-u-ru-uš (KUB 33.10, 10 (OH/MS)), case? [a]r-ša-ar-šu-u-ra-aš (KBo 26.135, 6 (OH/NS)). Once we find a spelling with Ú, however: ar-ša-a-aš-šu-ú-ri-i=š-ši-it (KUB 36.55 ii 20 (OH/MS)). It is remarkable that only 6 lines below this form we find ar-ša-ar-šu-u-ri-i=š-ši-it with a plene U. In my view, this indicates that ar-ša-a-aš-šu-ú-ri-i=š-ši-it must be erroneous.

uni(n)k-zi ‘to batter, to crash’ is often spelled with plene U: 3sg.pres.act. u-u-ni-ik-zi (often), 3sg.pres.midd. u-u-ni-ik-ta-ri, 3sg.pret.midd. u-u-ni-ik-ta-at, part. u‑u-ni(-in)-kán-t- (often). Once, we find the spelling u-ú-ni-ik-zi, however, namely in KBo 6.2 i 16 (OS). Since this is the same tablet where we also find the aberrant a-pu-ú-un (instead of normal a-pu-u-un, see above) and since correct u‑u-ni-ik-zi is attested only three lines above (ibid. i 13), we must assume that this is an erroneous form. Moreover, it would be the only form where we find the sequence u-ú- in all of Hittite.[80]

kinūpi, a portable container, is usually spelled without plene ‑u‑, but once we find the spelling ki-nu-ú-pí (KUB 29.2 ii 7) and twice ki-nu-u-pí (KUB 29.1 ii 41, KBo 21.22, 10).[81] Since this word is likely to be of foreign origin, these forms are irrelevant.

kūn ‘this (one)’ ( is consistently spelled ku-u-un (more than 110x in my files). Once we find ku-ú-un, however, namely in KUB 48.125 ii? 4. Although this small fragment does not contain any other aberrancies, I regard this form as an error.

kūš ‘these (ones)’ ( is consistently spelled ku-u-uš (more than 120x in my files).[82] Once we find ku-u-ú-uš, however, in KUB 14.14 rev. 31. Since this form is found on the same tablet as where the aberrant a-pu-ú-uš is attested (instead of normal a-pu-u-uš, see above), I regard it as an error.

NINDAlalla(m)puri(a)‑, NINDAlal(l)am(m)uri(a)‑, a dish made of cereals, shows the following spellings: la-al-la-pu-u-ri-a-aš, la-al-la-am-pu-u-ri-a-aš, la‑al-la-am-pu-ri-iš, la-al-la-am-mu-ri-iš, la-al-la-mu-ú-ri-iš, la-la-mu-ri-a-aš, la-la-mu-ri, [la-l]a-am-mu-ri.[83] The spelling variancy (including the alteration between Cu-u- and Cu-ú‑) and the fact that this word is attested in Kizzuwatnaean rituals only, makes it likely that it is of foreign (Hattic?) origin. This makes the word irrelevant for our purposes here.

lūri- (c.) ‘disgrace’ and its derivatives lūriatar ‘disgrace’ and lūriaŸŸ-i ‘to disgrace’ are predominantly spelled with plene U: lu-u-ri-iš (MH/NS), lu-u-ri-in (OS), lu-u-ri (MS), lu-u-ri (NH), lu-u-ri-e-eš (OS), lu-u-ri-uš (OS); [l]u-u-ri-a-tar (NH)); 2sg.imp.act. lu-u-ri-a-a (NH), impf. lu-u-ri-a-aŸ-Ÿi-eš-ke/a- (NH)).[84] Twice we find a spelling with Ú, however: lu-ú-ri-eš (KUB 13.4 iii 34 (OH/NS)), lu-ú-ri-iš (KUB 13.18 iii 6 (OH/NS)). Since these tablets do not show other remarkable aberrancies, it is not easy to dismiss these examples as errors. Perhaps we are dealing with traces of an original ablaut. See further below.

mūgae-zi ‘to invoke’ and its derivative mūkēššar / mūkēšn- ‘invocation’, when spelled with a plene ‑u‑, are predominantly spelled with the sign U: 1sg.pres.act. mu-u-ga-a-mi (MH/NS), mu-u-ga-mi (MH/NS), 3pl.pres.act. mu-u-ga-a-an[‑zi], mu-u-ga-an-zi; mu-u-ki-iš-na-aš (Bo 6575 obv. 13), mu-u-keš-šarîI.A.[85] Once, we find the spelling mu-ú-ga-it (KBo 3.7 i 13). Since this text contains a number of aberrancies,[86] I regard this spelling as an error as well.

mūri(an)- ‘cluster of fruit’ and its possible derivative NINDAmūriala‑, a bread, are predominantly speled with plene U: instr. mu-u-ri-ni-it (MH?/NS), mu-u-ri-uš (OS), mu-u-ri-a-nu-uš (MH?/NS); mu-u-ri-a-la-aš (OH/NS), mu-u-ri-a-la-an (NS), mu-u-ri-a-lu-š=a (OS).[87] There are two exceptions, however, namely mu-ú-ri-iš (KUB 57.110 ii 8 (NS)), and mu-ú-ri-a-lu-uš (Bo 2689 ii 11 (NS)). On the one hand, since the etymology of mūri(an)- and NINDAmūriala- is unclear, and since mūri(an)- shows a remarkable alternation between an i-stem mūri- and an n-stem mūrian‑, we could claim that these words are possibly of foreign origin and therefore irrelevant here. On the other hand, we could compare the situation to lūri‑, where was also aberrantly spelled lu-ú-ri-iš vs. lu-u-ri- elsewhere, and assume that in mūri(an)‑, too, we are dealing with traces of ablaut.

pūl- (n.) ‘lot’ is attested as follows: pu-u-ul (4x, OH/NS), pu-ú-ul (1x, NH), pu-u-la-aš (OH/NS), pu-la-aš (NH), pu-la-a-aš (NH), abl. pu-la-a[z] (NH), instr. pu-u-li-it (OH/NS), so predominantly with U, but once with Ú.[88] It has been suggested that it is a borrowing through Hurrian (compare Hurr. pulali ‘lot caster’), from Akk. pūru ‘lot’ (cf. e.g. Rieken 1999: 78). As a foreign word, it is irrelevant here.

pūdaaš(ša), puteaš(ša), designation of a festival, is spelled pu-te-a-a-aš-ša, pu‑ú-da-a-aš (NH), pu-u-du-a-aš (NH), pu-da-a-aš and pu-da-a-aš-ša, so both with plene U and Ú.[89] This word occurs almost exclusively as the designation of a festival that is performed in honour of Teššub and îepat, which makes it likely that the word is Hurrian. It is therefore irrelevant here.

punušš-zi ‘to ask’ is predominantly spelled without a plene vowel (pu-nu-uš‑), but sometimes we do find forms in which the first ‑u- is spelled plene: 3pl.pres.act. pu-ú-nu-uš-ša-an-zi (KBo 20.5 iii! 7 (OS)), 3sg.pret.act. pu-u-nu-uš-ta (KUB 36.35 i 8 (MH?/NS)), 1pl.pret.act. pu-u-nu-uš-šu-u-en (AT 454 ii 17, 21, iv 14 (NH)), 2pl.imp.act. [pu‑]ú?-nu-uš-ten (KUB 59.10 vi 2 (OH/NS)). Here we seem to be dealing with a chronological distribution: Ú in OH texts, U in younger texts. This fits the distribution that we established for /aun/ > /aon/ as well (see above).

pūria‑, uria‑, Hurrian offering term, is spelled as follows: gen. pu-u-ri-a-aš (MH/NS), pu-u-ri-a (often, MH/NS), pu-ú-ri-a (KBo 27.191 iii 3), uu-ri-a (MH/MS).[90] Since this word clearly is of foreign origin, it is irrelevant.

(MUN)pūti- (c.) ‘lump of salt’ is spelled pu-u-ti-iš (multiple times) as well as pu-ú-ti-iš (KUB 32.123 ii 18 (NS)).[91] Since this word likely is not native Hittite, it is irrelevant here.

tapūš- (n.) ‘side’ is usually spelled with plene Ú: ta-pu-ú-ša (KBo 4.2 iii 47, KBo 39.164 r.col. 6, KUB 20.99 ii 18, KUB 31.105, 19, KUB 55.45 ii 12, KUB 55.58 obv. 16, IBoT 2.112 obv. 9, etc.). Twice we find a spelling with U, however: ta-pu-u-ša (KUB 1.8 iv 19 (NH)) and abl. ta-pu-u-uš-za (KBo 30.58 iii 11 (OH/NS)). I must admit that I cannot explain these two forms otherwise than as scribal errors, although the texts in which they occur do not show other aberrancies.

tulia- ‘gathering’ is usually spelled without plene vowel: tu-li-a-an (KBo 3.1 ii 34, 51), tu-li-a-aš (KUB 9.34 i 33, iv 12, KUB 6.45 iii 11, KUB 6.46 iii 50, KUB 21.19 iv 10), tu-li-a[‑] (KUB 21.19 iv 25), tu-li-a (KBo 6.3 iii 21, KBo 4.10 obv. 50, KUB 6.45 iii 12, KUB 23.77a obv. 11, KBo 8.35 ii 9, KBo 5.4 rev. 55, KUB 21.1 iv 39, KUB 21.4 iv 9, Bronzetafel iii 79, KUB 21.19 iv 18, 19, KUB 4.1 ii 2, KUB 17.30 iii? 4), tu-li-a-aš (KBo 22.1, 16 (OS)). Occasionally we find a plene spelling, however, namely twice with U (tu-u-li-a (KUB 6.46 iii 51), tu-u-li-a-aš (KUB 33.110, 5)) and twice with Ú (tu-ú-li-a (KUB 21.1 iv 39), tu-ú-lị-[a] (KUB 21.5 iv 45)). This word occurs in CLuwian as well, and is there predominantly spelled tu-ú-li-a- (besides tu-li-a- once).[92] This could mean that the two Hittite spellings tu-ú-li-a- should be regarded as Luwianisms, and the spellings tu-u-li-a- as the ‘normal’ spelling.

zarzur- (n.) ‘concoction’ is attested thus: za-ar-zu-úr (KUB 42.107 iii 13 (OH/NS)), za-ar-zu-u-ur (KUB 31.57 iv 18 (OH/NS)), za-ar-zu-ú-úr (KUB 34.89 obv. 6 (OH?/MS)), [za-a]r-zu-úr (KUB 34.89 obv. 1 (OH?/MS)). Since this word can hardly be native Hittite, it is irrelevant here.

Accordingly, for the words of which we find forms with U as well as with Ú, we have seen that either (1) one of these spellings is a scribal error, (2) the two spellings represent different chronological stages, (3) the different spellings may reflect an original ablaut, or (4) that the word is of foreign origin and therefore irrelevant for our investigation. In all other words, we find a complementary distribution between U and Ú and I therefore conclude that we must assume the occurrence of two different phonemes in interconsonantal position, namely /u/ and /o/.

In the following section I will look more closely at the prehistory of the words under discussion in order to elucidate the origin of the difference between /o/ and /u/. In order to do so, I will treat the words according to the consonants that are adjacent to /o/ and /u/.


Whenever the consonant precedes a plene ‑u‑, this vowel is always spelled with the sign U.[93] This seems to indicate that all instances of */HuC/ have yielded Hitt. /HoC/. As we saw above (§, however, the combination îU-U- occurs so often in MS and NS texts that it has been suggested that we should interpret this combination as a sort of ligature (BÉ) in order to distinguish the sign îU (B) from the closely resembling sign RI (J). It therefore is not always clear how to interpret the combination îU-U‑. To make the problem more transparent, I have taken the liberty to cite the ‘ligature’ îU+U (in which the sign U only seems to have had an orthographic value and perhaps not so much a phonetic value) as îÚ in the following example.[94] For instance, pa-a-ur ‘fire’ must in my view be analysed phonologically as /páHwr/, because of the occasional spelling pa-a-u-a-ar. Once, we find a spelling pa-a-u-u-ur, however. Is this spelling suddenly to be interpreted as /páHor/, or do we have to read the form as pa-a-ú-ur = /páHwr/?

Another problem is that in ablauting verbs, we find e.g. u-e-ek-zi ‘he conjures’ vs. u-u-kán-zi ‘they conjure’. Since I do not reckon with a phonemic distinction between /o/ and /u/ in the sequence C_e (see above), the former should be interpreted /Huégtsi/ < *h2œh-ti whereas the latter is /Hogántsi/ < *h2uœhénti. This means that we seem to be dealing with an ablaut /Hueg- / Hog‑/, which may not be very convenient. Similarly in u-a-ap-p° / u-u-up-p° ‘to harass’, which seems to stand for /Huap- / Hop-/. Perhaps we should conclude that in the full grade forms we are dealing with /o/ as well: /Hoeg-/ and /Hoap‑/, the latter then perhaps expressed in the spelling u-u-a-ap-p°. If so, then we should also interpret e.g u-a-an-t- ‘wind’ as /Hoánt‑/, which then perhaps is expressed in the spelling u-u-a-an-t‑. Since, however, there is no phonemic distinction between /o/ and /u/ after , one could also choose to write /Hu/ everywhere. Yet on the basis of the fact that the Hittites themselves never wrote u-ú- and apparently did not perceive these sequences as [Hu] but as [Ho], I will write /Ho/ in my phonemic analysis, also in the sequences /Hoa/, /Hoe/ and /Hoi/.

Some examples of ūC: u-u-uk-ki-iš-ke/a- ‘to conjure (impf.)’ /Hokɨské/á-/ < *h2uœh-sé/ó-; u-u-uk-ki-iš-ke/a- ‘to butcher (impf.)’ /Hokɨské/á-/ < *h2ug(h)-sé/ó-; u-u-uk-ma-a-uš ‘conjurations (’ /Hogmā́us/ < *h2uœh-mói-; u-u-ul- ‘to smash’ /HoL-/ < *h2ul-n-; u-u-ma-an-t- ‘all’ /Hómant-/; u-u-up‑[pa-an-du] ‘they must harass’ /Hopántu/ < *h2uph1-éntu; DUGu-u-up-pár ‘bowl’ /Hópr/; u-u-ur-da‑a‑in ‘curse (’ /Hortā́in/ < *h2urt-ói-m; u-u-ur-za-ke/a- ‘to curse (impf.)’ /Hortské/á-/ < *h2urt-sé/ó-; u-u-uš-ke/azi ‘to wait’ /Hoské/á-/ < *h2u-sé/ó-; u‑u‑da- ‘readiness’ /Hoda-/ < *h2uh1do-; iš-u-u-na-u- ‘upper arm’ /ɨsHonau-/ < sh2u-nóu‑.


When a follows, we always find U as well. In some cases we are dealing with *Ceuh2-: šu-u-u-za ‘roof (abl.)’ /sóHts/ < *séuh2-ti; in some with *Cóuh2-: u-u-i ‘I saw’ /ʔóHi/ < *h2óu-h2ei, mu-u-u-i ‘I fell’ /móHi/ < *móu-h2ei; in others with *Cuh2-: an-tu-u-u-ša-an ‘human being (’ /ʔndoHsan/ < *h1n-dhuh2-s-om.

Other cases of /CoH/ are: lu-u-a- ‘?’ /loha-/; mu-u-u-ra-i‑, a body part of animals /moHrai-/; šu-u-u-mi-li- ‘firm(?)’ /soHmili-/; (UZU)pa-an-tu-u-a- ‘bladder’ /p(a)ntoha‑/; pár-aš-tu-u-u-a‑, an earthenware cup(?) /prstoHa-/; pu-u-u-ga-ri- ‘substitute’ /póhogari-/; tu-u-u-ši-a-e- ‘to await’ /tohwsiae-/?; MUNUSzi-in-tu-u-i- ‘girl’ /tsintohi-/.


I only know of one case, namely ue-zi / u- ‘to send’, which is consistently spelled u‑i-e- / u-i-a- = /ʔoié- / ʔoi‑/. This verb is a univerbation of the preverb *h2ou and the verbal root *h1ieh1- ‘to send’ (cf. pee-zi / pe- ‘to send (away)’), and shows that *h2ou > */ʔu/ has been lowered to /ʔo‑/ in front of ‑i‑.

Note that the case of ie/a-zi ‘to scream’ is quite different. This verb, which is consistently spelled ú-i-a‑, is a secondarily thematicized form of the verb ai-i / i- ‘to scream’. The spelling of 3sg.pres.act. a-a-i ‘he screams’ shows that there was no initial glottal stop (otherwise we would have expected a spelling **ú-a-a-i), so I would phonologically interpret the spelling ú-i-a- as /uiá‑/, phonetically realized as [wijá‑].



Here we must distinguish between different ablaut grades: *CuK > /CuK/, cf. *h1éœH >> *h1úœ > Hitt. ú-uk ‘I’ /ʔúg/, *iugom > i-ú-kán /iugan/ ‘yoke’ and *dolugh- > ta-lu-ú-ga ‘long’; *CeuK > /CūK/, cf. *méug-r > mu-ú-kar ‘rattle’ /mÜgr/, *iéug > i-ú-uk /iÜg/ ‘yoke’ and *h2téug-om > a-tu-ú-ga-an /HdÜgan/ ‘terrible’; but *CouK > /CoK/, cf. *mougó-e/o- > mu-u-ga-e- /mogae/ ‘to invoke’ /mogā́e-/.


The situation around C_l is quite complicated, especially because the etymology of many words containg ‑Cul- is unclear. A sequence *Ceul is clear in the words i-mi-ú-ul ‘horse feed’ < *im-ié-ul and iš-i-ú-ul ‘binding’ < *sh2-ié-ul, which show that *Ceul > /Cul/. The words aš-šu-ú-ul ‘favour’, tak-šu-ú-ul ‘agreement’ and uštūl- / aštūl- ‘sin’ (cf. a-aš-du-ú-li) are usually regarded as showing the accented suffix *‑úl‑, and would show that *Cul yields /Cul/ as well. This would also fit the word ga-az-zar-nu-ú-ul, a certain cloth, although its etymology is less clear. It has been claimed that pittūla- ‘loop, knot’ is a thematization of the suffix *‑ul‑, but this word is consistently spelled pit-tu-u-la‑. We could assume that we are dealing with a lowering to /o/ here due to the back-vowel that follows l‑, which also fits the words ka-lu-u-lu-pa- ‘finger’, mu-u-la-tar, an evil quality, NINDAmu-u-la-ti‑, a bread, and pár-šu-u-la-a-an-t- ‘crumbling’.[95] It would imply that before a front vowel, we would expect /u/. This is certainly the case for lu-ú-li- ‘pond’, lu-ú-li-a-aš-a- ‘marshland’, mi-i-lu-ú-li ‘skin(?)’ and mu-ú-li-li, a plant. Note that tu-u-li-a- ‘gathering’ does not fit into this picture: it shows /o/ inspite of the following front vowel. Since the etymology of this word is not fully clear, it is difficult to judge this form. Perhaps we are dealing with *tuHl-i-o‑, in which *CuHl yields /Col/. Also šu-u-ul-le-e-et ‘he became arrogant’ shows /o/ while a front vowel follows. Perhaps we must conclude that here the geminate ‑ll- < *‑lH- was the crucial factor and that it caused lowering as well. This does not work for šu-ú-ul-lu-uš, of šulla- ‘hostage’, however, but here we might be dealing with *seul°. The reflex of *Coul may be /Caul/ if a-ú-li‑, a tube-shaped organ in the neck, indeed reflects *h2ou-li- (see s.v.). Note that the /o/ in GIŠu-u-la-li, GIŠu-u-lu-ga-an-ni- and GIŠšar‑u-u-li- is determined by the preceding .


On the basis of a-a-nu-ú-mi ‘I make turn’ /uəhnÜmi/ < *‑néu-mi, we must conclude that *Ceum > Hitt. /Cūm/. It must be noted that ú-me-e-ni ‘we see’ and a‑ú-me-en ‘we saw’ are non-probative since the ‑m- in these forms is recent: the forms go back to *Hu-éni and *Hóu-en. Inf.I pát-tu-u-ma-an-zi ‘to dig’ (KUB 55.45 ii 4) < *bhdhh2-uén-ti shows that *CHuV > Hitt. /ComV/,[96] which means that e.g. tu-me-e-ni ‘we take’ < *dh3uéni stands for /toméni/, tar-nu-me-ni ‘we let go’ < *trk-n-h1/3-uéni stands for /trnoméni/, etc. This probably also goes for the appurtenance suffix ‑umen- / ‑umn‑, which is spelled with U in i-iš-tu-u-ma-aš, é-eš-tu-u-um-ni ‘person pertaining to the ištā’, URUKa-a-ta-pu‑u-me-né-eš ‘persons from Kātapa’, URULu-ú-i-u-ma-na-aš ‘person from Lūi̯a’, URUŠa-lam-pu-u-me-né-eš ‘persons from Šalampa’, URUZa-al-pu-u-ma-aš ‘person from Zalpa’, mŠu-up-pí-lu-li-u-ma ‘man from the pure well’ and ta-me-u-ma- ‘being from somewhere else’. The etymology of nu-u-ma-an (negation of man) is not fully clear. The words Éka-ru-ú-um-mi ‘sanctuary’, Étu-u-ma-an-ti-a-at‑t-, a building, NINDAtu-u-ma-ti‑, a bread, and al-al-tu-u-ma-ri ‘corner’ are likely of foreign origin. The U in u-u-ma-an-t- is determined by the preceding .


As we saw above, *Coun yields OH /Caun/, which develops into /Caon/ from the MH period onwards. A similar chronological distribution may underly the difference between OS pu-ú-nu-uš-š° and younger pu-u-nu-uš-š° ‘to ask’ (although the etymology of this verb is not clear yet). This would also explain the spelling ki‑nu‑u‑n=a (KUB 14.17 ii 14 (NH)) /kinon/ < *i-num. Also the NH attestation e-ep-pu-u-un ‘I grabbed’ (KBo 3.6 ii 7) shows that in NH times the 1sg.pret.-act. ending ‑un in fact was /‑on/ < older /‑un/, cf. OH pa-a-ú-un /pā́un/ > NH pa-a-u-un /pā́on/ (§ It does not apply to all positions, however: the fact that unu-zi ‘to decorate’ < *h3u-neu- is spelled ú-nu- = /ʔunu-/ throughout Hittite shows that in initial position this lowering did not take place. In the case of ku-ú-na-aš ( ‘dog’ (KBo 7.48, 12 (MS?)) we are in my view dealing with a restored /kunas/ that replaced expected **/konas/ < *unos on the basis of the full grade stem /kuan-/ ( ku-a-aš /kuā́s/ < *uÖns, ku-a-na-an /kuā́nan/ < *uón-om).

The lowering of /u/ to /o/ seems to have taken place in front of geminate ‑nn- as well, as is apparent in ūnna-i / ūnni- ‘to send (here)’ that is consistently spelled u‑un‑n° = /ʔoN°/ < *h2ou + *n(o)iH-.[97] Since this word is attested in MS and NS texts only, we do not know whether the lowering has taken place in OH times as well. The plene spellings ku-u-un-na- ‘right’ = /koNa-/ are attested in NS texts only and do not give information about the OH pronunciation of this word. Although emended, 2pl.imp.act. šu-u-‹un-›ni-iš-tén ‘you must fill’ (KUB 13.3 ii 27 (OH/NS)) also points to /soN°/ < *su-n-H‑. The hapax spelling mu-ú-un-na-a-it ‘he hid’ (KUB 17.5 i 4 (OH/NS))[98] may show a reflex of an OH form that still shows /muNā́it/ (< *mu-n-H-?), instead of younger /moNā́it/ (although we do not have any spelling **mu-u-un-n° of this verb).

The verb šūnie/a-zi ‘to dip’ is consistently spelled šu-ú-ni- throughout Hittite. Because its etymology is rather unclear, we cannot determine its preform. Since *Coun > OH /Caun/ > NH /Caon/ and *Cun > NH /Con/, the only reasonable possibility is *Ceun. This may go for a-ru-ú-ni ‘sea’ ( (KUB 36.41 i 13 (MS)) as well, which therefore perhaps should be reconstructed as *h3reuni.[99]

The forms ku-u-un ‘this (one)’ ( = /kón/, a-pu-u-un ‘that (one)’ ( = /ʔabón/ and u-ni ‘him (there)’ ( = /ʔóni/ are special cases. They are spelled with U from the oldest texts onwards, and therefore cannot be derived from older **/kún/, **/ʔabún/ and **/ʔúni/ through a MH lowering in front of n. In my view, these forms show that the reflex of *Cóm was /Cón/ in the oldest stages of Hittite already.[100] (MUŠEN)pár-tu-u-ni-uš (StBoT 25.3 iv 37 (OS), StBoT 25.4 iv 33 (OS), StBoT 25.7 iv 2 (OS)), a certain bird, shows a remarkable U in front of ‑n- in OS texts. The similarity to (UZU)partāar / partaun- ‘wing, feather’ is striking, but since the prehistory of this latter word is not fully known, the interpretation of partūniuš remains unclear.[101]

The interpretation of ū(n)-zi ‘?’ is not entirely clear. It is spelled u-un-°, which is attested in an OS text already: u-un-a-an-zi (KUB 32.94 i 3 (OS)), as well as u-u°. The /o/, which might be unexpected in front of ‑n- in OH times, is in my view due to the following - in the allophonic stem ū-.[102]

The /o/ in [a]r-u-u-un ‘I arrived’ /ʔarHon/, da-a-u-u-un ‘I took’ /tā́Hon/, al‑ze-e-u-u-un ‘I screamed’ /HltséHon/, tar-na-a-u-u-un ‘I let go’ /trnáHon/ and iš-u-u-na-u- ‘upper arm’ /ɨsHonau-/ is automatic due to the preceding .



On the basis of u-up-zi /ʔóptsi/ ‘(the sun) comes up’ < *h1eup-ti, we can conclude that *Ceup- > Hitt. /Cop/. The adjective šuppi- / šuppa, which is spelled with U in the name mŠu-u-up-pí-lu-li-u-ma (KUB 19.10 iv 2) can hardly reflect *séup-i- or *sóup-i‑, since in these forms we would have expected lenition to **/b/. This means that šu-u-up-pí- /sopi-/ reflects *sup-i‑, which shows that *Cup yields /Cop/ as well. The verb uppa-i / uppi- ‘to send (here)’, which represents /ʔup°/ (see §, reflects *h2ou + *h1p-oi‑. Since this verb is a quite recent univerbation of the preverb *h2ou, which in isolation yielded /ʔu/, and the verb pai-i / pi‑, this example is non-probative for the outcome of *Coup.

The words kinūpi, a container, which is spelled ki-nu-ú-pí as well as ki-nu-u-pí, lu‑u-pa-an-ni- ‘royal cap’ (also luanni‑), dam-pu-u-pí- ‘barbaric’ and ša-ru-ú-pa ‘?’ are all probably of a foreign origin, and do not shed any light on this matter.


First we should keep in mind that *Cour yields Hitt. /Caur/, e.g. a-ú-ri- ‘lookout’ < *Hou-ri‑. The sequence *Cur seems to yield Hitt. /Cor/, as is found in e.g. an-tu-u-ri-a- ‘interior’ /ʔntoria-/ < *h1n-dhur-o-(?), ar-ša-ar-šu-u-r° ‘stream’ /ʔrsʔarsor-/ < *h1ers-ur‑, pár-šu-u-ur ‘cooked dish’ /p(a)rSor/ < *bh(e)rs-ur, pu-u-ru-ut ‘mud’ /porut/ < *bhur-u-t(?), tu-u-ri-a- ‘to harness’ /torie/a-/ < *dhuh1r-ie/o‑, u-ur-ki- ‘trace’ /ʔorgi-/ < *h1/3urg-i‑, a-ak-šu-u-ur, a vessel /uaksor/ < *ueKs-ur(?), and ú-i-šu-u-ri-a- ‘to press together’ /uisorie/a-/ < *uis-ur-ie/o-(?). This implies, however, that the one attestation ši-iš-šu-ú-ra-aš ‘irrigation (’ (KBo 6.26 iii 5 (OH/NS)) < *h1si-h1s-ur- is a mistake, cf. correct ši-iš-šu-u-r° (KUB 31.100 rev.? 17 (MH/MS)).[103] With *Cour yielding /Caur/ and *Cur > /Cor/, the only way to explain lu-ú-ri- ‘disgrace’ /lūri-/ is by reconstructing *leh1u-ri-.[104] The forms within the paradigm of this word that are spelled lu-u-ri- may then reflect *lh1u-ri- > luh1ri- > /lori-/. Note that *eur yielded /or/ in a-ni-u-ur ‘ritual’ /ʔniór/ < *h3n-ié-ur and ši-iš-ši-u-r° ‘irrigation’ /siSior/ < h1si-h1s-ié-ur (see above), but here the preceding i- may have been crucial.


First we should keep in mind that *Cous > Hitt. /Caus/, e.g. a-uš-te-en ‘you must see’ < *Hou-sten.[105] A sequence *Ceus yields Hitt. /Cūs/, as follows from e.g. ku‑ú‑ša- ‘daughter-in-law, bride’ /kÜsa-/ < *œéus-o- and ka-ru-ú-uš-ši-a- ‘to be silent’ /krūSie/a-/ < *greus-e/o‑. On the basis of the spelling pa-an-ku-ú-š=a (KUB 35.136 iv 9 (NS)) for of panku- ‘multitude’, which represents /pngus/ < *bhnœh-u-s, we can conclude that *Cus in principle yields Hitt. /Cus/.

Our findings that *Cous > /Caus/, *Ceus > /Cūs/ and *Cus > /Cus/, seem to imply that in Hittite the spelling Cu-u-š° or °Cu-u-uš° cannot exist. This is not entirely the case: pu-u-uš-zi ‘to be eclipsed’ /pós-/ may reflect *ph2u-s‑, in which the *h2 may have caused lowering; a-a-ru-u-uš-i ‘incense vessel’ is likely of a foreign origin; šu-u-uš ‘full (’ /sós/ is a contraction of /sóus/ < /sóʔus/ < *sóuH-u-s, see at C_ʔ.

The ending ‑uš is a special case. It is predominantly spelled °Cu-uš, but occasionally we find forms with plene spelling. It is spelled with plene Ú in: al‑pu‑ú-uš (KUB 28.5 rev. 7 (NS)), a-ú-li-ú-uš (KBo 25.178 i 2 (OH/NS), KUB 24.3 ii 11 (MH/NS)), a-ú-li-ú-š=a (KUB 17.21 ii 18 (MH/MS)), NINDAar-ša-ú-uš (KBo 17.4 ii 17 (fr.) (OS), KUB 7.8+ ii 11 (NS)), e-e-mu-ú-uš (KBo 43.137, 7 (NS)), kap-pí-ú-uš (KBo 34.47 ii 8 (MH/MS)), ku-i-ú-uš (HKM 23 obv. 9 (MH/MS), KBo 18.57a + 57 obv. 2, rev. 42 (MH/MS)), ma-ši-ú-u[š]? (KBo 9.109 rev. 4 (OH/NS)), pu-u-ri-ú-uš (KBo 19.163 i 23, iv 4 (OH/NS)), ta-lu-ga-ú-uš (KBo 17.22 iii 6 (OS)). It is spelled with U in [al‑]lu-ú-a-u-uš (KBo 3.8 iii 4 (OH/NS)), al-lu-a-u-uš (KBo 26.135, 2 (OH/NS)), [i-da-a]‑la-mu-u-š=a (KBo 15.10 iii 54 (OH?/MS)), [i?-da?]‑a-la-mu-u-uš! (KUB 8.67 iv 14 (MH/NS)), pár-ga-u-uš (KBo 3.8 iii 22 (OH/NS)). Although the MS attestation [i-da-a]‑la-mu-u-š=a is awkward, it seems that we are dealing with a development of OH /‑us/ to NH /‑os/. It is not fully clear to me whether we must assume every OH /us/ (also when reflecting *Cus) to develop into NH /os/, or that here we are dealing with a special development of *°Coms and *°Cms, yielding first OH /°Cųs/, which subsequently develops into NH /°Cos/.[106]

The forms ku-u-uš ‘these (ones)’ ( and a-pu-u-uš ‘those (ones)’ ( must be treated separately as they show plene spelling with U throughout Hittite, which indicates /kós/ and /ʔabós/. I regard these as the regular reflexes of *Cóms (just as /Cón/ is the regular reflex of *Cóm, see above).


It must be borne in mind that *Cout > Hitt. /Caut/, e.g. a-ut-ta ‘you (sg.) saw’ < *Hóu-th2e.[107] On the basis of u-e-nu-ú-ut (KBo 3.28 ii 19) ‘he made run’ (or u‑e‹‑eš›-nu-ú-ut ‘he rescued’ ?) = /HoinÜt/ < *‑néu-t, we must conclude that *Ceut > Hitt. /Cūt/.[108] This is confirmed by ku-ú-uz-za ‘wall’ /kÜts/ < *œhéu-t-s. For *Cut I have found no conclusive evidence. The forms u-u-da- ‘readiness’ /Hóda-/ and u-u-da-a-ak ‘immediately’ /Hodā́k/ probably reflect *h2uh1-do- in which the initial is the determining factor for the reflex /o/.

The hapax spelling ku-u-ut-ru-a-a-iz-zi ‘he provides testimony’ < *kwtru- may show that the labial element of the labiovelar /kw/ was perceived more as /o/ than as /u/. Nevertheless, I will write the labiovelars as /kw/ and /gw/ in phonemic transcription (so /kwtru-/ here).


The only case is ku-ú-uz-za ‘wall’, for which see under C_t.


The stem of the adjective šūu- / šūa- ‘full’ is consistently spelled with U ( šu-u-uš, šu-u-un, šu-u-ú, šu-u, šu-u-a-mu-uš) which points to /so-/. The remarkable spelling of šu-u-ú in my view represents /sóu/, which must reflect /sóʔu/ < *sé/óuh1/3-u. In younger times, this form is spelled šu-u = /só/, the result of the contraction of šu-u-ú = /sóu/. This probably goes for šu-u-uš = /sós/ < /sóus/ < /sóʔus/ < *sé/óuH-u-s and šu-u-un = /són/ < /sóun/ < /sóʔun/ < *sé/óuH-u-m as well, whereas šu-u-a-mu-uš presents /sóamos/ < /sóʔamus/ < pre-Hitt. */sóʔauus/ < virtual *sé/óuh1/3‑eu‑ms.

The spellings with Ú in šu-ú-il, šu-ú-i-il ‘thread’ and mu-ú-i-il ‘spade(?)’ seem to represent /sÜil/ and /mÜil/ respectively. Since these words are derived from the roots *seuh1- ‘to sow’ and *meuh1- ‘to move’ (although the latter is not fully certain), they originally must have contained /ʔ/: */sÜʔil/ and */mÜʔil/. These then must reflect *sé/óuh1-el and *mé/óuh1-el here.

Since we are dealing with two reflexes, /o/ and /u/, and two possible reconstructions, *eu and *ou, it is not possible to decide which one reflects which. For the sake of parallelism with C_k, where *Ceuk > /Cūk/ and *Couk > /Cok/, I assume that šūu- /sóu-/ reflects *sóuh1-u- and that šūil- /sÜil-/ and mūil- /mÜil-/ reflect *séuh1-el- and *méuh1-el- respectively.

Overview of interconsonantal reflexes

On the basis of the treatments above, we arrive at the following overview of the interconsonantal refelxes of PIE *ou, *eu and *u. Note that when - is the preceding consonant, the reflexes of these are always /o/.

                                *ou              *eu              *u                                other

    C_                     /o/                /o/                /o/

    C_i                      /o/                --                 --

    C_k                      /o/                /ū/                /u/

    C_l                      /au/              /ū/                /u/[109]

    C_m                    --                 /ū/                --

    C_n         OH      /au/              /ū/                /u/                /o/ < *CuHn ?         /o/ < *Cóm

                                  ↓                ↓                 ↓                 ↓                             ↓

                    NH      /ao/[110]          /ū/                /o/[111]            /o/121                        /o/

    C_p                     /o/                /o/                /o/

    C_r                      /au/              /o/[112]            /o/


    C_s          OH      /au/              /ū/                /u/                /u/[113] < *°C(o)ms    /o/ < *Cóms

                                                                                             ↓                             ↓

                    NH                                                                  /o/[114]                        /o/


    C_t                      /au/              /ū/                --

    C_ʔ                     /o/                /ū/                -- Word-finally after consonants

There are only a few relevant forms here, namely a-aš-šu-u ‘goods’, šu-u ‘full’ (from older šu-u-ú, see above) and the adverb ka-ru-ú ‘early, formerly’. Since these forms are consistent in their spelling,[115] they point to a phonological difference between °Cu-u and °Cu-ú. We must keep in mind that a third spelling of course is °Cu without a plene vowel (e.g. a-aš-šu ‘good’). I therefore assume that °Cu-u stands for /°Co/, °Cu-ú stands for /°Cū/ and °Cu stands for /°Cu/. So, a-aš-šu = /ʔáSu/ < *Cu, a-aš-šu-u = /ʔáSo/ < *Cuh2, šu-u = /só/ (a contraction of /sóu/) and ka-ru-ú = /krÜ/ < *Céu.[116] Word-finally after vowels

a_# : The sequence °a-u only occurs in the words (GIŠ)za-a-u ‘?’, ši-i-iš-a-u (KBo 3.2 obv. 26) ‘sweat’, GIŠma-ra-a-u (KBo 20.86, 9), a wooden object used as seat, and ga-ra-a-u (KBo 40.176 obv. 11) ‘?’. Although the last three words occur in this form only once,[117] the word (GIŠ)za-a-u is attested thus several times, and never spelled **za-a-ú.

The sequence °a-ú occurs often. It is found in the of diphthong-stems (e.g. ar-na-a-ú, ar-ga-na-ú, GIŠta-na-a-ú), in 2sg.imp.act. a-ú ‘look!’ and in 3sg.imp.act. of dāi/tianzi‑, mēma‑/mēmi- and tarn(a)-class verbs and of dā-i / d- (al-za-a-ú, a-an-na-ú, a-at-ra-a-ú, u-u-a-a-ú, [u]‑et-ti-a-an-na-ú, e-eš-ša-ú, iš-u-a-a-ú, iš-kal-la-ú, la-a-ú, ma-a-ú, me-e-ma-ú, me-em-ma-ú, na-a-ú, na‑an-na-ú, pa-a-ú, pé-e-da-ú, da-a-ú (of dā-i / d- ‘to take’), da-a-ú (of dai‑i / ti- ‘to place’), da-la-a-ú, tar-na-ú, du-a-ar-na-a-ú, u-un-na-ú, up-pa-ú, ú-da-ú, ú-i-ta-ú, zi-in-na-a-ú). All these forms, too, are consistent in their spelling; they never show **°a-u.

It is clear that the spellings °a-u and °a-ú are used complementarily, and therefore it is likely that they denote different sounds. I consequently propose to interpret °a-u as /°ao/ and °a-ú as /°au/. Note that the words that are spelled °a-u do not have a good IE etymology,[118] which shows that the diphthong /ao/ in word-final position is not inherited, but probably is of foreign origin.


e_# : Neither the spelling °e-u nor °e-ú is attested in the Hittite texts.


i_# : The spelling °i-u occurs only in Éa-le-en-ti-u ‘palace’, which is not coincidental if we compare the fact that the spelling °i-u-i° only occurs in this word as well. Apparently, it is pronounced /Halentio/. The contrasting spelling °i-ú is found in mi-ú and mi-i-ú ‘soft’ < *mih1-u, which must represent /mÏu/. Conclusions regarding U and Ú

From the treatment above it is clear that the signs U and Ú, which are both traditionally interpreted as u- only, in fact can be used to represent three different phonemes, namely /u/, /ū/ and /o/. Note that I do not distinguish a fourth phoneme, /ō/, for several reasons. First, the fact that the spelling of /o/ automatically requires the use of a plene vowel, namely the sign U, makes it graphically impossible to distinguish between a short /o/ and a long /ō/. Secondly, it is likely that /o/ behaves symmetrically to /e/, which does not show a phonemic distinction in length: when accented, /e/ is phonetically long in open syllables and monosyllabic words, but this lengthening is automatic and therefore subphonemic. I assume a similar behaviour of /o/. Epenthetic vowels

In Hittite we can distinguish three epenthetic vowels that emerge in specific consonant clusters.

(1) In clusters of the shape *CRC, i.e. containing syllabic resonants, an epenthetic vowel spelled -a- can emerge that cannot be identical to /a/ and phonetically may have been [6] or [ə] (cf. § 1.3.7).

(2) In some clusters involving -s- and stops or laryngeals an epenthetic vowel spelled -e- or -i- emerges: *-TsK- > Hitt. -ze/ik(k)-; *-PsK- > Hitt. -p(p)e/išk(k)‑; *‑KsC- > Hitt. ‑k(k)e/išC‑; *‑Vh2s > Hitt. ‑VŸŸe/iš; *-VKs > Hitt. ‑Vk(k)e/iš; *ClHsV > Hitt. Cale/iššV; *CmHsV > Hitt. Cane/iššV; *CnHsV > Hitt. Cane/iššV; *CrHsC > Hitt. Care/išC; *ClHsC > Hitt. Cale/išC; *CmHsC > Hitt. Came/išC; *CnHsC > Hitt. Cašše/iC; *VrHsC > Hitt. Vre/išC; *VlHsC > Hitt. Vlle/išC; *VmHsC > Hitt. Vmme/išC (cf. § and § We also find this vowel in secondary initial clusters *T- (in which T = any stop), e.g. e/itēn- (obl.-stem of ātar ‘water’) < *d-én- << *ud-én- or e/ikk- (weak stem of ekk-zi ‘to wish’) < *¾§- << *u- (see their respective lemmata). This vowel cannot be identical with /i/ or /e/ because these are consistently spelled -i- and -e- respectively. The vowel e/i therefore may phonetically have been [ɨ] or [ɘ].

(3) Before initial clusters of the shape *sT- (in which T = any stop and /H/) a prothetic vowel spelled i- emerges: e.g. *stu- > Hitt. ištu-, *sh2oi- > Hitt. ai-, etc. This i- cannot be identical to /i/ because it does not partake in the NH lowering of OH /i/ to /e/ before /s, n, m/ and clusters containing /H/ (cf. § It can neither be identical to the epenthetic vowel e/i, because it is never spelled e-. Phonetically we may think of e.g. [I].

Because these three vowels occur in specific environments that are complementarily distributed, we could in prinicple regard them all as allophones of a single phonemical epenthetic vowel, which we could write as /ə/.

Note that there potentially is one environment in which -a- = [ə] and -e/i- = [ɨ] have to be phonologically distinguished, however, namely in /K_sC/. The cluster *KsC regularly yields Hitt. [kɨsC], spelled -ke/išC- (e.g. h3rg-sé/ó- > Hitt. ar-ki‑iš‑ke/a-, ar-ke-eš-ke/a-, *téks-ti > ták-ki-iš-zi, ták-ke-eš-zi, etc.), whereas a cluster *KnsC would yield pre-Hitt. *[kənsC], which with the regular loss of *n before *s would further develop in Hitt. [kəsC], spelled -kašC-. If we would interpret -a- = [ə] and -e/i- = [ɨ] as allophones of a single phoneme /ə/, it would in this environment become impossible to explain on the basis of synchronic reasoning only why the phonological form /kəsC/ is spelled in one form as ‑ke/išC- and in the other as ‑kašC‑. It should be noted, however, that thus far the development *KnsC > -kašC- is only attested in the verb *gwhnsé/ó- > Hitt. [kwəsk:e/a-], spelled ku-a-aš-ke/a- ‘to kill (impf.)’, which has an initial labiovelar. Because in the cluster *KsC labiovelars behave differently, yielding not **-kue/išC- but -kušC- (cf. *h1gwhsé/ó- > Hitt. ak-ku-uš-ke/a-), there is thus far no minimal pair attested where -a- = [ə] and -e/i- = [ɨ] have to be distinguished. Nevertheless, I do not think it improbable that such a minimal pair may have existed and one day will surface in the texts.[119]

In this book I therefore have rendered the vowel -a- = [ə] as /ə/ and the vowel ‑e/i- = [ɨ] as /ɨ/, without specifically claiming that they must be regarded as separate phonemes: the reader should bear in mind that on the basis of the Hittite material that is known so far it is fully justified to regard these vowels as allophones of a single phoneme /ə/. For sake of convenience, I have rendered prothetic i- = [I] as /ɨ/ as well.

So, the phonological vowel chart can be given as follows:

                                            /ī/  /i/                                 /u/  /ū/


                                                    /e/                           /o/

                                                                 /ā/  /a/

It should be noted, however, that in the case a form would surface in which a cluster -kašC- = [kəsC] < *KnsC is attested, which then forms a minimal pair with -ke/išC- = [kɨsC] < *KsC, this chart should be adapted to:

                                            /ī/  /i/              /ɨ/               /u/  /ū/

                                                    /e/                           /o/


                                                                 /ā/  /a/



1.3.10 Overview of the Hittite phoneme inventory

After having treated all evidence available from the Hittite orthography, discussing spelling conventions and complementary distributions, I conclude that the Hittite phoneme inventory was as follows:

                                        /p/               /t/                /k/               /kw/                     (fortis)


                                        /b/               /d/               /g/               /gw/                     (lenis)

    affricate                                                /ts/

    glottal stop                                            /ʔ/

                                                /H/              /Hw/            /S/                                   (fortis)


                                                /h/               /hw/             /s/                                    (lenis)

                                        /R/              /L/               /N/              /M/                     (fortis)


                                        /r/                /l/                /n/               /m/                      (lenis)




                                            /ī/  /i/                                 /u/  /ū/


    vowels                                    /e/                           /o/

                                                                 /ā/  /a/

1.4 Changes from PIE to Hittite

In this chapter I will treat the phonological developments that took place from Proto-Indo-European to Hittite. First I will treat some basic phonological phenomena that are important for Hittite historical phonology. Then I will systematically treat the PIE phonemes and discuss their reflexes in Hittite in different phonological surroundings. Note that I will only refer to the intermediate Proto-Anatolian stage when necessary. Sometimes I will use more vague terms like pre-Hittite (i.e. any stage between PIE and attested Hittite), post-PAnat. (i.e. the stage between PAnat. and attested Hittite) or pre-PAnat. (i.e. the stage between PIE and PAnat.).

1.4.1 Lenition


Lenition is the phenomenon that an original fortis consonant becomes lenis. We can distinguish two situations in which lenition regularly takes place.[120]

(1) Intervocalic fortis consonants are lenited after an accented long vowel.[121] Note that this includes the reflexes of the monophthongizations of *ei, *oi, *eu and *ou[122] as well as the reflex of *ó, which yields Hitt. /ā́/ through PAnat. /Ö/.[123] Examples: *h1éih2ou > Hitt. /ʔéhu/, e-u ‘come!’; *méih2ur > Hitt. /méhur/, me-e-ur ‘period, time’; *sókwo- > Hitt. /sā́gwa‑/, ša-a-ku-a- ‘eye’; *h2ómsei > Hitt. /Hā́si/, a-a-ši ‘she gives birth to’;[124] *nóh2ei > Hitt. /nā́hi/, na-a-i ‘he fears’.[125]

It should be noted that in many occasions the fortis consonant was restored, especially when we are dealing with verbal endings (e.g. te-e-e-i /téHi/ ‘I take’ < *dhh1óih2ei should regularly have been **/téhi/, **te-e-i, etc.).

(2) Intervocalic fortis consonants are lenited between two unaccented vowels.[126] There are only a few examples from Hittite because in many occasions the fortis consonant has been analogically restored. Real examples include: *CÕC-h2eh2e > Hitt. /CÕCHaha/, °C-a-a (1sg.pres.midd. ending);[127] *sépitos > OH /sépidas/, še‑ep-pí-da-aš, a kind of grain ( >> OH /sépitas/, še-ep-pí-it-ta-aš, with restored /t/.

It should be borne in mind that lenition only affects intervocalic consonants, i.e. not consonants that are part of a cluster (compare e.g. the treatment of ekk-zi).

1.4.2 Fortition


Fortition is the phenomenon that an original lenis consonant becomes fortis. Usually, this is due to contact with an adjacent consonant. For instance, it seems to be a general rule that lenis consonants are fortited before the cluster ‑sk-: ak-ku-uš‑ke/a-zi = /ʔkwské/á‑/, the ‑ške/a-imperfective of eku-zi / aku- = /ʔegw- / ʔgw‑/ ‘to drink’; la-ak-ki-iš-ke/a- = /ləkɨské/á‑/, the ‑ške/a-imperfective of lāk-i / lak- = /lāg- / ləg‑/ ‘to knock down’;[128] etc. Sometimes, fortition can be interpreted as assimilation, e.g. *Vgh2V > Hitt. /VkV/. For more examples, see below at the treatment of the separate phonemes.



1.4.3 Stops     *p

The normal reflex of PIE *p is Hitt. /p/: *pédom > Hitt. /pédan/, pé-e-da-an ‘place’; *prh1-ói-ei > Hitt. /prʔā́i/, pa-ra-a-i ‘he blows’; *h1ép-ēr > Hitt. /ʔéper/, e‑ep-pé-er ‘they seized’; *sup-óri > Hitt. /supári/, šu-up-pa-ri ‘he sleeps’; *h1épti > Hitt. /ʔéptsi/, e-ep-zi ‘he seizes’, etc.. Like all other consonants, *p could fell victim to lenition and then yields /b/: *stóp-ei > Hitt. /ɨstā́bi/, iš-ta-a-pí ‘he plugs up’, cf. *stpénti > /ɨstpántsi/, iš-tap-pa-an-zi ‘they plug up’.     *b

Since already in PIE *b was a rare phoneme, only one example can be found in Hittite, namely *ghróbh1-ei > Hitt. /krā́bi/, ka-ra-a-pí ‘he devours’, which shows that PIE *b yields Hitt. /b/.     *bh

The normal reflex of *bh is Hitt. /b/: *nébhes > Hitt. /nébis/, ne-e-pí-iš ‘heaven’; *dhébh-u > Hitt. /tébu/, te-e-pu ‘little’, *h3érbh-to > Hitt. /Hárbta/, ar-ap-ta ‘he changes alliance’. In initial position, all labial stops have merged in /p/: *bhérh2/3-ti > Hitt. /párHtsi/, pár-a-zi, pár-a-zi ‘he chases’. Fortition of *bh to /p/ seems to have taken place in *h3rbh-sé/ó- > Hitt. /Hrpɨské/á‑/, ar-ap-pí[‑iš-ke/a‑], impf. of arp‑tta ‘to change alliance’ and in *kmbh-i- > Hitt. /kapi‑/, kap-pí- ‘small’.     *t

The normal reflex of *t is Hitt. /t/: *terh2-u- > Hitt. /tarHw‑/, tar-u‑, ta-ru-u- ‘to conquer’; *tué-om > Hitt. /tuékan/, tu-ek-ka-an ‘body’; *h2ét-o > Hitt. /Háta/, a‑at-ta ‘he pierces’; *melit- > Hitt. /milit‑/, mi-li-it-t° ‘honey’; *h1p-ént-om > Hitt. /ʔpántan/, ap-pa-an-ta-an ‘seizing’. In front of *i, *t is assibilated to /ts/: *tíH-ge > Hitt. /tsÏg/, zi-i-ik ‘you’; *ti-n-h1-énti > Hitt. /tsiNántsi/, zi-in-na-an-zi ‘they finish’; *h2t-ié/ó- > Hitt. /Htsié/á‑/, a-az-zi-e°, a-az-zi-a- ‘to pierce’;[129] *h2entetio- > Hitt. /Hantetsia-/, a-an-te-ez-zi-a- ‘first’; *h1és-ti > OHitt. /ʔésts/, e-eš-za >> Hitt. /ʔéstsi/, e-eš-zi;[130] see § for a more detailed treatment. Assibilation also takes place in word-initial position in front of *l: *tlh2-Öi > Hitt. /tslHā́i/, za-al-a‑a‑i, a vessel, lit. ‘carrier’. Lenition of *t may be found in some forms of the paradigm of šeppitt‑, a grain, e.g. še-ep-pí-da-aš /sépidas/, if this reflects *sépitos. Also in *h2tug- > Hitt. /Hdug‑/, a-tu-uk‑, we seem to be dealing with a lenition of *t in the initial cluster *h2tV.     *d

The normal reflex of *d is Hitt. /d/: *uódr > /uā́dr/, a-a-tar ‘water’; *h2dént- > Hitt. /Hdánt‑/, a-da-an-t- ‘dried up’. In front of *i, *d gets assibilated to /s/[131]: *diËus > Hitt. /sÏus/, ši-i-ú-uš ‘god’; *diéuot- > Hitt. /síuat‑/, ši-i-a-at-t° ‘day’; see at § for a more detailed treatment. We do not know whether this development took place word-internally as well: in all examples *d is in initial position. Word-initially, in front of *l, *d gets assibilated to /ts/: *dlugh-nu- > Hitt. /tslugnu‑/, za-lu-uk-nu- ‘to postpone’, *dlugh-éh1sh1- > Hitt. /tslugéS‑/, za-lu-keš- ‘to take long’. After the assibilation has taken place, all word-initial dental stops have merged in /t/: *dóru > Hitt. /tā́ru/, ta-a-ru ‘wood’; *dóh3-ei > Hitt. /tā́i/, da-a-i ‘he takes’.     *dh

The normal reflex of *dh is Hitt. /d/: *móldh-ei > Hitt. /mā́ldi/, ma-a-al-di ‘he recites’. In initial position, the dental consonants merge in /t/: *dhéh1-ti > Hitt. /tétsi/, te-e-ez-zi ‘he speaks’; *dhéœhom > Hitt. /tégan/, te-e-kán ‘earth’; etc. If panku- / panga- ‘entire’ reflects *dhbhnœh-(e)u- (see s.v.), it shows loss of word-initial *dh before another stop. This may indicate that its preservation in e.g. ták-na-a-aš /tgnā́s/ ‘earth (’ < *dhœh-m-ós is analogical after the full grade *dhéœh-m > te-e-kán. Fortition of *dh to /t/ is found in *bhódhh2-ei >> Hitt. /patái/, pád-da-i ‘he digs’, where it is due to the following *h2.

The behaviour of *dh in front of *i is important for our understanding of PAnatolian: if in PAnatolian the PIE ‘voiced’ and ‘voiced aspirated’ series indeed merged into a lenis series, we would a priori expect that *dh, just as *d, gets assibilated to /s/ in front of *i. Unfortunately, all examples where we seem to be dealing with *dhi are non-probative: titta-i / titti- ‘to install’ goes back to virtual *dhi‑dhh1-ói- / *dhi-dhh1-i- but could very well be a recent formation that was created after the assibilation ceased to operate; išpartie/a-zi ‘to escape’ seems to reflect *sprdh‑ié/ó‑, but is a NH formation; the 2sg.imp.act. ending ‑t (e.g. īt ‘go!’, arnut ‘deport!’, ašnut ‘take care!’, uešnut ‘rescue!’, etc.) reflects *‑dhi, but may have lost its word-final *‑i before the assibilation took place. This means that there is no solid evidence to prove or disprove that *dh, too, would have been assibilated before *i.[132]     *

The normal reflex of * is Hitt. /k/: *ós > Hitt. /kā́s/, ka-a-aš ‘this (one)’, *éito >> Hitt. /kíta/, ki-it-ta ‘he lies’; *h2rto- > Hitt. /Hrtka‑/, ar-tág-ga- ‘bear’; *-r > Hitt. /sā́kr/, ša-ak-kar ‘dung’.[133] Lenition of * to /g/ may have occurred in za‑ma‑kur ‘beard’ if this reflects /tsmā́gur/ < *smóur (but perhaps this form is a defective spelling for za-ma-‹an-›kur).

In the cluster *RC, * is regularly dropped, as can be deduced from ar-zi /Hártsi/ ‘he holds’ < *h2érti and iš-tar-zi /ɨstártsi/ ‘it ails’ < *stérti.     *œ

The normal reflex of *œ is Hitt. /g/: *h2rœ-i- > Hitt. /Hrgi‑/, ar-ki- ‘white’, *sléiœ-o > Hitt. /slÏga/, ša-li-i-ga ‘he touches’. In initial position the palatovelars merge into /k‑/: *œnh3sénti > Hitt. /knɨSántsi/, ka-ni-eš-ša-an-zi ‘they recognize’; *œénu- > Hitt. /kénu‑/, ge-e-nu- ‘knee’. Fortition of *œ is found in *œh2om > Hitt. /mékan/, me-e-ek-kán ‘great (’, where it is due to the following *h2.     *œh

The normal reflex of *œh is Hitt. /g/: *dhéœhom > Hitt. /tégan/, te-e-kán ‘earth’; *stélœhti > Hitt. /ɨstálgtsi/, iš-tal-ak-zi ‘he flattens’. In initial position the palatovelars merge into /k‑/: *œhésr > Hitt. /kéSr/, ke-eš-šar ‘hand’; *œhimro- > Hitt. /kiMra‑/, gi‑im-ra- ‘field’.   *k

The normal reflex of *k is Hitt. /k/: *kérsti > Hitt. /kárStsi/, kar-aš-zi ‘he cuts’; *skór-ei > Hitt. /ɨskā́ri/, iš-ka-a-ri ‘he cuts’; *tuéko- > Hitt. /tuéka‑/, tu-e-ek-ka- ‘body’; *tuk-ó-r(i) > Hitt. /tukā́ri/, du-ug-ga-a-ri ‘he is visible’; *mrk-ié/ó- > Hitt. /mrkié/á‑/, mar-ki-a- ‘to disapprove of’. Lenition of *k to /g/ has possibly taken place in a-ki /ʔā́gi/ ‘he dies’ < *-ei if this form reflects *k. In the cluster *RkC, *k is probably regularly dropped, as may be deduced from ar-zi /Hártsi/ ‘he holds’ < *h2érti, if this form contained *k.   *g

The normal reflex of *g is Hitt. /g/: *h2tugo- > Hitt. /Hduga‑/, a-tu-ga- ‘terrible’; *iugom > Hitt. /iugan/, i-ú-kán ‘yoke’; *h3érg-ti > Hitt. /Hárgtsi/, ar-ak-zi ‘he gets lost’.   *gh

The normal reflex of *gh is Hitt. /g/: lógh-ei > Hitt. /lā́gi/, la-a-ki ‘he knocks down’; *lgh-ó-ri > Hitt. /ləgā́ri/, la-ga-a-ri ‘he is felled’. In initial position all velars merge into /k‑/: *ghróbh1-ei > Hitt. /krā́bi/, ka-ra-a-pí ‘he devours’. Fortition of *gh to /k/ is found in la-ak-ki-iš-ke/a- /ləkɨské/á‑/, impf. of lāk-i / lak- ‘to knock down’ < *lgh-sé/ó‑.   *kw

The normal reflex of *kw is Hitt. /kw/[134]: *kwis > Hitt. /kwis/, ku-iš ‘who’; *kwér-ti > Hitt. /kwértsi/, ku-e-er-zi ‘he cuts’; *kwt-ru-en- > Hitt. /kwtruen‑/, ku-ut-ru-e-n° ‘witness’; *nekwe > Hitt. /nekw/, ne-ek-ku ‘not?’; *prkw-i- > Hitt. /prkwi‑/, par-ku-i- ‘clean’; *dekws-ié/ó- > Hitt. /tekwSié/á‑/, te-ek-ku-uš-ši-a- ‘to show’. If išpant- ‘night’ indeed reflects *kwspént- (cf. s.v.), it would show loss of initial *kw before obstruents.[135] Lenition of *kw to /gw/ is visible in ša-a-ku-a- /sā́gwa‑/ ‘eye’ < *sókw‑o‑. In the cluster *RkwC the buccal part of */kw/ is regularly lost: *térkwti > */tárʔwtsi/ > Hitt. /tárutsi/, tar-ú-zi ‘he dances’; *trkwsé/ó- > */trʔwské/ó‑/ > Hitt. /truské/á‑/, ta-ru-uš-ke/a- ‘to dance (impf.)’ (similarly in *RC and *RkC, see above).   *gw

The normal reflex of *gw is Hitt. /gw/: *negw-m-ent- > Hitt. /negwmant‑/, ne-ku-ma-an-t- ‘naked’; *dhngw-i- > Hitt. /tngwi‑/, da-an-ku-i- ‘dark’; *trgw-ent- > Hitt. /trgwant‑/, tar-ku-a-an-t- ‘looking angrily’. In initial position the labiovelars merge into /kw‑/: *gwel-uon- > Hitt. /kweluan‑/, ku-e-lu-a-n° ‘washbasin’.   *gwh

The normal reflex of *gwh is Hitt. /gw/: *h1égwhti > Hitt. /ʔégwtsi/, e-ku-zi, e-uk-zi ‘he drinks’; *négwhti > Hitt. /négwtsi/, ne-ku-zi ‘it becomes evening’; *h2lgwh-éh1sh1-r > Hitt. /HlgwéSr/, al-ku-e-eš-šar ‘supplies’. In initial position, the labiovelars merge into /kw‑/: *gwhénti > Hitt. /kwéntsi/, ku-en-zi ‘he kills’. Fortition of *gwh to /kw/ is found in ak-ku-uš-ke/a- /ʔkwské/á‑/, impf. of eku-zi / aku- ‘to drink’ < *h1gwhsé/ó‑.

1.4.4 Fricative     *s

Word-initially before vowels, *s is retained as such: *sV- > Hitt. /#sV‑/: *sésti > Hitt. /séstsi/, še-eš-zi ‘he sleeps’; *sókh1-ei > Hitt. /sā́ki/, ša-a-ak-ki ‘he knows’; *sup-ó-ri > Hitt. /supā́ri/, šu-up-pa-ri ‘he sleeps’; *sih2- > Hitt. /sī‑/, ši-i- ‘one’.

Word-initially before consonants, the reflex depends on the nature of the consonant. Before stops and *h2, *sC- > Hitt. /ɨsC‑/, spelled iš-C°: *sh2-ói-ei > Hitt. /ɨsHā́i/, iš-a-a-i ‘he binds’; *skórei > Hitt. /ɨskā́ri/, iš-ka-a-ri ‘he cuts’; *sph1-ói-ei > Hitt. /ɨspā́i/, iš-pa-a-i ‘he gets full’; *stélœhti > Hitt. /ɨstálgtsi/, iš-tal-ak-zi ‘he flattens’.[136] Before all other consonants (i.e. resonants, *h1, *h3 and *s), *sC- remains unchanged: *sró > Hitt. /srā́/, ša-ra-a ‘upwards’; *sléiœ-o > Hitt. /slÏga/, ša‑li-i-ga ‘he touches’; *sménti > Hitt. /sméntsi/, ša-me-en-zi ‘to pass by’; *snh2énti > Hitt. /snHéntsi/, ša-an-a-an-zi; *sh1-ói-ei > Hitt. /sā́i/, ša-a-i ‘he impresses’; *sh3ngh-u-oi- > Hitt. /snguai‑/, ša-an-ku-a-i- ‘nail’; *ssénti > Hitt. /ssántsi/, ša‑ša‑an-zi ‘they sleep’.

The outcome /ts‑/ as found in za-ma(-an)-kur /tsmá(n)gur/ ‘beard’ < *smóur and za-ak-kar, za-aš-ga-r° /tskar/ ‘excrement’ < *s§Ör is not phonetic. See the lemmata zama(n)kur and šakkar, zakkar / šakn- for a possible explanation.    

In word-internal position, it is best to treat the specific environments separately.

*VsV > Hitt. /VsV/: *nébhesos > Hitt. /nébisas/, ne-e-pí-ša-aš ‘heaven (’, *h1és-ēr > Hitt. /ʔéser/, e-še-er ‘they were’.

*VsPV (where P = any labial consonant): the only example, *uos-bho- > Hitt. /uaSba‑/, a-aš-pa‑, a-aš-ša-pa- ‘clothing’, seems to show that the outcome is /VSPV/, but here the geminate could easily have been secondarily taken over from the verb ešš-tta, ašše/a-zi ‘to wear’.

*VsKV > Hitt. /VsKV/ (where K = any velar stop): *h3rnusé/ó- > Hitt. /ʔrnuské/á‑/, ar-nu-uš-ke/a- ‘to transport (impf.)’; *Hu-sé/ó- > Hitt. /ʔuské/á‑/, ú-uš-ke/a- ‘to see (impf.)’; *kwis-i > Hitt. /kwiski/, ku-iš-ki ‘anyone’.

*VsTV > Hitt. /VsTV/ (where T = any dental stop): *h1ésti > Hitt. /ʔéstsi/, e-eš-zi ‘he is’; *uósth2ei > OH */uā́sti/ > NH /uásti/, a-aš-ti ‘you buy’; *sostos > Hitt. /sastas/, ša-aš-ta-aš ‘bed (’.

*VssV > Hitt. /VssV/: *h1és-si > Hitt. /ʔéssi/, e-eš-ši ‘you are’.

*Vsh1V > Hitt. /VSV/: *h2ltish1énti > Hitt. /HltsiSántsi/, al-zi-iš-ša-an-zi ‘they call (impf.)’; si-sh1-i-ént- > Hitt. /siSiánt‑/, ši-iš-ši-a-an-t- ‘sealed’.

*Vsh2V > Hitt. /VsHV/: *h1esh2enós > Hitt. /ʔisHanā́s/, iš-a-na-a-aš ‘blood (’; *h1esh2ó- > Hitt. /ʔisHā́‑/, iš-a-a- ‘master’.

*Vsh3V: no examples.

*VsrV > Hitt. /VSrV/: *h1és-ri- > Hitt. /ʔéSri‑/, e-eš-ri‑, e-eš-ša-ri- ‘shape, image’. Similarly in *VsrC > Hitt. /VSrC/ (*œhésr-t > Hitt. /kéSrt/, ke-eš-šar-ta ‘with the hand’) and *Vsr# > Hitt. /VSr#/ (*œhésr > Hitt. /kéSr/, ke-eš-šar ‘hand’).[137]

*VslV > Hitt. /VslV/: *h1és-l+ > Hitt. e-eš-li-it, e-eš-lu-ut ‘I must be’.

*VsmV > Hitt. /VsmV/: *h1ésmi > Hitt. /ʔésmi/, e-eš-mi ‘I am’ (never spelled **e‑eš-ša-mi, so not **/ʔéSmi/); *sésmi > Hitt. /sésmi/, še-eš-mi ‘I sleep’ (and not **še-eš-ša-mi = **/séSmi/). Similarly in *Vsm# > OH /Vsun#/: *sésm > OH /sésun/, še-e-šu-un ‘I slept’.

*VsnV > Hitt. /VSnV/: e.g. *usnié/ó- > Hitt. /uSnié/á‑/, uš-ni-a‑, uš-ša-ni-a- ‘to put up for sale’.

*VsiV > Hitt. /VSV/: *h2msósio- > Hitt. /Hntsā́Sa‑/, a-an-za-a-aš-ša- ‘descendant’; *iugosio- > Hitt. /iugaSa‑/, i-ú-ga-aš-ša- ‘yearling’.[138]

*VPsV > Hitt. /VPSV/ (where P = any labial stop): *h1épsi > Hitt. /ʔépSi/, e-ep-ši ‘you seize’; *dhebhsu- > Hitt. /tebSu‑/, te-ep-šu- ‘?’.[139]

*VKsV > Hitt. /VKSV/ (where K = any velar stop): *h1égwhsi > Hitt. /ʔégwsi/, e-uk-ši, e-ku-uš-ši ‘you drink’; *dekws-ie/o- > Hitt. /tekwSie/a‑/, te-ek-ku-uš-ši-e°, te-ek-ku-uš-ši-a- ‘to show, to present (oneself)’; *no-nogwhs-ie/o- > Hitt. /nanagwSie/a‑/, na-na-ku-uš-ši-a- ‘to become dark’.

*VTsV > Hitt. /VTSV/ (where T = any dental stop): *h1édsi > Hitt. /ʔédSi/, e-ez-ši, [e-ez-za-a]š-ši ‘you eat’.

*Vh1sV. Here we must take the accentuation into account, namely *Õh1sV yields Hitt. /ÕsV/ whereas *Vh1sÕ > Hitt. /VSÕ/: *h1éh1so > Hitt. /ʔésa/, e-ša ‘he sits down’ vs. *h2eh1séh2- > Hitt. /HaSā‑/, a-a-aš-ša-a- ‘hearth’ (see s.v. āššā- for an extensive treatment of this word); *h2ih1/3s-éh2- > Hitt. /HiSa‑/, i-iš-ša- ‘carriage pole’.

*Vh2sV > Hitt. /VHsV/:[140] *péh2s-o > Hitt. /páHsa/, pa-a-ša ‘he protects’; *pleh2so- > Hitt. /plaHsa‑/, pa-la-a-ša‑, a garment.

*Vh3sV. Here we must take the accentuation into account as well, namely *Õh3sV yields Hitt. /ÕsV/ whereas *Vh3sÕ > Hitt. /VSÕ/: *póh3sei > Hitt. /pā́si/, pa-a-ši ‘he sips’ vs. *poh3s-uén-ti > Hitt. /pāSuántsi/, pa-a-aš-šu-an-zi ‘to sip’ (with analogical ā); *h2ih1/3s-éh2- > Hitt. /HiSa‑/, i-iš-ša- ‘carriage pole’.    

The reflexes of clusters involving *Rs and *RHs need special attention, especially the difference between clusters with and without laryngeals.

*VrsV > Hitt. /VRV/: *h1orso- > Hitt. /ʔaRa‑/, a-ar-ra- ‘arse’.

*VlsV > Hitt. /VlsV/?: *polso- (or *plso-?) > Hitt. /palsa‑/ (or /plsa‑/?), pal-ša- ‘road’.

*VmsV > Hitt. /VSV/: *h2ems-u- > Hitt. /HaSu‑/, a-aš-šu- ‘king’. When lenited, the outcome is /VsV/, however: *h2ómsei > Hitt. /Hā́si/, a-a-ši ‘he procreates, she gives birth’.[141]

*VnsV > Hitt. /VSV/?: de/ons-u- (or *de/oms-u‑) > Hitt. /taSu‑/, da-aš-šu- ‘powerful’.

*VrHsV > Hitt. /VrsV/ if the reconstruction of *ārši (as inferred from 3sg.pret.act. a-a-ar-aš-ta) ‘he tills (the soil)’ as *h2órh3-s-ei is correct.

*VlHsV: no examples.

*VmHsV > Hitt. /VnsV/: *h2ómh1-s-ei > Hitt. /ʔā́nsi/, a-a-an-ši ‘he wipes’.

*VnHsV > Hitt. /VntsV/: *œénh1-su- > Hitt. /kéntsu‑/, ge-en-zu- ‘lap’.

*VrHsC > Hitt. /VrɨsC/: *uerh1-sé/ó- > Hitt. /uerɨské/á‑/, ú-e-ri-iš-ke/a- ‘to call (impf.)’.[142]

*VlHsC > Hitt. /VLɨsC/: *élh1st > Hitt. /káLɨsta/, kal-li-iš-ta ‘he called’.

*VmHsC > Hitt. /VMɨsC/: *demh2sh2ó- > Hitt. /taMɨsHā́‑/, dam-me-iš-a-a- ‘damage’.[143]

*VnHsC: no examples.

*CrsV > Hitt. /CrSV/: *krs-éntu > Hitt. /krSántu/, kar-ša-an-du, kar-aš-ša-an-du ‘they must cut’.

*ClsV > Hitt. /ClSV/: *kwlsénti > Hitt. /kwlSántsi/, gul-ša-an-zi, gul-aš-ša-an-zi ‘they carve’.

*CmsV > Hitt. /CntsV/: *h2msósio- > Hitt. /Hntsā́Sa‑/, a-an-za-a-aš-ša- ‘descendant’.

*CnsV > Hitt. /CntsV/: *nsós > Hitt. /ntsā́s/, an-za-a-aš ‘us’.

*CrHsV: no examples.

*ClHsV > Hitt. /ClɨSV/: *lh1sénti > Hitt. /klɨSántsi/, ga-li-iš-ša-an-zi ‘they call’.

*CmHsV > Hitt. /CnɨSV/: *h2mh1s-énti > Hitt. /HnɨSántsi/, a-ni-eš-ša-an-zi ‘they wipe’.

*CnHsV > Hitt. /CnɨSV/: *œnh3sénti > Hitt. /knɨSántsi/, ka-ni-eš-ša-an-zi ‘they recognize’.

*CrHsC > Hitt. /CrɨsC/: *pri-prh1-sé/ó- > Hitt. /priprɨské/á‑/, pa-ri-ip-ri-iš-ke/a- ‘to blow (impf.)’, *h1rh1-sé/ó- > Hitt. /ʔrɨské/á‑/, a-ri-iš-ke/a‑, a-re-eš-ke/a- ‘to consult an oracle (impf.)’.

*ClHsC > Hitt. /ClɨsC/?: *mlh2sk-u-(?) > Hitt. /mlɨsku‑/, ma-li-iš-ku- ‘weak, light’.

*CmHsC > Hitt. /CmɨsC/: *dmh2sé/ó- > Hitt. /tmɨské/á‑/, da-me-eš-ke/a- ‘to oppress (impf.)’.[144]

*CnHsC > Hitt. /CəSɨC/: *h3nh3sé/ó- > Hitt. /HəSɨké/á‑/, a-aš-ši-ke/a- ‘to sue (impf.)’.    

In clusters containing *s and stops we often see the rise of the anaptyctic vowel /ɨ/[145] (sometimes only within the Hittite period): *dhh1sé/ó- > OH /tské/á‑/, za-aš-ke/a- > OH /tsɨké/á‑/, zi-ik-ke/a- ‘to place (impf.)’; *h1d-sé/ó- > MH /ʔdské/á‑/, az-za-ke/a- (MH/MS) > MH/NH /ʔdsɨké/á‑/, az-zi-ke/a- (MH/MS) ‘to eat (impf.)’; *h2t-sé/ó- > OH /Htsɨké/á‑/, a-az-zi-ik-ke/a- ‘to pierce, to prick (impf.)’;[146] *h1p-sé/ó- > Hitt. /ʔpɨské/á‑/, ap-pí-iš-ke/a- ‘to seize (impf.)’; *téks-ti > Hitt. /tákɨstsi/, ták-ki-iš-zi ‘he devises’; *lgh-sé/ó- > Hitt. /ləkɨské/á‑/, la-ak-ki-iš-ke/a- ‘to fell (impf.)’;[147] *œ(h)sd‑uént- > Hitt. /kɨsduánt‑/, ki-iš-du-a-an-t- ‘hungry’; *œhsréi > Hitt. /kɨSrí/, ki‑iš-ša-ri-i, ki-iš-ri ‘hand (’; and compare the reflexes of *CRHsC (above), *‑Vh2s and *‑Vks (below).    

The reflex of word-final *s is /s/.

*-Vs > Hitt. /‑Vs/: *‑os > Hitt. /‑as/, ‑aš, ending; *sés > Hitt. /sés/, še-e-eš ‘sleep!’; *ós > Hitt. /kā́s/, ka-a-aš ‘this (one)’.

*-VKs > Hitt. /VKɨs/: *h1ó-s > Hitt. /ʔā́kɨs/, a-ak-ki-iš ‘he died’.

*-VPs > Hitt. /VPs/: *stóp-s > Hitt. /ɨstā́ps/, iš-tap-pa-aš ‘he plugged up’; *ghróbh1-s > Hitt. /krā́bs/, ga-ra-pa-aš ‘he devoured’.

*-VTs > Hitt. /‑VTs/: *ghéu-t-s > Hitt. /kÜts/, ku-ú-uz-za ‘wall’, *diéuot-s > Hitt. /síuats/, ši-i-a-az ‘day’; *œhrh1ód-s > Hitt. /krʔā́ds/, ka-ra-a-az ‘entrails’.

*-Vh1s > Hitt. /‑Vs/: *dhéh1s > Hitt. /tés/, te-e-eš ‘you spoke’.

*-Vh2s > OH /‑VHs/ > NH /VHɨs/: *‑éh2-s > OH /‑aHs/, °Ca-a-a-aš (OH/NS) > NH /‑aHɨs/, °Ca-a-i-iš (NS), 3sg.pret.act. forms of verbs in ‑aŸŸ-i.

*-Vh3s > Hitt. /Vs/: *dóh3-s > Hitt. /tā́s/, da-a-aš ‘he took’.

*‑Vrs > Hitt. /‑Vrs/: h1órs > Hitt. /ʔā́rs/, a-ar-aš ‘he arrived’; *kers > Hitt. /kárs/, kar-aš ‘cut!’.[148]

*-Vls: no examples.

*-Vms > Hitt. /‑Vs/: *‑oms > OH /‑us/, (ú‑)uš, > NH /‑os/, ‑(u‑)uš, of o‑stems.

*-Vns > Hitt. /‑Vs/: *‑uen-s > Hitt. /‑uas/, a-aš, of the verbal noun in ar.

1.4.5 Laryngeals

Since the PIE laryngeals, *h1, *h2 and *h3, show some mergers in the pre-PAnatolian period already, it is in my view best to first treat their PAnatolian reflexes and then see what these yield in Hittite.


The reflexes of word-initial laryngeals in PAnatolian have been treated in detail in Kloekhorst 2006b., where the following overview has been given (note that the order of the laryngeals is not numerical in order to make the mergers more transparent: R = r, l, m, n, i, u; T = any stop and s):

            PIE                     PAnat.                            PIE                            PAnat.

            *h2e-       >          *Ha-                              *h2o-         


            *h3e-       >          *Ho-                              *h3o-                          *ʔo-


            *h1e-       >          *ʔe-                               *h1o-         


            *h2R-      >           *HR-                             *h2T-          >              *HT-


            *h3R-                                                         *h3T-                         

                                         *ʔR-                                                                  *T-

            *h1R-                                                          *h1T-                         

The reflexes of the PAnatolian sequences in Hittite are as follows:

PAnat. */Ha‑/ > Hitt. /Ha‑/: *h2ent- > PAnat. */Hant‑/ > Hitt. /Hant‑/, a-an-t- ‘forehead’; *h2erœi- > PAnat. */Hargi‑/ > Hitt. /Hargi‑/, ar-ki- ‘white’; *h2emsu- > PAnat. */HaSu‑/ > Hitt. /HaSu‑/, a-aš-šu- ‘king’.

PAnat. */Ho‑/ > Hitt. /Ha‑/: *h3érōn+s > PAnat. */Hórōns/ > Hitt. /Hā́ras/, a-a-ra-aš ‘eagle’; *h3épr > PAnat. */Hópr/ > Hitt. /Hā́pr/, a-a-ap-pár ‘business’.

PAnat. */ʔe‑/ > Hitt. /ʔe‑/: *h1ésmi > PAnat. */ʔésmi/ > Hitt. /ʔésmi/, e-eš-mi ‘I am’; *h1érmn > PAnat. */ʔérmn/ > Hitt. /ʔérmn/, e-er-ma-an ‘illness’; *h1ésh2r > PAnat. */ʔésHr/ > Hitt. /ʔésHr/, e-eš-ar ‘blood’.

PAnat. */ʔo‑/ > Hitt. /ʔa‑/: *h2ómh1sei > PAnat. */ʔÖmsei/ > Hitt. /ʔā́nsi/, a-an-ši ‘he wipes’; *h2óuth2ei > PAnat. */ʔóutai/ > Hitt. /ʔáuti/, a-ut-ti ‘you see’; *h2óro > PAnat. */ʔÖro/ > Hitt. /ʔā́ra/, a-a-ra ‘right, properly’; *h3órœhei > PAnat. */ʔÖrgei/ > Hitt. /ʔā́rgi/, a-ar-ki ‘he mounts’; *h1órei > PAnat. */ʔÖrei/ > Hitt. /ʔā́ri/, a-a-ri ‘he arrives’.

PAnat. */HR‑/ > Hitt. /HR‑/: *h2rto- > PAnat. */Hrtḱo‑/ > Hitt. /Hrtka‑/, ar-ták‑ka- ‘bear’; *h2lt-i- > PAnat. */Hlti‑/ > Hitt. /Hltsi‑/, al-zi- ‘to call’; *h2méh1sh2o- > PAnat. */HméʔsHo‑/ > Hitt. /HmésHa‑/, a-me-eš-a- ‘spring’; *h2nénti > PAnat. */Hnánti/ > Hitt. /Hnántsi/, a-na-an-zi ‘they draw water’; *h2imno- > PAnat. */Himno‑/ > Hitt. /HiMa‑/, i-im-ma- ‘imitation’; *h2uh1ént- > PAnat. */Huʔánt‑/ > Hitt. /Hoánt‑/, u-a-an-t- ‘wind’.

PAnat. */ʔRC‑/ > Hitt. /ʔRC‑/: *h1mn- > PAnat. */ʔmn‑/ > Hitt. /ʔM‑/, am-m° ‘me’; *h1ndom > PAnat. */ʔndom/ > Hitt. /ʔndan/, an-da-an ‘inside’; *h1idhi > PAnat. */ʔidi/ > Hitt. /ʔid/, i-it ‘go!’; *h1rsénti > PAnat. */ʔrsánti/ > Hitt. /ʔrSántsi/, ar-ša-an-zi ‘they flow’; *h2u-s§é/ó- >> PAnat. */ʔusḱé/ó‑/ > Hitt. /ʔuské/á‑/, ú‑uš‑ke/a- ‘to see (impf.)’; *h3rtóri > PAnat. */ʔrtóri/ > Hitt. /ʔrtári/, ar-ta-ri ‘he stands’; *h3rœh-i- > PAnat. */ʔrgi‑/ > Hitt. /ʔrgi‑/, ar-ki- ‘testicle’; *h3niéti > PAnat. */ʔniéti/ > Hitt. /ʔniétsi/, a-ni-e-ez-zi ‘he works’.

PAnat. */ʔRV‑/ > Hitt. /RV‑/ (except PAnat. */ʔrV‑/, see below): *h1lenghti > PAnat. */ʔlengti/ > Hitt. /líktsi/, li-ik-zi ‘he swears’; *h3néh3mn > PAnat. */ʔlóʔmn/ > Hitt. /lā́mn/, la-a-ma-an ‘name’; *h1uorœ- > PAnat. */ʔuarg‑/ > Hitt. /uarg‑/ in a‑a-ar-ki-ma‑, object in which the door-axle is fixed and turns; *h1uorso- > PAnat. */ʔuorso‑/ > Hitt. /uarsa‑/, a-ar-ša- ‘fog, mist’.[149]

PAnat. */ʔrV‑/ > Hitt. /ʔrV‑/: *h1rénti > PAnat. */ʔránti/ > Hitt. /ʔrántsi/, a-ra-an-zi ‘they arrive’; *h3rénto > PAnat. */ʔránto/ > Hitt. /ʔránta/, a-ra-an-ta ‘they stand’; *h3róiei > PAnat. */ʔrÖiē/ > Hitt. /ʔrā́i/, a-ra-a-i ‘he rises’.

PAnat. */HT‑/ > Hitt. /HT‑/: *h2stËr > PAnat. */HstËr/ > Hitt. /Hstér/, a-aš-te-er- ‘star’; *h2dént- > PAnat. */Hdánt‑/ > Hitt. /Hdánt‑/, a-da-an-t- ‘parched’; *h2tié/ó- > PAnat. */Htié/ó‑/ > Hitt. /Htsié/á‑/, a-az-zi-e°, a-az-zi-a- ‘to pierce, to prick’; *h2téugti > PAnat. */HdÜgti/ > Hitt. /HdÜgtsi/, a-tu-uk-zi ‘he is terrible’.

PIE *h1/3T- > PAnat. /T‑/ > Hitt. /T‑/: *h1p-ói-ei > PAnat. */pÖiē/ > Hitt. /pā́i/, pa‑a-i ‘he gives’; *h1siéti > PAnat. */siéti/ > Hitt. /siétsi/, ši-i-e-ez-zi ‘he shoots’; *h1t-i-sténi > PAnat. */tisténi/ > Hitt. /tsisténi/, zi-iš-te-e-ni ‘you (pl.) cross over’.


The reflexes of word-internal laryngeals are as follows (note that the three laryngeals always colour a neighbouring *e; a PAnat. */H/ followed by /u/ gets phonemicized as /Hw/, cf. Kloekhorst 2006b):

                  PIE                 PAnat.            early OH            late OH

                *Vh2V     >      */VHV/      >      /VHV/        >      /VHV/


                                        */VʔV/       >      /VʔV/         >      /VV/



*Vh2V: *péh2ur > PAnat. */páHwr/ > Hitt. /páHwr/, pa-a-ur ‘fire’; *tieh2-oi- > PAnat. */tiaHoi‑/ > Hitt. /tsaHai‑/, za-a-a-i- ‘battle’; *h1ndhuéh2ōs > PAnat. */ʔnduáHōs/ > Hitt. /ʔnduáHas/, an-tu-a-a-a-aš ‘human being’; *h2uh2o- > PAnat. */HuHo‑/ > Hitt. /HuHa‑/, u-u-a- ‘grandfather’; *méih2ur > PAnat. */mËhwr/ > Hitt. /méhwr/, me-e-ur ‘period, time’ (with lenition); *nóh2ei > PAnat. */nÖhē/ > Hitt. /nā́hi/, na-a-i ‘he fears’ (with lenition); etc.

*Vh1/3V: *h2éih3-eu-eies > PAnat. */HËʔeuēs/ > early OH /Héʔaues/, é-e-a-u-e-eš (OS) > late OH /Héaues/, é-e-a-u-e-š=a (OS) ‘rains (’;[150] *néih1/3-o > PAnat. */nËʔo/ > early OH /néʔa/, ne-e-a > younger Hitt. /néa/, ne-e-a ‘he turns, leads’; *dóh3ei > PAnat. */dÖʔē/ > early OH /tā́ʔi/, da-a-i > younger Hitt. /tā́i/, da‑a‑i ‘he takes’.


                  PIE                         PAnat.                          Hitt.


                *Vh3T                     */VʔT/       >                  /ÅT/


                                                               (T = any stop, but not *s)

Note that PAnat. */ÕʔDV/ (in which D = any lenis stop) yields Hitt. /ÕDV/, whereas */VʔDÕ/ yields Hitt. /VTÕ/, showing fortition of */D/ due to assimilation with /ʔ/.

Examples: *dhéh1t > PAnat. */déʔt/ > Hitt. /tét/, te-e-et ‘he said’; *léh2pt(o) > PAnat. */láʔpto/ > Hitt. /lā́pta/, la-a-ap-ta ‘it glowed’; *séh2goi- > PAnat. */sáʔgai‑/ > Hitt. /sā́gai‑/, ša-a-ga-i- ‘sign, omen’; *uóh2gei > pre-Hitt. */uÖʔgē/ > Hitt. /uā́gi/, a-a-ki ‘he bites’; *uh2génti >> */u̯ʔgánti/ > pre-Hitt. /uəʔgánti/ > Hitt. /uəkántsi/, a-ak-kán-zi ‘they bite’; *dóh3th2e > PAnat. */dÖʔta/ > Hitt. /tā́ta/, da-a-at-ta ‘you took’.


                  PIE                         PAnat.                          Hitt.

                *Vh2s      >              */VHs/       >                  /VHs/

                *Vh3s                                                              /ÅsC/ and /Ås#/

                                                */Vʔs/                                /ÐsV/

                *Vh1s                                                              /VSÕ/


*Vh2s: *péh2so > PAnat. */páHso/ > Hitt. /páHsa/, pa-a-ša ‘he protects’; *h1ndhuh2sos > PAnat. */ʔnduHsos/ > Hitt. /ʔnduHsas/, an-tu-u-ša-aš ‘human being (’; *pleh2so- > PAnat. */plaHso‑/ > Hitt. /plaHsa‑/, pa-la-a-ša‑, a garment.

*Vh1/3s: *dhéh1s > PAnat. */déʔs/ > Hitt. /tés/, te-e-eš ‘you said’; *dhéh1si > PAnat. */déʔsi/ > Hitt. /tési/, te-ši ‘you say’; *h1éh1so > PAnat. */ʔéʔso/ > Hitt. /ʔésa/, e-ša ‘he sits down’; *póh3sei > PAnat. */pÖʔsē/ > Hitt. /pā́si/, pa-a-ši ‘he sips’; *poh3suénti > PAnat. */poʔsuánti/ > Hitt. /pāSuántsi/, pa-a-aš-šu-an-zi ‘to sip’ (with analogical ‑ā‑); *h2eh1seh2- > PAnat. */Haʔsáʔ‑/ > Hitt. /HaSā́‑/, a-aš-ša-a- ‘hearth’.


                  PIE                         PAnat.                          Hitt.

                *Vh2R     >              */VHR/ ?   >                  /VHR/ ?

                *Vh3R                                                             /ÅRC/

                                                */VʔR/                           /ÐRV/

                *Vh1R                                                             /VRRÕ/



*Vh2R:[151] *meh2roi-(?) > PAnat. */maHroi‑/ > Hitt. /maHrai‑/, ma-a-ra-i‑, a body part of animals; *tieh2roi- > PAnat. */tiaHroi‑/ > Hitt. /tsaHrai‑/, za-a-ra-i- ‘knocker(?)’; *móh2lo-(?) > PAnat. */mÖHlo‑/ > Hitt. /mā́Hla‑/, ma-a-a-la- ‘branch of a grapevine’.

*Vh1/3R: *h3néh3mn > PAnat. */ʔlóʔmn/ > Hitt. /lā́mn/, la-a-ma-an ‘name’; *tiéh1no- > PAnat. */tiéʔno‑/ > Hitt. /tséna‑/, zé-e-na- ‘autumn’; *dhéh1mi > PAnat. */déʔmi/ > Hitt. /témi/, te-e-mi ‘I say’; *tiéh1ri- > PAnat. */tiéʔri‑/ > Hitt. /tséri‑/, ze‑e-ri- ‘cup’; *h3eh3nóh3- > PAnat. */Hoʔnáʔ‑/ Hitt. /HaNá‑/, a-an-na- ‘to sue’.


                  PIE                         PAnat.                          Hitt.


                *Th3V                     */TʔV/           >                  /TV/


                                                               (T = any stop, but not *s)

Note that *VDh2V (in which D = any lenis stop) yields Hitt. /VTV/, i.e. the *D has undergone fortition to /T/ due to assimilation to the following *h2.[152]

Examples: *‑th2e > PAnat. */‑tʔa/ > Hitt. /‑ta/, °t-ta, 2sg.pret.act. ending of the i-conjugation; *dh3énti > PAnat. */dʔánti/ > Hitt. /tántsi/, da-an-zi ‘they take’; *dhh1‑ói-ei > PAnat. */tʔÖiē/ > Hitt. /tā́i/, da-a-i ‘he places’; *dhh1iénti > PAnat. */dʔiánti/ > Hitt. /tiántsi/, ti-an-zi ‘they place’;[153] *œh2-i- > PAnat. */mékʔi‑/ > Hitt. /méki‑/, me-ek-ki- ‘much, many’; *bhodhh2ei > PAnat. */batʔai/ > Hitt. /patai/, pád-da-i ‘he digs’; *ghróbh1-ei > PAnat. */grÖbʔē/ > Hitt. /krā́bi/, ka-ra-a-pí ‘he devours’.[154]


                  PIE                         PAnat.                          Hitt.

                *sh2V      >              */sHV/       >                  /sHV/


                                                */sʔV/            >                  /SV/



*sh2V: *sh2óiei > PAnat. */sHÖiē/ > Hitt. /ɨsHā́i/, iš-a-a-i ‘he binds’; *h1ésh2r > PAnat. */ʔésHr/ > Hitt. /ʔésHr/, e-eš-ar ‘blood’; *h1esh2ó- > PAnat. */ʔesHó‑/ > Hitt. /ʔisHā́‑/, iš-a-a- ‘master’.

*sh1/3V: *h2ltish1énti > PAnat. */Hltisʔánti/ > Hitt. /HltsiSántsi/, al-zi-iš-ša-an-zi ‘they call (impf.)’; si-sh1-i-ént- > PAnat. */sisʔiánt‑/ > Hitt. /siSiánt‑/, ši-iš-ši-a‑an‑t- ‘sealed’.


                  PIE                                             Hitt.


                *#Rh3                                    /#RV/


Examples: *lh1énti > Hitt. /lántsi/, la-an-zi ‘they loosen’; *lh1uti- > Hitt. /lutsi‑/, lu‑uz-zi- ‘public duty’; *mh2óiei > Hitt. /mā́i/, ma-a-i ‘he grows’.


                  PIE                             PAnat.                          Hitt.


                                                    */CRHV/       >          /CRHV/



                *CRh1V          >          */CRʔV/        >          /CRʔV/


PAnat. */CRHV/: *plh2-i- > PAnat. */plHi‑/ > Hitt. /plHi‑/, pal-i- ‘broad’; *bhrh2/3énti > PAnat. */brHánti/ > Hitt. /prHántsi/, pár-(a‑)a-an-zi ‘they pursue’; *trh2uénti > PAnat. */trHwánti/ > Hitt. /trHwántsi/, tar-u-a-an-zi ‘they conquer’; *ulh3énti > PAnat. */ulHánti/ > Hitt. /uəlHántsi/, a-al-(a‑)-a-an-zi ‘they hit’.

PAnat. */CRʔV/: *pri-prh1-ói-ei > PAnat. */priprʔáiē/ > Hitt. /priprʔā́i/, pa-ri-pa‑ra-a-i ‘he blows’;[155] *œrh1ód- > PAnat. */grʔÖd‑/ > Hitt. /krʔā́d‑/, ka-ra-a-t° ‘entrails’; *h1rh1iéti > PAnat. */ʔrʔiéti/ > Hitt. /ʔrʔiétsi/, a-ri-e-ez-zi ‘he consults an oracle’.


                  PIE                                             Hitt.


                *VRh3                                    /VRRV/


Examples: *h1órh1ei > Hitt. /ʔā́Ri/, a-ar-ri ‘he washes’; *tinh1énti > Hitt. /tsiNántsi/, zi-in-na-an-zi ‘they finish’; *molh2ei > Hitt. /maLai/, ma-al-la-i ‘he mills’; *h2orh3ei > Hitt. /HaRai/, ar-ra-i ‘he grinds’.


Interconsonantally, the three laryngeals are all lost,[156] but there are only a few good examples: *plth2sh2o- > Hitt. /pltsHa‑/, pal-za-a-a‑, pal-za-aš-a- ‘pedestal’; *dhh1sé/ó- > early OH /tské/á‑/, za-aš-ke/a- > late OH /tsɨké/á‑/, zi-ik-ke/a‑, ‘to place (impf.)’; *h1/3uenh1-ti > Hitt. /uentsi/, ú-en-zi ‘he copulates’; *uorh1œent- > Hitt. /uargant‑/, a-ar-kán-t- ‘fat’; *dhh1-sh2-oi- > Hitt. /tsHai‑/, za-aš-a-i- ‘dream’; *œénh1-su- > Hitt. /kéntsu‑/, ge-en-zu- ‘lap’.[157]

In clusters containing resonants and ‑s‑, there is often a different outcome when a laryngeal is part of it, e.g. *VmsV > Hitt. /VSV/, but *VmHsV > Hitt. /VnsV/. See §, the treatment of *s, for details.


Word-finally, the laryngeals are all lost. Note that in *Cuh2# the *u is lowered to /o/ and in *Cih2# the *i to /e/.

Examples: *léh1 > Hitt. /lé/, le-e ‘not’ (prohib.); *lóh1 > Hitt. /lā́/, la-a ‘let go!’; *duoiom *h3esth1ih1 > Hitt. /tānHasti/, da-a-an-a-aš-ti ‘double-bone’; *mih1éh1sh1 > Hitt. /miés/, mi-i-e-eš ‘be gentle!’; *sókh1 > Hitt. /sā́k/, ša-a-ak ‘know!’; *œh2 > Hitt. /mék/, me-e-ek ‘many, numerous (’; *‑eh2 > Hitt. ‑a, ending; *sókweh2 > Hitt. /sā́gwa/, ša-a-ku-a ‘eyes (’;

*h1oh1suh2 > Hitt. /ʔáSo/, a-aš-šu-u ‘goods (’; *ih2 > Hitt. /ke/, ke-e ‘these (nom.-acc. pl.n.)’; *dóh3 > Hitt. /tā́/, da-a ‘take!’.[158]

1.4.6 Liquids     *l

PIE *l in principle is retained in Hittite as /l/, except in the environments *VlHV > Hitt. /VLV/ and *VlHsC > Hitt. /VLɨsC/ where assimilation of the laryngeal to the preceding *l results in a fortis /L/.

Examples: *lóh1ei > Hitt. /lā́i/, la-a-i ‘he releases’; *lóghei > Hitt. /lā́gi/, la-a-ki ‘he makes fall down’; *léuk-to > Hitt. /lÜkta/, lu-uk-ta ‘it dawns’; *plh2-i- > Hitt. /plHi‑/, pal-i- ‘wide, broad’; *h2l(e/o)ugho- > Hitt. /Hluga‑/ or /Hlūga‑/, a-lu-ka- ‘message’; *séuh1-el > Hitt. /sÜil/, šu-ú-il ‘thread’; *molh2ei > Hitt. /maLai/, ma-al-la-i ‘he mills’; *élh1st(o) > Hitt. /káLɨsta/, kal-li-iš-ta ‘he called’.   Fortition

From MH times onwards, we occasionally find fortition of intervocalic /l/ to /L/, e.g. uš-tu-la-aš (OS) > a-aš-túl-la-aš (MH/MS) ‘sin (’ ; a-aš-du-ú-li (MH/MS) > a-aš-túl-li (NS) ‘sin (’; iš-i-ú-la-a-° (NS) > iš-i-ul-la-a-° (NH) ‘to bind by treaty’. Whether we are dealing with a phonetically regular process is unclear, cf. Melchert 1994a: 165.     *r

PIE *r is in principle retained in Hittite as /r/, except in *VrHV > /VRV/ and *VrsV > Hitt. /VRV/ where assimilation of the laryngeal and of *s to the preceding *r results in a fortis /R/.

Examples: *h1rénti > Hitt. /ʔrántsi/, a-ra-an-zi ‘they arrive’; *sro > Hitt. /srā/, ša‑ra-a ‘upwards’; *h2rœ-i- > Hitt. /Hrgi‑/, ar-ki- ‘white’, *sr-li- > Hitt. /srli‑/, šar‑li- ‘superior’; *spórei > Hitt. /ɨspā́ri/, iš-pa-a-ri ‘he spreads out’; *kérsti > Hitt. /kárstsi/, kar-aš-zi ‘he cuts’; *h1ésri- > Hitt. /ʔéSri‑/, e-eš(-ša)-ri- ‘shape, image’; *supr-ié/ó- > Hitt. /suprié/á‑/, šu-up-pa-ri-a- ‘to sleep’; *h1ésh2r > Hitt. /ʔésHr/, e‑eš-ar ‘blood’; *œhésr > Hitt. /kéSr/, ke-eš-šar ‘hand’; *h1p-i-Ër > Hitt. /piér/, pí-i-e-er ‘they gave’; *§Ër > Hitt. /kér/, ke-er ‘heart’; etc.

Assimilation: *h1órh1ei > Hitt. /ʔā́Ri/, a-ar-ri ‘he washes’; *h2orh3ei > Hitt. /HaRai/, ar-ra-i ‘he grinds’; *h1orso- > Hitt. /ʔaRa‑/, a-ar-ra- ‘arse’.   Loss

Word-final *r is lost after unaccented *o or *ō.[159] This is only attested in the endings of the middle (for which see especially Yoshida 1990: 112f.) and in the forms of neuters in ‑r/n-.

Examples: *h1éh1s-or > Hitt. /ʔésa/, e-ša ‘he sits down’ vs. *tuk-ór +i > Hitt. /tukā́ri/, du-ug-ga-a-ri ‘is visible’; *uoh2œéh1sh1-ōr > Hitt. /uagéSa/, a-ag-ge-eš-ša, a kind of bread (, and *h2t-ótōr > Hitt. /Htā́da/, a-at-ta-a-da ‘wisdom (’ vs. *udÖr > Hitt. /uɨdā́r/, ú-i-ta-a-ar ‘waters (’.   Fortition

From MH times onwards we occasionally find fortition of intervocalic /r/ to /R/: an-tu-u-ri-a‑, an-tu-ri-a- vs. an-dur-ri-a- (NS) ‘interior’; a-ku-ut-ta-rV‑, a-ku-ut-tar‑a° vs. a-ku-ut-tar-ra- (NS) ‘drinker’; a-a-ap-pa-ra-az (OS) vs. a-ap-pár-ra-az (NS) ‘business (abl.)’; a-a-ra-na-aš (OS) vs. ar-ra-n[a-aš] (NS) ‘eagle (’; iš-pa-ra-an-zi (OS) vs. iš-pár-ra-an-zi (MS, NS) ‘they strew’; iš-pa-ru-uz-zi (OH/NS) vs. iš-pár-ru-uz-zi (MH/MS) ‘rafter’; kat-te-ra- (MS), kat-te-e-ra- (MS) vs. kat-te-er-ra- (NS) ‘lower, inferior’; ge-nu-uš-ša-ri-a-an-t- vs. ge-nu-šar-ri-a‑an-t- (NS) ‘kneeling’; me-re-er (OS) vs. me-er-ra-a-an-ta-ru (NS), me-er-ra‑an‑t- (NS) ‘to disappear’; na-a-ša-ra-az (MH/MS) vs. na-a-šar-ra-az (NS) ‘fear (’; pa-ra-a-an-ta (OS) vs. pár-ra-an-ta (MH/MS) ‘across (to)’; ta‑pa‑ri-a- vs. ta-pár-ri-a- (NH) ‘to rule’; etc.

If we compare words like pé-e-ra-an, which remains unchanged throughout Hittite, it is difficult to interpret this phenomenon as a phonetically regular development. Perhaps we are dealing with mere variation in spelling (note that most examples show a sign CVC: dur, tar, pár, ar, šar; but this does not go for me-er-ra- and kat-te-er-ra‑). Cf. Melchert (1994a: 165-6) for another interpretation.   Dissimilation

In OH ú-ra-a-ni /urʔā́ni/, MH/NH a-ra-a-ni /uərʔā́ni/ ‘burns’ < *urh1-ór+i we encounter dissimilation of the second *r to Hitt. /n/.

1.4.7 Nasals     *m

Word-initially before vowels, *m is retained as /m/: *mégh2 > Hitt. /mék/, me-e-ek ‘many’ (; *mérti > Hitt. /mértsi/, me-er-zi ‘he disappears’; *móldhei > Hitt. /mā́ldi/, ma-a-al-di ‘he recites’; etc.

Word-initially before stops, we would expect that *mC- yields Hitt. /mC‑/, phonetically realized as [əmC‑]. The only example is *mdhró- > Hitt. /ndrā́‑/, an‑ta‑ra-a- ‘blue’, where */m/ is assimilated to /n/ due to the following /d/, however. In all other cases where we are dealing with *mT- (in which T = any stop), this sequence is the zero grade of a root *meT‑, on the basis of which the regular outcome /mT‑/ = [əmT‑] has been altered to /məT‑/: *mgh2éh1sh1- > Hitt. /məkéS‑/, ma-ak-ke-e-eš-š° ‘to become abundant’ (belonging to mekk(i)- ‘many, much’). Note that if mai-i / mi- ‘to grow’ indeed reflects *mh2-oi- / *mh2-i‑, it would show that *mh2V > Hitt. /mV‑/.

Word-initially before resonant, *mR- yields Hitt. /mR‑/: *mlit- > Hitt. /mlit‑/, ma-li-it-t° ‘honey’; *mnieh2- > Hitt. /mniaH‑/, ma-ni-a-a-° ‘to distribute’; *mréntu > Hitt. /mrántu/, ma-ra-an-du ‘they must disappear’; *mr-nu- > Hitt. /mrnu‑/, mar-nu- ‘to cause to disappear’.  

For the word-internal position, it is best to treat the different phonetic surroundings separately. I have given here only the reflexes of clusters where *m is the first member.

*CmV > Hitt. /CmV/: *h2méh1sh2o- > Hitt. /HmésHa‑/, a-me-eš-a- ‘spring’; *smén-ti > Hitt. /sméntsi/, ša-me-en-zi ‘he passes by’[160]; *tménkti > Hitt. /tméktsi/, ta-me-ek-zi ‘he attaches’.

*CmT (in which T = any stop) > Hitt. /CaT/ with fortition of the stop: *kmt-o > Hitt. /kata/, kat-ta ‘downwards’; *kmb(h)-i- > Hitt. /kapi‑/, kap-pí- ‘small, little’.

*CmsV > Hitt. /CntsV/: *h2msósio- > Hitt. /Hntsā́Sa‑/, a-an-za-a-aš-ša- ‘offspring’.

*Cms# > OH /°Cus/, °Cu(-ú)-us > NH /°Cos/, °Cu(-u)-uš ( ending).

*CmHsV > Hitt. /CnɨSV/: *h2mh1sénti > Hitt. /HnɨSántsi/, a-ne-iš-ša-an-zi ‘they wipe’; *tmh2sénti > Hitt. /tmɨSántsi/, da-me-iš-ša-an-zi ‘they press’ (with restored /m/ on the basis of strong stem /tmāS‑/).

*CmnC > Hitt. /CmɨnC/: *h2mnœhénti > Hitt. /Hmɨngántsi/, a-me-in-kán-zi ‘they betroth’; *tmnkénti > Hitt. /tmɨnkántsi/, ta-me-en-kán-zi ‘they attach’.

*CmnV > Hitt. /CMnV/: smnoé- > Hitt. /sMnāé‑/, ša-am(-ma)-na-a-e° ‘to create’; *smnénti > Hitt. /sMnántsi/, ša-am-na-an-zi ‘they pass by’;[161] *h1rmnié/ó- > Hitt. /ʔrMnió‑/, ar-ma-ni-a- ‘to become ill’; *h3nh3mnié/ó- > Hitt. /ləMnié/á‑/, lam(-ma)-ni-a- ‘to name’.

*CmiV > Hitt. /CmiV/, *septmió- > Hitt. /siptmiá‑/, ši-ip-ta-mi-a- ‘seven-drink’.

*VmV > OH /VmV/: *imié/ó- > OH /imié/á‑/, i-mi-e°, i-mi-a- (see § below for the conditioned fortition of OH /VmV/ > NH /VMV/).

*VmT (in which T = any stop) > Hitt. /VmT/: *tomp-u- > Hitt. /tampu‑/, dam-pu ‘blunt’.

*VmsV > Hitt. /VSV/: *h2éms-u- > Hitt. /HáSu‑/, a-aš-šu- ‘king’. That this assimilation took place very early is deducible from the fact that the outcome /S/ is affected by the lenition rules: *h2óms-ei > */HÖSē/ > Hitt. /Hā́si/, a-a-ši ‘she gives birth’.

*Vms# > Hitt. /Vs#/: *°Coms > OH /°Cus/, °Cu(-ú)-uš > NH /°Cos/, °Cu(-u)-uš ( ending).

*VmHsV > Hitt. /VnsV/: *h2ómh1sei > Hitt. /ʔā́nsi/, a-a-an-ši ‘he wipes’.

*VmHsC > Hitt. /VMɨsC/: *demh2sh2ó- > Hitt. /taMɨsHā́‑/, dam-me-eš-a-a- ‘damaging’.

*VmHs# > Hitt. /Vnɨs/: *h2ómh1s > Hitt. /Hā́nɨs/, a-a-ni-iš, a-a-ni-eš ‘wipe!’ (with restored ).

*VmnV > Hitt. /VMV/, unless when part of a paradigm[162]: *h2imno- > Hitt. /HiMa‑/, i-im-ma- ‘imitation’; *œhimnént- > Hitt. /kiMánt‑/, gi-im-ma-an-t° ‘winter’.

*Vmn# > Hitt. /Vmn/: *h1érmn > Hitt. /ʔérmn/, e-er-ma-an ‘illness’; *h3néh3mn > Hitt. /lā́mn/, la-a-ma-an ‘name’.

*VmrV > Hitt. /VMrV/: *œhimro- > Hitt. /kiMra‑/, gi-im(-ma)-ra- ‘field’.

*Vmr# > Hitt. /VMr/: *nómr > NH /láMr/, lam-mar ‘moment’.[163]  

Word-finally after vowel: *Vm# > Hitt. /Vn#/: *pérom > Hitt. /péran/, pé-e-ra-an; *h2érœ-i-m > Hitt. /Hárgin/, ar-ki-in ‘white’ (; *óm > Hitt. /kón/, ku-u-un ‘this (one)’ (; etc.

Word-finally after consonant: *°Cm > OH /°Cun/[164] > NH /°Con/[165]: *h1épm > OH /ʔépun/, e-ep-pu-un > NH /ʔépon/, e-ep-pu-u-un ‘I seized’.   Fortition

From MH times onwards, we encounter fortition of intervocalic /m/ to /M/: a-mi-a‑an-t- (MS) > am-mi-a-an-t- (NS) ‘small’; i-mi-a- (OH/MS) > im-mi-a- (MH/MS, NS) ‘to mingle’; i-mi-ú-l=a-aš (MS) > im-mi-ú-ul (NS) ‘grain mix’; kar-di-mi-a- (OS) > kar-tim-mi-a- (MS, NS) ‘to be angry’; la-a-ma-an (MS) > la‑a‑am-ma‑a(n)=m-mi-it (NS) ‘name’; mu-mi-a- (MS) > mu-um-mi-a- (NS) ‘to fall’; ‑nu-me-e-ni (OS, MS) > ‑nu-um-me-ni (NS) 1pl.pres.act. ending of causatives in ‑nu-; pé-e-tu-me-e-ni (OS) > pé-e-du-um-me-e-ni (NS), pé-e-tum-me-e-ni (NS) ‘we transport’; šu-me-eš (OS) > šu-um-me-eš (NH) ‘you’; šu-ma-an-za-n° (OS, MS) > šum-ma-an-za-n° (NS) ‘bulrush’; etc.

Because in none of these words the gemination has been carried out consistently, it is not fully clear whether we are dealing with a phonetic development or not. Prof. Melchert informs me that due to the loss of mimation in Akkadian (i.e. signs of the shape CVm could now also be used in words where no -m- is present), these cases of gemination of -m- can be accounted for as a graphical phenomenon. Although this indeed would work for e.g. kar-tim-mi-a-, pé-e-tum-me-e-ni and šum-ma-an-za-n°, I do not see how this practice can explain im-mi-ú-ul, mu-um-mi-a- or šu-um-me-eš, where an extra sign with the value Vm is used. In view of the virtual absence of gemination in cases like mēma/i- ‘to speak’ (although some occasional spellings with me-em-m° do occur in NS texts), we could perhaps assume that at the end of the NH period the beginning of a phonetic development can be observed through which an intervocalic -m- following an unaccented vowel is being geminated (compare also the occasional NH fortition of intervocalic -n- in §     *n

Word-initially before vowels, *n is retained as /n/: *nébhes > Hitt. /nébis/, ne-e-pí-iš ‘heaven’; néuo- > Hitt. /néua‑/, ne-e-a- ‘new’; *nóh2ei > Hitt. /nā́hi/, na-a-i ‘he fears’; *nukwe > Hitt. /nukw/, nu-uk-ku ‘and now’; etc.

Word-initially before consonants, *nC- regularly yields Hitt. /nC‑/, phonetically realized as [ənC‑]: *nsós > Hitt. /ntsā́s/, an-za-a-aš ‘us’. If amiant- ‘small’ really reflects *n- + miant‑, it would show /əmiánt‑/ < */ənmiánt‑/ (cf. e.g. ku-e-mi < */kwénmi/ for loss of ‑n- in *VnmV).

In cases where *nC- is the zero grade of a root *neC‑, the regular outcome /nC‑/ = [ənC‑] has been secondarily altered to /nəC‑/ in analogy to the full grade: *nh2ént- > Hitt. /nəHánt‑/, na-a-a-an-t° ‘fearing’ (on the basis of *nóh2ei > na-a-i).  

For word-internal position, it is best to treat the different environments separately.

*CnV > Hitt. /CnV/ : *h2nénti > Hitt. /Hnántsi/, a-na-an-zi ‘they draw water’; smnoé- > Hitt. /sMnāé‑/, ša-am(-ma)-na-a-e° ‘to create’; *h2rœnou- > Hitt. /Hrgnau‑/, ar-ga-na-u- ‘palm, sole’; etc.

*CnT (in which T = any stop) > Hitt. /CnT/: *h1ndo > Hitt. /ʔnda/, an-da ‘inwards’; *dhbhnœh-u- > Hitt. /pngu‑/, pa-an-ku- ‘entire, complete’ (if not reflecting *dhbhonœh‑u‑).

*CnsV > Hitt. /CntsV/: *nsós > Hitt. /ntsā́s/, an-za-a-aš ‘us’.

*CnHsV > Hitt. /CnɨSV/: *œnh3sénti > Hitt. /knɨSántsi/, ka-ni-eš-ša-an-zi ‘they recognize’.

*CnHsC > Hitt. /CəSɨC/: *h3nh3sé/ó- > Hitt. /HəSɨké/á‑/, a-aš-ši-ke/a- ‘to sue (impf.)’.

*CnmV: this sequence may be attested in ša-a-a‑[m]e-et (KBo 6.2 + KBo 19.1 ii 24) // ša-a-a-mi-it (KBo 6.3 ii 38) ‘my šaŸŸan’ < *séh2n=met. If correct, these forms must be phonologically interpreted as /saHəmet/ parallel to a-mi-a‑an-t- = /əmiant‑/ < */ənmiant‑/ < *n-mi¤ant-.

*CnnC > Hitt. /CnɨnC/: *h3rnngénti > Hitt. /Hrnɨngántsi/, ar-ni-in-kán-zi ‘they make disappear’.

*CniV > Hitt. /CniV/: *h3nié/ó- > Hitt. /ʔnié/á‑/, a-ni-e°, a-ni-a- ‘to work’.

*ClnC > Hitt. /ClɨnC/?: *h1lnœhénti > Hitt. /lɨngántsi/, li-in-kán-zi ‘they swear’ (if the preform is correct).

*CmnC > Hitt. /CmɨnC/: *h2mnœhénti > Hitt. /Hmɨngántsi/, a-me-in-kán-zi ‘they betroth’; *tmnkénti > Hitt. /tmɨnkántsi/, ta-me-en-kán-zi ‘they attach’.  

*VnV > Hitt. /VnV/: *h3ér-on-os > Hitt. /Hā́ranas/, a-a-ra-na-aš ‘eagle (’; *h2ónei > Hitt. /Hā́ni/, a-a-ni ‘he draws water’; *œénu- > Hitt. /kénu‑/, ge‑e‑nu- ‘knee’; *h1esh2enós > Hitt. /ʔisHanā́s/, iš-a-na-a-aš ‘blood (’.

*VnP (in which P = any labial stop): no examples.

*VnT (in which T = any dental stop) > Hitt. /VnT/: *h1sénti > Hitt. /ʔsántsi/, a-ša-an-zi ‘they are’; *gwhént(o) > Hitt. /kwénta/, ku-e-en-ta ‘he killed’; *spóndei > Hitt. /ɨspā́ndi/, iš-pa-a-an-ti ‘he libates’; *spóndh2ei > Hitt. /ɨspā́ndHe/, iš-pa-an-ta-é ‘I libate’.

*VnKV (in which K = any velar stop) > Hitt. /VnKV/: *ónkei > Hitt. /kā́nki/, ka‑a‑an-ki ‘he hangs’; *h1h-ēr > Hitt. /línker/, li-in-ke-er ‘they swore’

*VnKC (in which K = any velar): If the preceding vowel is /ā/, then */ānKC/ > Hitt. /ānKC/: *ónkh2ei > Hitt. /kā́nkHe/, ga-a-an-ga-a-é ‘I hang’. If the preceding vowel is not /ā/, then */VnKC/ > Hitt. /VKC/: *h1lénœhti > Hitt. /lígtsi/, li-ik-zi ‘he swears’; *srnénkmi > Hitt. /srníkmi/, šar-ni-ik-mi ‘I compensate’.

*VnHV > Hitt. /VNV/: *tinh1énti > Hitt. /tsiNántsi/, zi-in-na-an-zi ‘they finish’; *munh2énti > Hitt. /muNántsi/, mu-un-na-an-zi ‘they hide’; sunh1/3énti > Hitt. /suNántsi/, šu-un-na-an-zi ‘they fill’.

*Vns > Hitt. /VS/[166]: de/ons-u- (or *de/oms-u‑) > Hitt. /taSu‑/, da-aš-šu- ‘powerful’; *uéns-ti > Hitt. /kuáStsi/, ku-a-aš-zi ‘he kisses’; *h3érōns > Hitt. /Hā́ras/, a‑a-ra-aš ‘eagle’; *uÖns > Hitt. /kuā́s/, ku-a-aš ‘dog’; *kwtruËns > Hitt. /kwtruás/, ku-ut-ru-a-aš ‘witness’; *‑uén-s > Hitt. /‑uás/, a-aš, of the verbal noun in ar.[167]

*VnHsV > Hitt. /VntsV/: *œénh1-su- > Hitt. /kéntsu‑/, ge-en-zu- ‘lap’.

*VnnV > Hitt. /VNV/: *un-no- > Hitt. /koNa‑/, ku-u-un-na- ‘right, favourable’.[168]

*VnmV > Hitt. /VMV/: see Kimball (1999: 324) for examples like ma-a-am-ma-an < *mān-man, ad-da-am-ma-an < *attan=man ‘my father’, tu-ek-kam-ma-an < *tuekkan=man ‘my body’, etc.[169]

*VnuV > Hitt. /VuV/: *gwhén-uen > Hitt. /kwéuen/, ku-e-u-en ‘we killed’; mān + =a > Hitt. ma-a-a, ma-a-u-a.[170]  

*VPnV (in which P = any labial stop) > Hitt. /VPnV/: *h3epnos > Hitt. /Hapnas/, a‑ap-pa-na-aš ‘baking kiln (’; *dhebh-n(e)u- > Hitt. /tebnu‑/, te-ep-nu- ‘to diminish’.

*VtnV > Hitt. /VNV/: ‑ótno > Hitt. /‑ā́Na/, °Ca-a-an-na (inf.II ending); *‑otnos > Hitt. /‑ā́Nas/, °Ca-a-an-na-aš ( of abstracts in ‑ātar / ‑ānn‑).

*Vd(h)nV > Hitt. /VdnV/: *h2uidnos > Hitt. /Huidnas/, u-it-na-aš ‘game, wild animals’ (; *(h3)ud-nei- > Hitt. /(ʔ)udné‑/, ut-ne-e- ‘land’.

*VKnV (in which K = any velar stop) > Hitt. /VKnV/: *dlughnéuti > Hitt. /tslugnÜts/, za-lu-uk-nu-za ‘he delays’.

*ÕHnV > Hitt. /ÐnV/: *tiéh1no- > Hitt. /tséna‑/, ze-e-na- ‘autumn’.

*VHnÕ > Hitt. /VNÕ/: *h3eh3nóh3- > Hitt. /HaNá‑/, a-an-na- ‘to sue’.

*VsnV > Hitt. /VSnV/: e.g. *usné/ó- > Hitt. /uSnié/á‑/, uš-ni-a‑, uš-ša-ni-a- ‘to put up for sale’.

*VmnV > Hitt. /VMV/: *h2imno- > Hitt. /HiMa‑/, i-im-ma- ‘imitation’.

*VrnV > Hitt. /VrnV/: *h3ernou- > Hitt. /Harnau‑/, ar-na-u- ‘birthing chair’.

*VlnV > Hitt. /VLV/: *uélnu- > Hitt. /uéLu‑/, ú-e-el-lu- ‘pasure’ (if this etymology is correct).  

Word-finally, *n is retained as such, so *‑Vn > Hitt. /‑Vn/ and *‑Cn > Hitt. /‑Cn/.

Examples: *h2ón > Hitt. /Hā́n/, a-a-an ‘draw water!’; *h1érmn > Hitt. /ʔérmn/, e‑er-ma-an ‘illness’; *h3néh3mn > Hitt. /lā́mn/, la-a-ma-an ‘name’; *séh2n > Hitt. /sáHn/, ša-a-a-an ‘feudal service’.   Fortition

Fortition of OH intervocalic /n/ to NH /N/ seems to have taken place in the following examples: a-ap-pa-na-an-da (OS) > a-ap-pa-an-na-an-da (NS) ‘behind’; a-ra-a-ni- (OS) > a-ra-a-an-ni- (NS) ‘free’; i-na-ra- (OS) > in‑na‑ra- (MS, NS) ‘vigour’. If we compare cases like ini ‘this (’, genu- ‘knee’, šīuna- ‘god’ and zēna- ‘autumn’, in which intervocalic /n/ remains throughout the Hittite period, it seems that fortition only took place when /n/ did not follow the accented vowel.    Dissimilation

In the words lāman /lā́mn/ ‘name’ < *h3néh3mn, lammar /láMr/ ‘moment’ < *nómr and armalie/a-tta(ri) /ʔrmlié/á-/ ‘to be ill’ < *h1rmn-ié/ó- (besides armanie/a-tta(ri)) we seem to be dealing with dissimilation of *n to /l/ due to the nasal consonant *m in the same word.

1.4.8 Semi-vowels     *i

Word-initially before vowels *i is retained, except before *e: *iugom > Hitt. /iugan/, i-ú-ga-an ‘yoke’ vs. *ieg-o- > Hitt. /éga‑/,[171] e-ka- ‘ice’; *iéu-on-[172] > Hitt. /éuan‑/,[173] e-a-n°, a kind of grain.[174] If Hitt. i-ú-uk ‘yoke’ represents /iÜg/ < *iéug, it would show that loss of *i in front of *e must postdate the monophthongization of *eu to /ū/ here.

Word-initially before consonant *i is retained as such: *imié/ó- > Hitt. /imié/á‑/, i‑mi-e°, i-mi-a- ‘to mingle’.  

Interconsonantally, *i is in principle retained (but see below at ‘assibilation’): *h1ité+n > Hitt. /ʔitén/, i-it-te-en ‘go!’; *h2imno- > Hitt. /HiMa‑/, i-im-ma- ‘imitation’; *h2érœis > Hitt. /Hárgis/, ar-ki-iš ‘white’; *œhim-n-ént- > Hitt. /kiMánt‑/, gi-im-ma-an-t- ‘winter’; *kwis > Hitt. /kwis/, ku-iš ‘who’.

In the sequence *CiV, *i in principle is retained as well (but see below at ‘assibilation’): *h3nié/ó- > Hitt. /ʔnié/á‑/, a-ni-e°, a-ni-a- ‘to work’, *h3rgié/ó- > Hitt. /Hrgié/á‑/, ar-ki-e°, ar-ki-a- ‘to get lost’, *rsié/ó- > Hitt. /krSié/á‑/, kar‑aš‑ši-i-e° ‘to cut’; etc.

Note that *VsiV yields Hitt. /VSV/, however, as is seen in e.g. *iugosio- > Hitt. /iugaSa‑/, i-ú-ga-aš-ša- ‘yearling’; *h2msósio- > Hitt. /Hntsā́Sa‑/,[175] a-an-za‑a‑aš‑ša- ‘offspring’; *usié/ó- >> */u̯sié/á-/[176] > */uəsié/á-/ > OH /uəSé/á‑/, a-aš-še/a-[177] ‘to put on clothes’.

For the sequence *ViC, cf. the treatment of the diphthongs *ei, *ēi, *oi and *ōi below.

Intervocallicaly, *i is dropped without a trace, e.g *CéC-ei-os > pre-Hitt. */CéCaias/ > Hitt. /CéCas/ ( ending of i-stem adjectives).[178] When the *i is surrounded by non-identical vowels (e.g. *‑aie‑, *‑aii‑, *‑aiu‑), the loss of *i causes compensatory lengthening of the preceding vowel:[179] *CÕC-ei-i > pre-Hitt. /CÕCaii/ > Hitt. /CÕCāi/, °Ca-a-i ( ending of i-stem adjectives); *CÕC-ei-ēs > pre-Hitt. */CÕCaies/ > Hitt. /CÕCāes/, °Ca-a-eš ( ending of i-stem adjectives); *CÕC-ei-ms > pre-Hitt. */CÕCaius/ > Hitt. /CÕCāus/, °Ca-a-uš ( ending of the i-stem adjectives); *°Co-ié-mi > pre-Hitt. /°Caiémi/ > Hitt. /°Cāémi/, °Ca-a-e-mi (1sg.pres.act. ending of the atrae-class).[180]

An intervocalic cluster *VHiV yields OH /ViV/ with lengthening of the preceding vowel (if possible).[181] The newly created intervocalic /i/ is again lost in NH times: *h2uh1iénti > OH /Hoiántsi/, u-a-an-zi > NH /Hoántsi/, u-a-an-zi ‘they run’; *teh2iéti > OH /tāiétsi/, ta-a-i-ez-zi, da-i-e-ez-zi ‘he steals’.[182]

In words with the sequence *ViHV, we first find monphthongization of the diphthong to /e/ (see also below at the treatment of the diphthongs *ei and *oi): *méih2ur > OH /méhur/, me-e-ur ‘period, time’; *néih1/3-o > OH /néʔa/, ne-e-a ‘he turns’; *h2éih3um > OH /Héʔun/, é-e-un ‘rain (’; *h2éih3-eu-ēs > OH /Héʔaues/, é-e-a-u-e-eš ‘rains (’. Note that in the latter three examples intervocalic /ʔ/ is lost only in the late OH period, as we can see from spellings like OS é-e-a-u-e-š=a = /Héaues/, phonetically realized as [Héaues] and MH/MS ne-e-a-ri = /néari/, phonetically realized as [néari]  

Forms like i-it ‘go!’ < *h1idhi and te-e-et ‘speak’ < *dhéh1dhi seem to show regular loss of word-final *i. This means that the synchronic word-final ‑i as attested in the ending and the verbal present endings (mi, ‑ši, ‑zi, etc.) must be the result of a wide-scale restoration.[183]

If ke-e ‘these’ indeed reflects *ih2, as will be suggested s.v. kā- / ki- / kū‑, then it shows that in word-final *Cih2#, *i is lowered to /e/ due to the following *h2 (just as *°Cuh2 yields Hitt. /°Co/, cf. a-aš-šu-u /ʔáSo/ < *°s-u-h2). This implies that zi-i-ik ‘you’ < *tíH+œ reflects *tíh1.   Assibilation

As is well-known, *i causes assibilation of preceding dental consonants. We must distinguish between two separate instances of assibilation, however. On the basis of an etymological connection between Hitt. šarāzzi(a)- ‘upper’ and Lyc. hrzze/i- ‘upper’, we must assume that already in PAnatolian *t is assibilated to */ts/ in a cluster *VtiV (cf. s.v. ‑(e)zzi(a)‑). Note that here the *i is retained (although it eventually is lost in Lyc., where no synchronic trace of *i can be found). The assibilation of dentals in the sequences *TiV-, *CTiV and *-Ti is a specific Hittite development. Note that here in principle *i is lost: *tiéh2-oi- > Hitt. /tsáHai‑/, za-a-a-i- ‘battle’; *tióh2-ei > Hitt. /tsā́hi/, za-a-i ‘he hits’; *tiéh1-o > OH /tséʔa/, ze-e-a ‘cooks’; *h2t-ié/ó- > Hitt. /Htsié/á‑/, a-az-zi-e°, a-az-zi-a- ‘to pierce’;[184] *h2énti > Hitt. /Hánts/, a-an-za ‘in front’; *h1ésti > OH /ʔésts/, e-eš-za ‘he is’[185]; *diËus > Hitt. /sÏus/, ši-i-ú-uš ‘god’; *diéu-ot- > Hitt. /sÏuat‑/, ši-i-a-at-t- ‘day’. This means that we must reckon with a development */ti/ = *[t:i] > *[t:j] > *[tç] > *[tS] > [ts] = /ts/ and */di/ = *[ti] > *[tj] > *[ç] > *[S] > [s] = /s/. We therefore have to assume that in forms like e-ep-zi ‘he seizes’ < *h1épti, ap-pa-an-zi ‘they seize’ < *h1pénti, a‑az‑zi-e°, a-az-zi-a- ‘to pierce’ < *h2tié/ó- the ‑i- has been restored. In the first two cases in analogy to the other present endings in ‑i (mi, ‑ši, etc.), in the latter in analogy to other verbs with the e/a-suffix.[186]

Note that the verbs zinu-zi ‘to make cross’ < *h1t-i-neu- and zinni-zi / zinn- ‘to finish’ < *ti-ne-h1- / *ti-n-h1- may show that despite the fact that in *TiC *i caused assibilation of the dental consonant, it was retained as a vocalic element.[187]

On the basis of the equation between Hitt. dalugašti- ‘length’ and Pol. długość ‘length’ < *d(o)l(u)gh-osti‑, Joseph (1984: 3-4) argues that in a cluster *‑sti- assibilation did not take place, which would have a parallel in Greek, cf. εἶσι ‘he goes’ < *h1éiti vs. ἐστί ‘he is’ < *h1ésti.[188] If taišzi- ‘hay-barn’ indeed reflects *dhoh1-es-ti-, as argued s.v., however, then the postulation of non-assibilation in *‑sti- is incorrect.   Lowering

In some words we encounter lowering of OH /i/ to NH /e/:

apiniššan ‘thus’ > apeneššan: a-pí-ni-iš-š° (OS) > a-pé-e-ni-eš-š° (NS).

alzišša-i / alzišš- ‘to call (impf.)’ > alzešš-: al-zi-iš-š° (OS, MS) > al-ze-eš-š° (NS).

išša- ‘carriage pole’ > ešša-: i-iš-š° (OH/NS) > e-eš-š° (NS, 1x).

ištā, ištī ‘mausoleum(?)’ > eštā, eštī: i-iš-t° (OS, MS) > é-eš-t° (MS, NS).

ini ‘this’ > eni: i-ni (MS) > e-ni (MS, NS)

iniššan ‘thus’ > eniššan: i-ni- (MS) > e-ni- (NS).

īšša-i / īšš- ‘to do, to make (impf.)’ > ešša-i / ešš-: i-iš-š° (OS) > iš-š° (MS) > (e‑)eš‑š° (NS).

ā- ‘master’ > - : iš-° (OS, MS, NS) > (e‑)eš-° (NS).

an- ‘blood (obl.)’ > ēšan-: iš-a-n° (OS, MS) > e-eš-a-n° (MS, NS).[189]

iš(ša)nā- ‘dough’ > eššana-: iš-(ša‑)n° (OS, MS, NS) > (e‑)eš-ša-n° (NS).

kīnu-zi ‘to open (up)’ > kenu-: ki-i-nu- (MS) > ke-e-nu- (NS).

lilua-i / lilui- ‘to pour’ > lelua-: li-il-° (MS) > le-el-° (MS, NS).

li(n)k-zi ‘to swear’ > lenk-: li-in-k° (OS, MS) > le-en-k° (NS).

mimma-i / mimm- ‘to refuse’ > memm-: mi-im-m° (OS, MS) > mi-im-m° (NS).

mišriant- ‘perfect’ > mešriant-: mi-iš-ri- (MS) > me-eš-ri- (NS), me-iš-ri- (NS).

šīna- ‘figurine, doll’ > šena-: ši-i-n° (OS, MS) > še-(e‑)na- (NS).

šiš-zi ‘to proliferate’ > šeš-: ši-iš- (OS, MS) > še-iš‑, ši-eš‑, še-eš- (NS).

šišša-i / šišš- ‘to impress’ > šešš-: ši-iš-š° (MS, NS) > še-eš-š° (NS).

šiša-i / šiš- ‘to decide’ > šeš-: ši-iš-° (MS) > še-eš-° (NS).

šiššur- ‘irrigation’ > šeššur-: ši-iš-š° (MS) > še-eš-š° (NS).

tit-a ‘to thunder’ > tet-: ti-it-° (OS, MS) > te-e-et-° (OH/MS) > te-et-° (NS).

arrišša-i / arrišš- ‘to help’ > arrešš-: a-ar-ri-iš-š° (NS) > a-ar-re-eš-š° (NS).

zinni-zi / zinn- ‘to finish’ > zenn-: zi-in-n° (OS, MS) > ze-en-n° (NS).

Melchert (1984a: 154) explains these forms as showing “a simple assimilation: i is lowered to e before a low vowel a in the next syllable”. As he notices himself, there are a number of words that contradict this formulation, however, e.g. idālu- and iar. Moreover, this rule cannot account for the lowering in e.g. ini, ištī, kīnu-zi and šiš-zi.[190]

In my view, lowering in these words is determined by the intervening consonants: /i/ > /e/ in front of s, n, m and clusters involving /H/ (l- and ‑t).[191] It cannot be coincidental that exactly these consonants[192] also cause a preceding /u/ to get lowered to /o/ (cf. § Note that lowering of /i/ to /e/ in front of s and n also explains the high number of NS spellings of the and forms of i-stem nouns and adjectives with the vowel ‑e‑.     *u

Word-initially before vowels *u is retained as such: *uódr > Hitt. /uā́dr/, a-a-tar ‘water’; *ti > Hitt. /uéktsi/, ú-e-ek-zi ‘he wishes’.: *uei(e)s > Hitt. /ués/, ú-e-eš ‘we’; etc. Word-initially before consonants *u is retained as well: *urh1ór(i) > OH /urʔā́ni/, ú-ra-a-ni > MH/NH /uərʔā́ni/, a-ra-a-ni ‘it burns’;[193] *usnié/ó- > Hitt. /uSnié/á‑/, uš-(ša‑)ni-e°, uš-(ša‑)ni-a- ‘to put up for sale’.  

Interconsonantally, *u yields either /u/ or /o/: see § for an elaborate treatment.

In the sequence *CuV, *u in principle yields /u/ (see § for the fact that in Hittite there is no reason to assume a phonemic difference between /CuV/ and /CoV/). Note that in *Tuo (where T = any dental stop or *n), *u is lost: *duoiom > Hitt. /tā́n/, ta-a-an ‘for the second time’; *h1eduól-u- > Hitt. /ʔidā́lu‑/, i-da-a-lu- ‘evil’; *nu-os > Hitt. /nas/, na-aš ‘and he’.[194] Remarkable is the outcome of *CHuV, which yields Hitt. /ComV/[195]: *dh3uéni > Hitt. /toméni/, tu-me-e-ni ‘we take’; *bhdhh2-uén-ti > Hitt. /ptomántsi/, pát-tu-u-ma-an-zi ‘to dig’; *su-n-H-¾μ > Hitt. /suNomər/, šu-un-nu-mar ‘filling’.

For the development of *CVuC, see the treatment of the diphthongs *eu, *ēu, *ou and *ōu below.

Intervocalically, *u is retained as /u/ (see § for the demonstration that intervocalically there is no phonologic difference between /u/ and /o/): *néuo- > Hitt. /néua‑/, ne-e-a- ‘new’; *dhébh-eu-ēs > Hitt. /tébaues/, te-e-pa-u-e-eš ‘little (’; *HH-ió-ueni > Hitt. /ʔiáuani/, i-a-u-a-ni ‘we do’; *h2rœ-nóu-i > Hitt. /Hrgnā́ui/, ar-ga-na-ú-i ‘palm (’.

A special development is the change of *uuV to /umV/ and of *VuuC to /VmuC/:[196] *CÕC-eu-ms > *CÕC-au-us > Hitt. /CÕCamus/, °Ca-mu-uš ( ending of u-stems); *CC-nu-uéni > Hitt. /CCnuméni/, °nu-me-e-ni (1sg.pres.act. ending of causatives in ‑nu‑); *CC-nu-uénti > Hitt. /CCnumántsi/, °nu-ma-an-zi (inf.I ending of causatives in ‑nu‑); *h2óu-ueni > Hitt. /ʔáumani/, a-ú-ma-ni[197] ‘we see’.  

In word-final position, *u is retained as such: *dhébhu > Hitt. /tébu/, te-e-pu ‘little (’; etc. Note that a-aš-šu-u /ʔáSo/, which contrasts with a-aš-šu /ʔáSu/, shows that word-final *‑Cuh2 yields Hitt. /‑Co/


1.4.9 Vowels     *e

There are a number of positions in which PIE *e gets coloured.

When adjacent to *h2, *e gest coloured to PAnat. /a/, which yields Hitt. /a/ in both accented and unaccented position: *h2énti > Hitt. /Hánts/, a-an-za ‘in front’; *h2épti > Hitt. /Háptsi/, a-ap-zi ‘he attaches’; *péh2ur > Hitt. /páHwr/, pa-a-ur, pa-a-u‑a-ar ‘fire’; *péh2so > Hitt. /páHsa/, pa-a-ša ‘he protects’; *léh2pt(o) > Hitt. /lā́pta/, la-a-ap-ta ‘he glowed’; *h1esh2enós > Hitt. /ʔisHanā́s/, iš-a-na-a-aš ‘blood (’; *peh2uénos > Hitt. /paHwénas/, pa-a-u-e-na-aš ‘fire (’.

When adjacent to *h3, *e gets coloured to PAnat. /o/ and ultimately merges with the reflex of PIE *o, yielding /ā́/ when accented and /a/ when unaccented: *h3érōn+s > Hitt. /Hā́ras/, a-a-ra-aš ‘eagle’; *h3épr > Hitt. /Hā́pr/, a-a-ap-pár ‘business’; *h3néh3mn > Hitt. /lā́mn/, la-a-ma-an ‘name’. Note that when accented this vowel does not lenite a following consonant, whereas PIE *ó > PAnat. /Ö/ does. This explains the difference between *‑ótr > PAnat. /‑Ödr/ > Hitt. /‑ā́dr/, °a-a-tar (abstract suffix) and *h3épr > PAnat. /Hópr/ > Hitt. /Hā́pr/, a-a-ap-pár ‘business’.

In the position *énT (in which T = any dental consonant), *e yields Hitt. /a/: *h1sénti > Hitt. /ʔsántsi/, a-ša-an-zi ‘they are’; *gwhnénts > Hitt. /kwnánts/, ku‑na‑an‑za ‘killed’; *h2uh1énts > Hitt. /Hoánts/, u-a-an-za ‘wind’; *srbhuéns > Hitt. /srɨbuás/, ša-ri-pu-a-aš ‘of drinking’.

In the position *eRCC (in which R = any resonant and C = any consonant), *e becomes Hitt. /a/: *bhérsti > Hitt. /párstsi/, pár-aš-zi ‘he flees’; *kérsti > Hitt. /kárstsi/, kar-aš-zi ‘he cuts’; *stélœhti > Hitt. /ɨstálgtsi/, iš-tal-ak-zi ‘he levels’; *h1érsti > Hitt. /ʔárstsi/, a-ar-aš-zi ‘he flows’; *bhérh2/3ti > Hitt. /párHtsi/, pár-a-zi, pár-a-zi ‘he pursues’; *térh2-u-ti > pre-Hitt. */térHwtsi/ > Hitt. /tárHwtsi/, tar-u-uz-zi, tar-u-zi ‘he conquers’. Note however that *œénh1su- yields Hitt. /kéntsu‑/, ge-en-zu- ‘lap, abdomen’, which shows that a sequence *éRh1CV develops into *eRCV before *eRCC > /aRCC/.[198]

In the position *éKsC, *e yields Hitt. /a/: *téksti > */tákstsi/ > Hitt. /tákɨstsi/, ták‑ki-iš-zi, ták-ke-eš-zi ‘he unifies’; *négwh-s-ti > Hitt. /°nágwstsi/, na-na-ku-uš-zi ‘it becomes dark’.

In the position *enK (in which K = any velar), *e yields Hitt. /i/: *h1lénœhti > */língtsi/ > Hitt. /lígtsi/, li-ik-zi ‘he swears’.

In the position *ueT (in which T = any dental consonant), *e yields Hitt. /i/: *uet- > Hitt. /uit‑/, ú-i-it-t° ‘year’; *uedo- > Hitt. /uida‑/, ú-i-da- ‘wet’; *sóuetest- > šāitišt- ‘weanling’.[199]

In some positions we seem to be dealing with i-umlaut, causing *e to become Hitt. /i/: *mélit- > Hitt. /mílit‑/, mi-li-it-t° ‘honey’; *tinéh1ti > Hitt. /tsinítsi/, zi-in-ni-iz-zi ‘he finishes’; *dhurnéh1ti > Hitt. /tuərnítsi/, du-a-ar-ni-iz-zi ‘he breaks’. The exact conditions are not fully clear to me.

In word-final position, *é yields Hitt. /i/ when accented: *Hu-sé > Hitt. /ʔuskí/, uš-ki-i ‘look!’; *h1gwhsé > Hitt. /ʔkwskí/, ak-ku-uš-ki-i ‘drink!’; *h1dsé > Hitt. /ʔdsɨkí/, az-zi-ik-ki-i ‘eat!’. Note that e.g. u-it-ti ‘draw!’ < *h2ueTH-ié shows that *°Cié# yields Hitt. /°Cí/, probably through */°Cií/.  

It has been claimed that a sequence *eRH yields Hitt. /aRH/,[200] but I disagree with this assumption. Alleged examples in favour of this development like *uélh3ti > Hitt. /uálHtsi/ ‘hits’ and *bhérh2/3ti > Hitt. /párHtsi/ ‘pursues’ rather show the development *eRCC > /aRCC/, whereas šalli- / šallai- ‘big’ (usually reconstructed *selh2-i‑) and tarra-tta(ri) ‘to be able’ (usually reconstructed *terh2‑o‑) may be interpreted otherwise (see their respective lemmata). On the basis of er- / ara- / ar- ‘boundary’ < *h1er-h2- / *h1r-eh2- / *h1r-h2- and šera- (an object to rinse feet with) < *sérh2/3-o- (?), I assume that *e in a sequence *eRHV does not get coloured in Hittite.  

If *e does not get coloured due to one of the positions mentioned above, it shows the following developments.

When accented, *é merges with *Ë, *éh1 and *éi and develops into Hitt. /é/. Note that this vowel is often spelled with a plene vowel in open syllable and in monosyllabic words, which indicates that in these positions it was rather long phonetically. Phonologically, there is no use to indicate length, however, since the reflexes of *é, *Ë, *éh1 and *éi have merged under the accent, and the original distinction in length has been given up. It should be noted that *é does not lenite a following consonant, whereas *Ë, *éh1 and *éi do. This means that the merger of *é with the reflexes of *Ë, *éh1 and *éi is a rather recent phenomenon, which postdates the lenition of intervocalic consonants due to a preceding long accented vowel.

Examples: *sésti > Hitt. /séstsi/, še-eš-zi, še-e-eš-zi ‘he sleeps’; *nébhes > Hitt. /nébis/ ne-e-pí-iš, ne-pí-iš ‘sky’; *pédom > Hitt. /pédan/, pé-e-da-an, pé-da-an ‘place’; *kwérti > Hitt. /kwértsi/, ku-e-er-zi, ku-er-zi ‘he cuts’; *œhésr > Hitt. /kéSr/, ke-eš-šar ‘hand’; *dhéh1mi > Hitt. /témi/, te-e-mi, te-mi ‘I say’.

When unaccented, *e weakens to /a/ in post-tonic open syllables[201] and to /i/[202] elsewhere (in pretonic open and closed syllables and in post-tonic closed syllables).[203] In word-final position, unaccented *e is dropped.[204]

Examples: *h1esh2enós > Hitt. /ʔishanā́s/, iš-a-na-a-aš ‘blood (’; *nébhes > /nébis/ ne-e-pí-iš, ne-pí-iš ‘sky’;[205] *pesén¬s > Hitt. /pisénus/, pí-še-e-nu-uš ‘men (’; *CéC-ueni > Hitt. CeCuani (1pl.pres.act. forms); *CéC-teni > Hitt. CeCtani (2pl.pres.act. forms); *=kwe > Hitt. /=kw/, V=k-ku ‘and’; *tokwe > Hitt. /takw/, ták-ku ‘if, when’.[206]     *ē

When accented, the development of *Ë does not differ from the development of *é: I have not been able to find a spelling difference between the reflexes of *Ë and *é that would indicate a phonetic and/or phonological difference.

Examples: *h2stËr+s > Hitt. /Hstérts/, a-aš-te-er-za ‘star’; *§Ër > Hitt. /kér/, ke-er ‘heart’, *sËr > Hitt. /sér/, še-e-er, še-er ‘above, on top’; *°CËr > Hitt. /°Cér/, °Ce‑e‑er (3pl.pret.act. ending), *ud-nËi > Hitt. /udné/, ut-ne-e ‘land’.

There is one case, in which *Ë yields a different reflex, however: PIE *diËus yields Hitt. /sÏus/, ši-i-ú-uš ‘god’. It is not clear to me what caused the raising of *ē > /ī/ here. The fact that the sequences *di- > Hitt. /s‑/ precedes *ē hardly can be decisive, cf. *tiéh1-no- > /tséna‑/, zé-e-na- ‘autumn’. Perhaps the raising is comparable to the one seen in *ueT- > Hitt. /uiT‑/.

On the basis of the assumption that Hitt. “šumanza” would mean ‘cord, binding’ and reflects *sh1u-mēn+s (~ Gr. ὑμήν ‘sinew’), it was generally believed that *‑ēn+s yielded Hitt. /‑ants/, ‑anza, whereas ‑ōn+s > /‑as/, ‑aš (as in *h3érōn+s > Hitt. āraš). Since “šumanza” now must be interpreted as belonging to (Ú)šumanzan- ‘bulrush’, which has nothing to do with Gr. ὑμήν, the development *‑ēn+s > ‑anza cannot be upheld. Instead, on the basis of *kwtruËn+s > Hitt. /kwtruás/, ku-ut-ru-a‑aš ‘witness’, we should rather conclude that *‑ēn+s merged with *‑ōn+s and yielded Hitt. /‑as/.  

When unaccented, *ē yielded Hitt. /e/, and therewith differs from the reflex of unaccented short *e: *h1ésēr > Hitt. /ʔéser/, e-še-er ‘they were’.

This indicates that the distinction between *ē and *ĕ was present up to a quite recent stage: only after the weakening of unaccented *ĕ to /a/ and /i/,[207] unaccented *ē developed into /e/.  

None of the alleged instances where the sequences *h2ē, *h3ē, *ēh2 or *ēh3 are thought to have yielded Hitt. e- or ‑e- (Eichner’s Non-Colouration Law) can in my view withstand scrutiny (see s.v. ai(n)k-tta(ri), NAekur, enkan-, ippara‑, Éištā, Éištī, kane/išš-zi, ur / un‑, pia‑, šēur / šēun- and e-zi / a- for alternative interpretations).[208]     *o

In the case of *o it is important to note that it shows different reflexes when part of a diphthong (*oiC and *ouC). The diphthongs will be treated below.

When accented, *ó yields OH /ā́/ in initial and word-final syllables (but not in internal syllables, see below). Since *ó causes lenition, the development to a long vowel must antedate the period of lenition of intervocalic consonants due to a preceding accented long vowel.

In OS texts, the reflex /ā́/ is almost consistently spelled with plene ‑a‑, in open as well as closed syllables. In MH and NH originals, we hardly find plene spellings in closed, non-final syllables anymore, which indicates that /ā́/ has been phonetically shortened in these syllables in the post-OH period, and merged with /á/.[209] So *óCCV > OH /ā́CCV/ > NH /áCCV/, whereas *óCV and *óC# yield OH/NH /ā́CV/ and /ā́C#/.

Examples for initial syllable: *stóph2ei > OH /ɨstā́pHe/, iš-ta-a-ap-é > NH /ɨstápHi/, iš-ta-ap-a-i ‘I plug up’; *stópei > OH/NH /ɨstā́bi/, iš-ta-a-pí ‘he plugs up’; *ónkh2ei > OH /kā́nkHe/, ga-a-an-ga-a-é > NH /kánkHi/, ga-an-ga-a-i, kán-ga-a-i ‘I hang’; *dóh3-h2ei > OH /tā́He/, da-a-a-é > NH /táHi/, da-a-i ‘I take’; *ós > OH/NH /kā́s/, ka-a-aš ‘this’; *dóru > OH/NH /tā́ru/, ta-a-ru ‘wood’.

Examples for word-final syllables: *dhœhmós > Hitt. /tgnā́s/, ták-na-a-aš ‘earth (’; *h1esh2enós > Hitt. /ʔisHanā́s/, iš-a-na-a-aš ‘blood (’.  

In internal syllables (non-initial and non-final), *ó yielded Hitt. /á/, however: *tr‑nó-h1-h2ei > OHitt. /trnáHe/, tar-na-a-é ‘I release’; *mi-móh1-ei > Hitt. /miMái/, mi-im-ma-i ‘he refuses’.[210] This explains the difference between *dóh3-h2ei > OH /tā́He/, da-a-a-é ‘I take’ and *h1poi-dóh3-h2ei > OH /petáHe/, pé-e-ta-a-é ‘I bring’, etc.  

A special development of *ó is found in the following positions: *Cóm# > Hitt. /Cón/, Cu-u-un and *Cóms# > Hitt. /Cós/, Cu-u-uš.

Examples: *óm > Hitt. /kón/, ku-u-un ‘this one’ (; *h1o-bhóm > Hitt. /ʔabón/, a-pu-u-un ‘that one’ (; *óms > Hitt. /kós/, ku-u-uš ‘these ones’ (; *h1o-bhóms > Hitt. /ʔabós/, a-pu-u-uš ‘those ones’ (  

When unaccented, *o usually yields Hitt. /a/: *pédom > Hitt. /pédan/, pé-e-da-an, pé-da-an ‘place’, *pérom > Hitt. /péran/, pé-e-ra-an, pé-ra-an ‘before’, *h1éh1s-o > Hitt. /ʔésa/, e-ša ‘he sits down’.

A special development of *o is found in the ending *°Coms > OH /°Cus/ > NH /°Cos/, cf. §     *ō

The reflex of *ō seems to have merged with the reflex of *ŏ. When accented, *Ö yields Hitt. /ā́/: *udÖr > Hitt. /uɨdā́r/, ú-i-ta-a-ar, ú-e-da-a-ar ‘water’. Note that *Ön+s yields Hitt. /‑ás/, *uÖn+s > Hitt. /kuás/, ku-a-aš ‘dog(man)’.[211] When unaccented, *ō yields Hitt. /a/: *h3érōn+s > Hitt. /Hā́ras/, a-a-ra-aš ‘eagle’; *h1ndhuéh2ōs > Hitt. /ʔnduáHas/, an-tu-a-a-a-aš ‘human being’.


1.4.10 Diphthongs   *ei

When accented, *éi merges with the reflexes of *é, *éh1 and *Ë and yields Hitt. /é/ (but note that *éi lenites a following intervocalic consonant, whereas *é does not): *h1éi-h2ou > Hitt. /ʔéhu/, e-u ‘come!’; *h2éih3-u-s > Hitt. /Héus/, é-e-ú-uš ‘rain’;

*néih1/3-o > OH /néʔa/, ne-e-a > MH /néa/, ne-e-a ‘he turns’; *h2ou + *h1éiti > Hitt. /ʔuétsi/, ú-e-ez-zi ‘he comes’; *méih2ur > Hitt. /méhur/, me-e-ur ‘period’.

When unaccented, *ei yields Hitt. /e/: *órs-ei > OH /uā́rSe/, a-ar-aš-še ‘he harvests’; *dóh3-h2ei > OH /tā́He/, da-a-a-é. Note that as we can see from the examples, the reflex of *Cei is identical with the reflex of *h2ei (through pre-Hitt. */Hai/).

In the sequence *Kei- (in which K = any velar) the diphthong *ei was first raised to *‑ii‑, which yielded Hitt. /i/ in closed syllable and /ī/ in open syllable: *éito > Hitt. /kíta/, ki-it-ta ‘he lies’ (note that /t/ = [t:] counts as a geminate that closes the syllable); *œéis-h2o > Hitt. /kísHa/, ki-iš-a ‘I become’; *œéis-o > Hitt. /kÏsa/, ki-i-ša ‘he becomes’ (see s.v. ki-tta(ri) and kīš-a(ri) / kiš- for an elaborate treatment of this development). A similar raising may be found in ša-li-i-ga ‘he touches’ if this represents /slÏga/ < *sléiœ-o.   *ēi

The only secure example of *ēi that I know of is *ud-nËi > Hitt. /udné/, ut-ne-e ‘land’.   *oi

The diphthong *oi shows two reflexes. When preceding a dental consonant, *oi yields /ai/. It should be noted that when accented, *ói does not yield /ā́i/ in this environment, as one could expect on the basis of *ó > /ā́/, but rather /ái/, with a short ‑a‑. Before all other consonants and in absolute auslaut,[212] *oi becomes monophthongized to /e/. Note that in the sequence *óiV, we find the normal reflex of *ó, namely pre-Hitt. */ā́iV/ > Hitt. /ā́V/.

These developments explain the following paradigm:

    PIE                                                                      OH

*dhh1-ói-h2ei                                       >     /téHe/   te-e-e-é     >>    /téHi/  te-e-e-i

*dhh1-ói-th2ei                  > */táite/      >>   /táiti/     da-it-ti

*dhh1-ói-ei       > */tā́ie/   > */tā́e/        >>   /tā́i/      da-a-i


Other examples: *œróits > Hitt. /kráits/, ka-ra-i-iz ‘flood’; *dhh1-ói-s > Hitt. /táis/, da-iš ‘he placed’; *óinos > Hitt. /káinas/, ka-i-na-aš, ga-e-na-aš ‘in-law’; *ói > Hitt. /ké/, ke-e ‘these’ (   *ōi

To my knowledge, this diphthong only occurs in the diphthong-stems and yields /āi/: *tlh2Öi > Hitt. /tslHā́i/, za-al-a-a-i, a vessel; *h2urtÖis > Hitt. /Hurtā́is/, ur-da-a-iš, ur-ta-iš ‘curse’; *h2urtÖim > Hitt. /Hurtā́in/, ur-da-a-in, u-ur-ta-in ‘curse (’.   *eu

The diphthong *eu (i.e. *CeuC) is monophthongized to /u/ or /o/, depending on the surrounding sounds. For an elaborate treatment, cf. § Note that in *euV, we find the normal development of *e, e.g. *néuo- > Hitt. /néua‑/, ne-e-a- ‘new’, *dhébh-eu-os > Hitt. /tébauas/, te-pa-u-a-aš ‘little, few’ (   *ēu

The only possible instance of *ēu that I know of is *œhrh1-Ëu > Hitt. /krʔÜ/, ka-ru-ú ‘early’ if this reconstruction is correct.[213]   *ou

The diphthong *ou yields Hitt. /au/ (with short a) before dental consonants (including *r): *h2óuth2ei > Hitt. /ʔáuti/, a-ut-ti ‘you (sg.) see’; *h2óusten > Hitt. /ʔáusten/, a-uš-te-en ‘you (pl.) must see’; *h2óuri- > Hitt. /ʔáuri‑/, a-ú-ri- ‘look-out’. In other positions *ou is monophthongized to /u/ or /o/, depending on the surrounding sounds. See § for an elaborate treatment.   *ōu

To my knowledge, the diphthong *ōu only occurs in diphthong-stems like *h3ér-nōu > Hitt. /Hárnāu/, ar-na-a-ú ‘birthing chair’, *h2érœ-nōu > Hitt. /Hárgnāu/, ar-ga-na-ú ‘palm, sole’, etc., where it yields /āu/.

[1] Note that I do not reconstruct a PIE phoneme “a” or “ā”: all PIE forms for which some scholars reconstruct *a or *ā should be interpreted otherwise. For an extensive treatment of most of these words, cf. Lubotsky 1989. Eichner 1988: 132-3 adduces a few more forms that in his view must contain PIE *a or *ā, but these are incorrect as well. (1) “*ns-” ‘nose’ must reflect *neh2-s‑, *nh2-es‑, *nh2-s- (cf. Kortlandt 1985a: 119). (2) “*krh-” ‘to proclame’ is based on Skt. kārú- ‘singer’ and kīrtí- ‘fame’. The former may reflect *keh2rú-, the latter *krh2-tí- with metathesis from *kh2r-tí- (Schrijver 1991: 4). (3) “*haœ-” ‘to praise’ is based on Gr. ἁγνός, ἅγιος ‘holy’ besides Skt. yajñá- ‘sacrifice’. The former two words reflect *ih2œ- (cf. Beekes 1988c: 24-5) and the latter *ieh2œ-nó- (with loss of laryngeal before media + consonant, cf. Lubotsky 1981: 135). (4) “*h2ap-” ‘to harm’ is based on “heth. huapzi ‘schädigt’ (mit grundstufigem ă)” besides PGerm. *ubilaz ‘evil’. The cited form, uapzi, is the NH secondary replacement of an original i-conjugated form uappi. Because all i-verbs reflect *o/Ø-ablaut, the Hitt. stem uapp- must reflect *h2uoph1- with o-grade (s.v. uapp-i / upp-). (5) “kas-” ‘to kiss’ is based on “heth. kuašzi ‘küßt’ mit grundstufigem ă”. As I show under its lemma, the Hittite verb in fact is kuašš-zi with geminate ‑šš-, which cannot be explained by a reconstruction *kuaš-. I therefore reconstruct *kuens-, which would explain the vowel ‑a- as well as the geminate ‑šš- by regular sound laws. Note that the nasal is visible in Gr. κυνέω ‘to kiss’ as well.

[2] Melchert (1994a: 53, 63) works with PIE *[z] > PAnat. *[z] as well, a “voiced allophone of */s/”, giving e.g. “Hitt. ašduēr ‘twigs, brush’ < (virtual) *h2o-zd-wËr” as an example. Since I do not see any indication of voicedness as a distinctive feature in Proto-Indo-European, PAnatolian or Hittite (see especially § 1.3.2 below), I will not follow him in this regard.

[3] Cf. Melchert 1994a: 53, 62 for the observation that on the basis of the etymological connection between the Hitt. suffixes -(e)zzi(a)- (as in šarāzzi(a)- ‘upper’) and Lyc. -zze/i- (as in hrzze/i- ‘upper’) we must assume that assibilation of *t in the sequence *VtiV is already a Proto-Anatolian development, and that we therefore must reconstruct a PAnat. phoneme /ts/.

[4] Cf. Coulmas 2003: 41-9; Fischer 2001: 47-57.

[5] It has often been claimed that “diese Form der Keilschrift [= the Old-Babylonian cursive] im Zusammenhang mit Kriegszügen des hethitischen Großkönigs îattušili I. nach Nordsyrien (um 1550 v. Chr. gemäß der Kurzchronologie) von dort nach îattuša [...] gelangt sei” (HZL: 15). The discovery of a text (Kt k/k 4) at Kültepe (Kaniš) that palaeographically occupies “eine Position zwischen dem “Normal-aA [= altassyrischen]” Duktus einerseits und dem altsyrischen und dem althethitischen andererseits” (Hecker 1990: 57) shows that the transfer of the Syro-Babylonian scribal tradition into Asia Minor may have been a more gradual proces that predates the Hittites’ occupation of îattuša.

[6] In transliteration, phonetic signs are given in small italics.

[7] Sumerograms are transliterated in Roman capitals.

[8] Akkadograms are transliterated in italic capitals.

[9] Note that the Akkadian phonetic complement is transliterated in superscript.

[10] Determinatives are transliterated in superscript as well.

[11] But cf. MUNUSduttariata/i-.

[12] See the discussion s.v. *kuan-.

[13] Durham 1976: 364.

[14] Some signs are hardly used in the Hittite texts: e.g. BA predominantly occurs in names; GU is attested only once with a phonetic value ([p]a-an-gu-uš (StBoT 25.13 ii 9 (OS))); BE is used with the values pát, pít or pét only.

[15] E.g. Oettinger 1979a: 551f.

[16] Cf. Melchert 1994a: 13-4: “While a great number of words are spelled consistently with either the voiceless or voiced sign, this usage does not correspond in any meaningful way with the voicing quality of the sounds being indicated, based on their expected inherited value”.

[17] Compare ki-iš-a ‘I become’ /kísHa/ < *sHa < *œéis-h2o vs. ki-i-ša ‘he becomes’ /kÏsa/ < *sa < *œéis-o.

[18] Similarly, I use the term fortited for describing an original lenis stop that has become a fortis one (in analogy to lenited).

[19] Contra e.g. Melchert 1994a: 92, who calls this phenomenon a “regressive voicing assimilation”.

[20] Melchert (1994a: 20) is aware of this fact and therefore calls the “devoicing of word-initial stops”, which he assumes for Hittite as well as for Palaic and CLuwian, “an areal feature across Anatolia”.

[21] Cf. Melchert 1994a: 19.

[22] It is for instance possible that kikkiš- was created at a (post-Proto-Anatolian) period when the initial stops had merged into the fortis series, but that later on all initial stops became lenis again, so that attested kikkiš- in fact represents /gikis‑/.

[23] Which implies that we must assume that in forms like appeššar ‘limb’ < *h2p-éh1sh1r, attant- ‘clever’ < *h2t-ént-, or appanzi ‘they seize’ < *h1pénti, where the fortis stop at first sight seems to have been retained in a similar initial cluster, these consonants were in fact restored on the basis of the full grade stems *h2ep-, *h2et- and *h1ep-.

[24] Cf. Durham 1976: 109 for the observation that in the Akkadian texts written in Boğazköy the sign Ú could be used as ’ux, i.e. with initial ’aleph = [ʔ].

[25] Ibid.: 117 for the sign A as ’ax.

[26] See s.v. parai-i / pari- ‘to blow’ fur further treatment.

[27] Spelled with the signs ZA, ZE/I, ZÉ, ZU, AZ, E/IZ, UZ, GAZ, ZUL and ZUM, which in Akkadian are used for the emphatic º: ºa, ºé/í, ºe/i, ºú, aº, e/iº, uº, gaº, ºul and ºum respectively.

[28] Cf. Kouwenberg (2003: 83) who states that Akk. “emphatic” º in fact was glottalized /s/, which was realized as an affricate /ts/. Kimball’s suggestion (1999: 107) that “it is possible that °Z° represents a voiced pre- or postconsonantal /z/ resulting from voicing assimilation (e.g. za-ma-an-kur “beard” = [zmã(n)kur] (?) < IE *smówµ “beard” [...])” is entirely ad hoc: cf. cases where Hitt. ša-mV reflects etymological *smV.

[29] With /S/ as found in [e-ez-za-a]š-ši, cf. §

[30] E.g. ur-za-ke/a- = /Hortske/a‑/, which is the imperfective in ‑ške/a- of uart-i / urt- (cf. the one spelling ur-za-aš-ke/a-), or ‑an-za = /‑ants/, which is a in ‑s of the suffix ‑ant- (cf. the spelling ‑an-za-aš-ša /‑antSa/ = ‑anz + =()a).

[31] The only secure examples of assibilation of *di- in Hittite show an outcome š-, namely šīuš ‘god’ < *diËus and šīatt- ‘day’ < *diéot-.

[32] /Hw/ is the regular outcome of PIE *‑h2u-, On the basis of the fact that *‑h2u- yielded the Lycian monophoneme q = [kw], I conclude that /Hw/ was already phonemic at the Proto-Anatolian stage.

[33] So all PIE roots that seemingly had an initial *r‑, must in fact have had either *h1r‑, *h2r- or *h3r‑, the regular outcomes of which in Hittite were /ʔr‑/, /Hr‑/ and /ʔr‑/, spelled ar‑, ar- and ar‑, respectively.

[34] E.g. Melchert 1994a: 125.

[35] Note that the ‘plene’ spelling in this case does not indicate vowel length, but rather must be read as ’a-ar-aš- with the sign A = ’ax.

[36] One could argue that this rule has ceased to operate at the time that /ləgnuwántsi/ has become the phonemic form, but this is incorrect: the development “/uw/” > /um/ is synchronically still operative as can be seen from e.g. aumeni ‘we see’. This form is a MH creation that replaced OH umēni: if at that time the development /uw/ > /um/ had ceased to operate, the secondary form au- + eni should have yielded **aueni.

[37] At least in OS texts, cf. the regular development of OH ú-ra-a-ni /urʔā́ni/ > MH/NH a-ra-a-ni = /uərʔā́ni/ ‘burns’.

[38] A special case is the verb tar(k)u-zi ‘to dance’. As I will show s.v., this verb reflects *terkw-, of which the buccal part of *kw is lost in the cluster *rkwC (compare e.g. arzi ‘he has’ < *h2érkti). So *térkwti > Hitt. tar-ú-zi and, more importantly, impf. *trkwsé/ó- > OH ta-ru-uš-ke/a- > NH tar-ú-iš-ke/a-. Does the NH form tar-ú-iš-ke/a- have to be interpreted as /trwɨské/á‑/ and therefore OH ta-ru-uš-ke/a- as /trwské/á‑/ and tar-ú-zi as /tárwtsi/? Or can we assume that in NH tar-ú-iš-ke/a- the NH suffix-variant /‑ɨské/á‑/ has been secondarily introduced and that OH ta-ru-uš-ke/a- can be interpreted as /truské/á‑/ and tar-ú-zi as /tárutsi/?

[39] Which incidentally shows that */VuuV/ yields /VumV/, and not **/VmuV/).

[40] Except clusters that include labiovelars or the phoneme /Hw/: e.g. ku-ra-an-zi ‘they cut’ = /kwrántsi/, tar-u-uz-zi ‘he conquers’ = /tárHwtsi/.

[41] All examples are taken from Dercksen fthc.

[42] Note that this word often is spelled ne-pí-iš as well, without a plene ‑e-.

[43] Moreover, not every accented vowel gets lengthened, as we will see in the treatment of the historical phonological developments below.

[44] One could assume that these spellings are scribal errors (da-a-ga-an for da-ga!-a!-an and ta-a-ga-an for ta-ga!-a!-an), but this does not explain da-a-ga-a-an. One could alternatively assume that these spellings reflect phonetically real forms, e.g. with anaptyxis in the initial cluster and accent retraction (so /tágan/), but this is hardly credible and still does not explain da-a-ga-a-an.

[45] This goes for the signs PÉ/Í, DE/I, GE/I, KE/I, îE/I, RE/I, LE/I, ^E/I5, ZE/I, E/IP, E/IT, E/IK, E/Iî (which can be read Aî and Uî as well), E/IR, E/IM, E/IZ, KE/IP, KE/IR, KE/IŠ, KE/IT9, LE/IK, LE/IŠ, NE/IR, PE/IR, PE/IŠ, ŠE/IR, TÉ/ÍN, DE/IR, TE/IR and TE/IŠ, whereas the sign NI can be read NÉ as well and MI likewise MÉ (in spite of the separate signs NE and ME).

[46] Which has led the editors of CHD to the unfortunate choice to consider the two vowels equivalent for the purpose of alphabetization and to list them in the i position. Note that in the revised preface of CHD L-N: xii the tone is milder: “It is well-known that the vowels e and i often interchange in the spelling of Hittite words. In the earliest texts scribes clearly sought to maintain a distinction. What consistency underlies later usage and whether the post-OH spelling conventions also reflect a continuing phonological distinction between e and i are matters of controversy”.

[47] Note that the line reads (5) ku-iš-ki i-eš-zi, in which the preceding ‑i- of kuiški may have triggered this scribal error.

[48] Note that Kimball is not always careful in citing her examples. For instance, on p. 68-9 she cites the OS forms “a-ne-e-mi StBoT 25, 3 II 2, a-ne-ẹ-[nu-un KBo III 22 Rs, 48” as examples of words where the sign NE is used instead of NI. This is incorrect: the words are in fact a-ni-e-mi and a-ni-ẹ[-nu-un], and therewith are spelled just as all the other forms in the paradigm of anie/a-zi, namely with the sign NI.

[49] Certainly in li-i-ik < *h1lénœh.

[50] Thus zi-i-ik, which reflects *tíH-ge (cf. chapter 2.1).

[51] The sign Ù (Í) only occurs akkadographically as the conjunction Ù ‘and’ and sumerographically as Ù ‘dream’ and in LIBIR.RA (= Ù.RA) ‘old’; the sign U4 ()) only occurs as such in the sumerogram U4.SAKAR ‘crescent of the moon’ (its normal value in Hittite is ut, UD or UTU); U5 (C) is only used in GIŠLE-U5 ‘wooden tablet’; U8 (W) is only used as part of the sumerogram USDUîA (= U8.LU.îI.A) ‘sheep and goats’; U19 (–) is only used as such in the sumerograms DUMU.(NAM.)LÚ.U19.LU ‘human being’, IM.U19.LU ‘southwind, south’, LÚ.(NAM.)U19.LU ‘human being’ and NAM.LÚ.U19.LU ‘humanity’ (its normal value in Hittite is URU).

[52] E.g. Hart 1983: 124-132; Eichner 1980: 156f.

[53] E.g. Melchert 1994a: 26 states that “[c]ontrary to a number of claims, there is no good evidence that the Hittites use the signs u and ú to indicate phonemically distinct vowels”.

[54] As a comparable phonomenon, Rieken refers to the ‘breaking’ of *u to O in front of r, h and Ø in Gothic.

[55] For instance, Rieken assumes that in front of ‑s- an old /u/ remains /u/ and therefore is always spelled with Ú (a-šu-ú-ša-, a-ú-li-ú-šº, a-pu-ú-šº, etc.). This is contradicted by ku-u-uš and a-pu-u-uš, however, which are both attested thus hundreds of times. She acknowledges that these forms form “eine wirkliche Ausnahme” and states that “[e]ine überzeugende Erklärung hierfür sich nicht erkennen läßt”.

[56] Note that all alleged instances of a spelling u-up-p° of this verb and its derivatives are false: cf. s.v. uppa-i / upp-.

[57] In accordance with the view expressed above, we may expect that the oldest spelling of this form must have been **ú-uš-te-e-ni, cf. impf. ú-uš-ke/a- (OS).

[58] To my knowledge, the spelling ú-a° only occurs in ú-ar-aš-a-an-zi (KUB 10.66 vi 4), which duplicates [a-ar-aš-a-an-zi] (KBo 7.48, 12), and in ú-a-a-at-ten ‘you must send’ (KUB 14.14 ii 36), which clearly is an error for normal u-i-a-at-ten (see s.v. ue-zi / u- ‘to send’). The spelling u-aº is only attested in KBo 24.11 rev.? (10) [...]x-še-eš u-an-za an-na-n[e-kº], in which the interpretation of u-an-za (or 10-an-za?) is unclear.

[59] The attestation “u-a-al-lu-uš” (KUB 29.1 iv 9) in my view is to be read as 10 a-al-lu-uš (see s.v. UZUalla-, alli-). The spellings u-a-al-° and u-a-al-a-° are found in one text only, KBo 16.50 obv. 10, 15, 20, and are so exceptional when compared to the other spellings of al- (±300 times with a- in my files) that we can safely disregard them.

[60] Or /uə‑/, e.g. a-ra-a-ni = /uərʔā́ni/ ‘burns’ or a-al-a-an-zi = /uəlHántsi/ ‘they hit’.

[61] Of words that are normally spelled with a-, we find only a few forms that show ú-a-: ú-a-an-ti-a-an-ta-az ‘lightning(?)’ (KUB 17.10 ii 3) instead of normal a-an-t° (see s.v. ant-, antae-, antie/a‑); ú-a-ar-ra ‘help’ (KUB 31.4 obv. 3) instead of normal a-ar-r° (see s.v. arri- / arrai-); ú‑a-ar-ka-an-ta-an ‘fat’ (KBo 3.60 ii 3) instead of normal a-ar-k° (see s.v. arkant-); ú-a-ar-ša-ma-an ‘firewood’ (KUB 32.129 iv 3) instead of normal a-ar-š° (as attested in ibid. 4, see s.v. GIŠaršma-); and ú-a-aš-ta-i ‘offends’ (KBo 3.28 ii 10) instead of normal a-aš-t° (see s.v. ašta-i / ašt-). Since these are all unique forms that cannot compete with the manifold attestations with a- of the words to which they belong, I disregard them. The spelling ú-a-a-nu-a-ar (KBo 3.2 i 66 passim), instead of correct a-a-nu-mar is clearly due to the fact that the author of this horse-training text is non-native. The only word that shows genuine alteration is (u)āi- ‘woe’: a-a-i-in (StBoT 25.3 iv 14, 40, StBoT 25.7 iv 9); ú-a-a-i-in (StBoT 25.4 iv 27, 35, StBoT 25.7 iv 5); ú-a-a-i (KBo 3.6 i 29, StBoT 24 i 34, iii 56, KUB 21.12+ iii 39, Bronzetafel iv 9, 16, 27, KUB 26.32 i 14, KUB 22.70 obv. 16); and ú-a-i (VSNF 12.125 obv. 5, 10, 11, KUB 16.10, 7, KUB 23.1+ ii 32). Since this word is clearly onomatopoetic, it is irrelevant here.

[62] All other apparent examples like ue-zi / u- ‘to send’, ūk ‘I’, ūnna-i / ūnni- ‘to send (here)’, uni ‘that’, unu-zi ‘to decorate’, ūpp-zi ‘to come up (of the sun)’, uppa-i / uppi- ‘to send (here)’, ūrki- ‘trace’, ūššie/a-zi ‘to open (curtains)’ and uda-i / ud- ‘to bring (here)’ reflect *H(V)uC°, and therefore are treated under the paragraph ‘Word-internally between consonants’ (§

[63] From MH times onwards, this verb is spelled a-ra-a-ni = /uərʔā́ni/, but this is irrelevant here.

[64] Namely that every /u/ following /H/ or /h/ automatically turns into /o/.

[65] In a-ú-e-er and a-ú-er (never **a-u-e-er) ‘they saw’ and ma-ú-er (never **ma-u-er) ‘they fell’, the spelling with ‑ú- is influenced by the spelling of the diphthong /au/, which in these verbs is always spelled °a-ú-CV°. The word la-la-ú-e-ša- ‘ant’ (never **la-la-u-e-ša-) is the Luwian variant of Hitt. la-la-ku-e-ša-.

[66] The spelling ni-u-i-i[t] (KUB 31.91, 5), instr. of a-, must represent the same form as the spelling ne-e-u-it, and therefore should be read né-u-i-i[t].

[67] a-ša-ú-ni (KBo 6.2+ iii 49 (OS)), pár-ta-ú-ni-t=u-uš (KBo 17.1 i 6 (OS)), []r-ta-ú-na-aš (KUB 36.49 i 8 (OS?)).

[68] a-ša-ú-ni (KBo 6.3 iii 53 (OH/NS)), a-ša-ú-na-az (KUB 30.10 obv. 15 (OH/MS)), pár-ta-ú-ni-it (KUB 32.122, 6, 7 (MS?)) vs. a-aš-ši-a-u-ni-it (KUB 33.62 ii 20 (OH/MS)) and u-et-ti-a-u-ni (KUB 15.34 iv 61 (MH/MS)).

[69] a-ša-u-ni (KUB 13.5 ii 22 (OH/NS)), a-ša-u-na-az (KUB 13.4 iv 59 (OH/NS), KUB 24.3 ii 12 (MH/NS)), pár-ta-u-na-az (KBo 8.155 ii 9 (NS)), pár-da-u-na-az (KBo 27.163, 7 (MH/NS)), pár-da-a-u-na-za (KBo 33.188 iii? 14 (MH/NS)), pár-ta-a-u-ni-it (KBo 4.2 i 4 (OH/NS), KUB 15.31 i 35, ii 40 (MH/NS)), pár-ta-u-ni-it (KUB 15.32 i 37 (MH/NS), KBo 15.48 ii 6, 27 (MH/NS)), pár-ta-u-na-aš (VBoT 125, 3 (NS)), ša-ra-u-na-an-za (KUB 18.11 rev. 5 (NH)) vs. ar-ša-ú-n[i] (175/w obv. 8 (NS)), ar-ša-ú-na[-aš] (KBo 6.34 ii 39 (MH/NS)) and pár-ta-ú-ni-it (KUB 33.8 ii 16 (fr.), 17 (fr.) (OH/NS)).

[70] Cf. Rieken 1999: 352.

[71] The text is quite corrupt: KUB 9.22 ii (39) DUGkap-pí=ma-a=š-ša-an ku-iš (40) ar-na-a-ú-uš la-u-an-zi should actually have been ... ku-iš ar-na- la-u-an-za ‘what sap has been poured into the vessel’.

[72] Cf. CHD L-N: 1.

[73] Prof. Kortlandt informs me that from a typological point of view the lowering of /au/ to /ao/ before /n/ should be interpreted as the rise of nasal vowels: /aun/ > /a˜/.

[74] A sequence °aiu° does not originally occur in Hittite words: pa-a-i-ú ‘he must give’ is a secondary formation instead of more original pa-a-ú < *h1pói-u, in which the stem pāi- was restored. All other cases where we find °aiu°, we are dealing with either names or words of foreign origin.

[75] Note that CHD L-N: 307 incorrectly cites “mi-u-uš” (KUB 39.41 obv. 17 (NS), KUB 33.38 iv 10 (OH/MS)): these forms actually are mi-ú-uš.

[76] For /‑ʔ‑/, cf. § 1.4.5.b.

[77] Or /Héun/ > /Hé˜/, cf. note 83.

[78] Besides a few times a-pu-un, but these are irrelevant here.

[79] Besides a few times a-pu-uš, but these are irrelevant here.

[80] Except URULa-a-u-ú-i-a-aš-ši-iš (Bronzetafel i 69).

[81] Puhvel HED 4: 153 incorrectly cites the form of KBo 21.22, 10 as “ki-nu-ú-pí”.

[82] Besides a few times ku-uš.

[83] See CHD L-N: 26 for attestations.

[84] See CHD L-N: 86f. for attestations.

[85] See CHD L-N: 319f. for attestations.

[86] E.g. e-ša-a-ri (iv 13) instead of normal e-ša-ri, u-ma-an (i 15) instead of u-u-ma-an, é-u-uš (ii 22) instead of normal é-mu-uš.

[87] See CHD L-N: 333 for attestations.

[88] See CHD P: 373f. for attestations.

[89] See CHD P: 400 for attestations.

[90] See CHD P: 387 for attestations.

[91] See CHD P: 402 for attestations.

[92] Cf. Melchert 1993a: 232.

[93] The only exception in the whole Hittite corpus, u-ú-ni-ik-zi (KBo 6.2 i 16), must be a mistake, as we have seen above.

[94] Just as the ‘ligature’ I+A (†) is cited xA, the ligature ME+EŠ (^) is cited MEŠ, and SISKUR.SISKUR ("") is cited SÍSKUR.

[95] Cf. Rieken 2005 for a similar view of these words.

[96] Also in šar-lu-u-ma-aš /srlomás/, of the verbal noun of šarlae-zi ‘to exalt’, although in this case we are dealing with a secondary tarn(a)-class ending instead of expected šarlāar.

[97] Note that normally *ou yields au in front of *n, but in this case we are dealing with a pre-Hittite univerbation of the preverb *h2ou, which in isolation yielded /ʔu/, and the verb nai-i / ni- ‘to turn’.

[98] Incorrectly cited in CHD L-N: 330 as “mu-u-un-na-a-it”.

[99] Or this form, which is attested in a MS texts, represents /?(a)runi/ < *h3(o)ru-n-i, in which the lowering of /Cun/ to /Con/, which starts within the MH period, has not taken place yet.

[100] This means that in effect we are dealing with a preservation of PIE *o as Hitt. /o/ in the position *Cóm.

[101] One could think of e.g. *prtuHn- < *prtH-u-n- vs. *prtH-o-un- > partāun-.

[102] The original distribution must have been ūC° vs. ūnV°, so **ūzi / ūnanzi.

[103] Thus also Rieken 2005.

[104] Unless we assume that the two forms that are spelled lu-ú-ri- (cf. § as well as s.v.) are mistakes. Then, on the basis of a-ni-u-ur < *h3n-ié-ur, we should assume that /Ceur/ > /Cor/.

[105] Note that ūššie/a-zi ‘to draw open (of curtains)’, spelled ú-uš-ši- and uš-ši- = /ʔuSie/a‑/, reflects an univerbation of the preverb *h2ou and the verb *h1s-ié/ó-, which took place at a stage when *h2ou had already become /ʔu/ in isolation.

[106] Cf. note 83.

[107] Note that uda-i / ud- ‘to bring (here)’, which is spelled ú-d° = /ʔud°/, reflects a univerbation of the preverb *h2ou and the verb *deh3- which took place at a stage in which *h2ou had already become /ʔu/ in isolation.

[108] Note that the handcopy of KUB 23.8 seems to show a form a-a-nu-u-u[t] in line 7. The photograph of this tablet (available through Hetkonk), in my view rather shows a-a-nu-u[t], however. Compare also line 8 where the photograph clearly shows i-da-a-lu, which turns up in the handcopy as i-da-za-lu.

[109] Possibly /o/ when in *C_lVback and in *CulH.

[110] Or /a˜/, cf. note 83.

[111] Or /˜/, cf. note 87.

[112] On the basis of a-ni-u-ur < *h3n-ié-ur. If lu-ú-ri- indeed reflects *leh1uri-, then *Ceur > Hitt. /Cūr/ with the exception that *ieur > Hitt. /ior/.

[113] Or /ų/, cf. note 83.

[114] Or /˜/, cf. note 87.

[115] This consistency is also found in the spellings of names: e.g. mGa-aš-šu-ú or mKa-aš-šu-ú is never spelled **mKa/Ga-aš-šu-u; mUz-zu-u is never spelled **mUz-zu-ú.

[116] The two remarkable spellings ge-en-zu-ú (KUB 31.127 i 4) instead of normal ge-en-zu and šu-ú-a-ru-ú (KUB 12.29, 3) instead of normal šu-(ú-)a-ru do not contradict this: they are just occasional spellings that stress the u-ness of the word-final vowel.

[117] The words šīšau and garāu are real hapaxes, GIŠmarāu occurs in ma-ra-a-u-i (1256/v obv. 3) as well.

[118] The reconstruction of ši-i-iš-a-u as *si-sh2ou is far from certain, q.v.

[119] A concrete case could e.g. be the form sākan ‘oil’ followed by the enclitic possessive pronoun =šmi-, =šma-, =šme- ‘your (pl.), their’, which should have been spelled **šākašmet ‘your / their oil’, representing [sā́gəsmed] < *[sā́gənsmed] < *œ(h)n=smed.

[120] Lenition is found in the other Anatolian languages as well, under the same conditions (cf. Melchert 1994a: 60 for examples), which implies that this was a PAnat. phenomenon. See Adiego 2001 for the argumentation that viewed in moraic terms the two lenition rules can be regarded as one.

[121] This rule is first formulated by Eichner 1973: 79.

[122] This shows that at the moment that lenition took place, the result of the monophthongization of *ei and *oi was */ē/, which was still different from original short */ĕ/, which did not cause lenition. Only later on, probably as a result of the loss of a distinction in length between */ē/ and */ĕ/ in unaccented syllables (due to the weakening of unaccented */ĕ/ to /i/ and /a/), accented */ē/ and /ĕ/ merged into Hitt. /e/.

[123] This is the only reasonable way in which we can explain the frequent alternation in i-verbs between a lenis stem-final consonant in 3sg.pres.act. and a fortis one in 3pl.pres.act., e.g. ištāpi / ištappanzi, which reflects *stópei / *stpénti, where *ó > PAnat. /Ö/ > Hitt. /ā́/ caused lenition of the following *p to /b/.

[124] This example shows that the assimilation of *VmsV to /VSV/ antedates the process of lenition.

[125] Examples like ārri ‘he washes’ < *h1órh1ei and šākki ‘he knows’< *sókh1ei show that at the moment of lenition *‑rh1- and *‑kh1- phonologically still counted as clusters (lenition only affects intervocalic consonants!). So the assimilation of *Vrh1V to /VRV/ and the disappearance of *h1 in *Ch1V only took place after the lenition.

[126] First suggested by Eichner 1973: 10086.

[127] Compare the Lycian ending χaga < PAnat. */‑Haha/.

[128] This example shows that the rise of the epenthetic vowel /ɨ/ in a cluster *VKsC postdates the fortition of */g/ to /k/ before ‑šk-.

[129] Possibly through a stage */Htsé/á‑/ in which the suffix e/a- was secondarily restored.

[130] With secondary addition of ‑i in analogy to the other present verbal endings in ‑i (-mi, ‑ši, eni, ‑tteni).

[131] The assibilation of *d- > /s/ in front of *i must be post-PAnatolian, since it does not occur in the other Anatolian languages (e.g. Luw. (d)Tiad- ‘Sun-god’ < *diéuot-, Pal. tiuna- ‘god’ < dieu-). Note, however, that no examples of *VdiV > /VsiV/ (a development that on the basis of *VtiV > PAnat. */VtsiV/ could a priori be expected to have taken place in PAnatolian as well) are known.

[132] Contra Kimball 1999: 292 who explicitly states that “*dh was not assibilated before *y”.

[133] We would expect lenition of * to /g/ here (*r > **/sā́gr/ like *‑ótr > Hitt. /‑ā́dr/, °a-a-tar), but apparently /k/ was restored in analogy to the oblique cases /skn‑/ < *s-n-.

[134] Contra Melchert 1994a: 61, who claims that “[t]he PIE voiceless labiovelar */kw/ is [..] weakened to PA[nat.] */gw/ in medial position”.

[135] The preservation of *kw- in kutruen- < *kwtru-en- ‘witness’ then must be due to restoration in analogy to the expected full grade form *kwetur-, which is not attested in Hittite anymore, however.

[136] See at § below for the fact that this i- does not partake in the lowering of OH /i/ to NH /e/ before /s/, which indicates that this vowel was phonologically different from /i/ < *i.

[137] The geminate ‑šš- in the CLuwian cognate īš(ša)ra/i- ‘hand’ may show that fortition of *s to /S/ before *r is PAnatolian already.

[138] Note that all instances of V in Hittite must be of secondary origin, e.g. āššie/a-tta(ri) is a secondary e/a-extension of āšš-ari; peššie/a-zi and ūššie/a-zi are secondary formations of pe+šie/a-zi and u+šie/a-zi; ašie/a-zi ‘to buy’ is a secondary stem on the basis of original āš-i / aš-; ešie/a-tta(ri) is based on the noun eši- / ešai-; etc.

[139] Although the spelling ‑Vp-šV- does not reveal anything about whether the ‑s- is single or geminate and although no spellings can be found that expressly indicate singleness (never **-pa-šV-) or geminateness (never **-pa-aš-šV), the fact that the *s fortites to /S/ in *VKsV and *VTsV makes it highly plausible, in my view, that this happened in *VPsV as well.

[140] Melchert 1994a: 77 states that *Vh2sV > Hitt. VššV on the basis of his reconstruction of the genitival suffix ‑ašša- as *‑eh2so-. See s.v. ‑ašša- for a discussion of this etymology, which seems incorrect to me.

[141] Since lenition is a PAnat. feature, the assimilation of *VmsV to /VSV/ must be PAnatolian as well.

[142] Perhaps we must assume on the basis of *VlHsC > /VLɨsC/ and *VmHsC > /VMɨsC/ that the regular outcome of *VrHsC was /VRɨsC/ and that in ú-e-ri-iš-ke/a- the single ‑r- was introduced from the indicative er(ie/a)-.

[143] If a-a-ni-iš, a-a-ni-eš /Hā́nɨs/ ‘wipe!’ < *h2ómh1s and a-a-ni-iš-šu-a-ar, a-a-ni-eš-š[u-a-ar] /Hā́nɨSuər/ ‘wiping’ < *h2ómh1s-¾µ are really phonetically regular, the different outcome may be due to the preceding leniting *ó. Note that these forms may also show that the ‑m- in damme/išā- at one point has been restored in analogy to the verb tamāšš-zi / tame/išš-.

[144] Although on the basis of *CmHsV > Hitt. /CnɨSV/ we may rather have to assume that regularly *CmHsC yields /CnɨsC/ and that in dameške/a- the ‑m- has been restored in analogy to the verb tamāšš-zi / tame/išš-.

[145] See also Kavitskaya 2001: 278f. for a treatment of the anaptyctic vowel /ɨ/ and the factors that determine its place within a cluster.

[146] Note that there is no anaptyctic vowel in ur-za(-aš)-ke/a- /Hortské/á‑/ ‘to curse (impf.)’ < *h2urt-sé/ó-, iš-pa-an-za-aš-ke/a- /ispndské/á‑/ ‘to libate (impf.)’ < *spnd-sé/ó- and ma-al-za(-aš)-ke/a- /məldské/á‑/ ‘to recite (impf.)’ < *mldh-sé/ó-. This must be due to the fact that a resonant is preceding the dental consonant.

[147] Note that OH i-in-ga-aš-ke/a- /Hinkské/á‑/ yields MH i-in-ki-iš-ke/a- /Hinkɨské/á‑/.

[148] The ‑z in a-aš-te-er-za /Hstérts/ ‘star’ < *h2stËr + s must be due to the secondary attachment of the commune ending ‑s to the stem /Hstér/.

[149] Note that in ú-a-an-zi /ʔuántsi/ ‘they see’ < *Hu-énti and ú-a-a-tar /ʔuā́dr/ ‘inspection’ < *Hu-ótr the /ʔ‑/ must be restored on the basis of e.g. ú-me-e-ni /ʔuméni/ ‘we see’ and uš-te-e-ni /ʔusténi/ ‘you see’. The form ú-a-an-ši-ke/a- ‘to copulate (impf.)’ = /ʔuənsɨke/a‑/ goes back to *h1/3unsé/ó- in which the initial laryngeal is regularly retained before vocalic ‑u-.

[150] This form shows that the lost of intervocalic /ʔ/ is a late OH phenomenon.

[151] Since none of the examples is fully certain, this sound law must be regarded as provisional. Note however that Kimball’s example (1999: 400) in favour of a development *Vh2nV > Hitt. VnnV is incorrect, see s.v. annum(m)ia-. In word-final position, note the difference between *séh2n > Hitt. /sáHn/, ša-a-a-an ‘feudal service’ and *duéh2m > Hitt. /tuā́n/, tu-a-a-an ‘to this side’. The latter development may be PIE already, which is commonly referred to as ‘Stang’s Law’.

[152] This fortition may have been PAnatolian already, if indeed CLuw. ‑d(d)uar(i) (2pl.midd. ending) reflects *‑dhh2u-, see s.v. ‑ttuma(ri), ‑ttumat(i), and cf. Melchert 1994a: 77.

[153] This latter example shows that the laryngeal must have been present up to the times of assibilation of dentals by a following ‑i-. Because this assibilation is pre-Hittite only, and not Proto-Anatolian, the laryngeal was still present at the PAnatolian stage, hence the PAnat. reconstruction */TʔV/.

[154] This latter example shows that *h1 does not cause fortition of a preceding *D.

[155] This example is crucial. If *pri-prh1-ói-ei would have yielded Hitt. **/priprā́i/, without retention of *h1 as /ʔ/, it would have been spelled **pa-ri-ip-ra-a-i.

[156] I regard the apparent retention of interconsonantal laryngeals in e.g. pár-a-zi = /párHtsi/ < *bhérh2/3ti and a-al-a-zi = /uálHtsi/ < *uélh3ti as secondary: the laryngeal must have been restored on the basis of 3pl.pres. *bhrh2/3énti and *ulh3énti where it was regularly retained.

[157] This latter example shows that the loss of *h1 predates the development *eRCC > Hitt. /aRCC/.

[158] Apparent retention of *h2 and *h3 as /H/ in word-final position in words like ma-ni-a-a ‘distribute!’ < virtual *mniéh2 and a-al-a ‘strike!’ < virtual *uélh3 is of course due to restoration in analogy to the rest of the paradigm.

[159] Cf. Eichner 1973: 9878, Melchert 1994a: 87 and Kimball 1999: 354-5. Eichner formulates the rule too broad (“[a]uslautendes ‑r schwindet generell nach unbetontem Vokal”): compare Kimball who points out that loss is only found after the vowel ‑a- (retention of *‑r after unaccented *u is found in e.g. *péh2ur > Hitt. pa-a-ur ‘fire’; Eichner’s example *pérur > Hitt. pé-e-ru rather shows dissimilation due to the first ‑r-). Note that Eichner explains the final ‑r in ātar ‘water’ as analogical after itār, but this is unnecessary: Hitt. ātar represents /uā́dr/ < *uódr in which *‑r never stood after a vowel.

[160] Also attested as še-me-en-zi and ši-me-en-zi = /sɨméntsi/.

[161] Note that this form is not attested with the spelling **ša-am-ma-na-an-zi. This could either be ascribed to chance, or it could indicate that in this form the lenis /m/ of the singular, *smén- > /smén‑/, has been introduced. The form is also attested as še-em-na-an-zi = /sɨmnántsi/, cf. note 170.

[162] Cf. Melchert 1994a: 81.

[163] This form shows that fortition of *m in front of ‑r- postdates the lenition of intervocalic fortis consonants due to a preceding *ó (including *‑óCr#, cf. *‑ótr > Hitt. ‑ātar).

[164] Or /°Cų/, cf. note 83.

[165] Or /°C˜/, cf. note 83.

[166] In word-final position there is (at least graphically) no difference between /S/ and /s/, and I will therefore write /Vs#/ here.

[167] The 2sg.pres.act. form ku-e-ši ‘you kill’ reflects pre-Hitt. */kuénsi/, which must show restoration of the stem kuen- in expected **/kuáSi/ < *gwhénsi. So the fact that *VnsV here yields /VsV with single ‑š- is due to the fact that the disappearance of *n in this restored form postdates the development *‑ensV- > Hitt. /‑aSV‑/.

[168] The verb šamenu-zi ‘to make (something/-one) pass by’ reflects pre-Hitt. *smen-n(e)u-, which must be a secondary formation replacing expected *smn-n(e)u-. So the fact that *VnnV here yields /VnV/ with single ‑n- is due to the fact that the disappearance of *n in this form postdates the development *VnnV > /VNV/.

[169] Technically, most of these examples reflect *VmmV, however. The 1sg.pres.act. form ku-e-mi ‘I kill’ < pre-Hitt. */kuénmi/ seems to show a development *VnmV > Hitt. /VmV/ with single ‑m-. I assume that this */kuénmi/ was a restored form that replaced expected */kuéMi/ < *gwhénmi, and that the development pre-Hitt. *VnmV > Hitt. /VmV/ is due to the fact that the disappearance of *n in this form took place after the development *VnmV > /VMV/ had come to an end. This also explains forms like tu-uz-zi-ma-an < *tuzzin=man ‘my army’ and u-u-a-ma-an < *uŸŸan=man ‘my grandfather’.

[170] See CDH L-N: 144 for attestations of these forms. Occasionally, the ‑n- is restored, yielding ma-a-an-a. The preservation of ‑n- in *h3unuénti > Hitt. /ʔunuántsi/, ú-nu-a-an-zi ‘they adorn’ is due to restoration of the causative suffix ‑nu-.

[171] Or /ʔéga‑/?

[172] Note that the etymology is uncertain.

[173] Or /ʔéuan‑/?

[174] Note that e.g. a-an-zi ‘they go’ < *h1iénti and i-a-an-t- ‘sheep’ < *h1iént- (?) show that *i is not lost in the word-initial sequence *h1ie-.

[175] This form shows that the assimilation of *VsiV > /VSV/ postdates the lenition of intervocalic fortis consonants due to a preceding long accented vowel.

[176] With *- in analogy to the full grade stem *es-, cf. the lemma ešš-tta, ašše/a-zi.

[177] Attested from MH times onwards as aššie/a-zi, with restored suffix e/a-.

[178] See at the lemma of ending ‑eš for my view that contraction of *‑eie- to *‑ē- must have happened earlier than the loss *i between other vowels as described here.

[179] This explains the fact that in the oblique cases of i-stem adjectives the ‑a- of the suffix, which must go back to post-tonic *e in open syllable (*CÕC-ei-V), is often spelled plene (e.g. ar-ša-a-i, ar-ša-a-eš, ar-ša-a-uš), whereas such a plene spelling is virtually absent in the oblique cases of u-stem adjectives (e.g. a-aš-ša-u-i, a-aš-ša-a-az, a-aš-ša-u-e-eš, a-aš-ša-mu-uš < *CÕC-eu-V).

[180] Also intervocalic ‑i- from secondary sources is lost with lengthening of the preceding vowel: e.g OH /páiitsi/, pa-i-iz-zi ‘he goes’ > MH /pā́itsi/, pa-a-iz-zi (see s.v. pai-zi / pai- for an extensive treatment).

[181] Note that in the case of Hitt. /e/ and /o/ I do not reckon with phonemic length, and that therefore these vowels do not get lengthened.

[182] In this latter verb, intervocalic /i/ is nevertheless often found in NH forms because of restoration of the suffix e/a-: NH ta-a-i-e-ez-zi = /tāiétsi/ and NH ta-a-i-a-zi = /tāiátsi/. Compare, however, phonetically regular 2pl.pres.act. ta-a-et-te-ni (NH) = /tāéteni/.

[183] This implies that the loss of word-final *i took places in several stages, probably determined by the preceding consonant.

[184] Possibly through a stage */Htsé/á‑/ in which the suffix e/a- was secondarily restored. Note that Melcher’s equation (1994a: 62) between HLuw. ha-zi-mi-na and Hitt. azzie/a-, on the basis of which he assumes that assibilation in a cluster *CtiV took place in PAnatolian as well (with retention of *i), cannot be substantiated (cf. s.v. att‑a(ri), azzie/a-zi).

[185] Replaced by e-eš-zi with restored ‑i in OS texts already.

[186] Similarly in e/a-zi ‘to steal’ (cf. note 192) and in ašše/a-zi >> aššie/a-zi ‘to put on clothes’ (cf. note 187).

[187] The verb zanu-zi ‘to cook (trans.)’ < *tih1-neu- shows that *h1 at one point had enough vocalic quality to prevent *i from becoming vocalic too.

[188] This would mean that the forms e-eš-zi ‘he is’ < *h1ésti (instead of expected **e-eš-ti), še-eš-zi ‘he sleeps’ < *sésti (instead of expected **še-eš-ti), etc. all show (a trivial) generalization of the assibilated ending ‑zi.

[189] Although here we might be dealing with the secondary introduction of the full grade stem ēš- from the ēšar.

[190] Melchert’s reformulation (1994a: 133) of this rule as /‑iCCa‑/ > /‑eCCa‑/, i.e. “a kind of “a-umlaut” in closed syllables” (apparently to explain idālu- and iar) still does not account for these forms. Rieken’s treatment (1996: 294-7) of the lowering cannot fully convince either (her formulation of the conditioning environments, namely -iCCa- > -eCCa and i > e “zwischen zwei Dentalen (t, n, s) oder zwischen l und einem der genannten Laute”, does not account for all examples, like ištā). Moreover, many of the examples of e~i that she cites in fact are cases where I assume the epenthetic vowel /ɨ/.

[191] The absence of lowering in kiššan ‘thus’ (but compare the one spelling ke-eš-ša-an (KBo .4 iii 17)) and kinun ‘now’ seems to be caused by the connection with  ‘this (’ (although ki-nu-un in principle can be read ke-nu-un as well). Note that words like gimmant- and gimra- are spelled with GI-IM- that can in principle be read gi-im- as well as ge-em-. Cases like innara° and inan- are real counter-examples, however: they never show lowering.

[192] There are no examples of /um/, though. Note that the i- that emerges in initial clusters of the shape *sT- did not partake in this lowering and therefore must have been phonologically different from /i/ < *i and pretonic *e.

[193] This example shows that ‘vocalization’ of *R in the initial sequence *uRC- is an inner-Hittite phenomenon: PIE *uRC- > OH /uRC‑/ > MH/NH /uəRC‑/. This also fits the words ulkiššara- (OH/NS) > alkiššara- (NS) ‘skilled’ and alkua- ‘bad omen(?)’ (earliest attested in a MS texts) if this indeeds reflects *ulkwo-. This would mean that u-ur-ki- ‘track, trail’ cannot go back to *urK-, but must represent /ʔorKi‑/ < *h1/3urKi-

[194] E.g. tuān /tuā́n/ ‘to this side’ < *duéh2m shows that this development only occurs when *u precedes *o, and not when it precedes “*ā”. The loss of *u in *Tuo therefore must be dated to the period before the merger of *o and PAnat. *a/ā.

[195] Through */CouV/? See below for *CuuV > Hitt. /CumV/.

[196] Prof. Kortlandt (p.c.) points out to me that a development of *u to /m/ is phonetically incomprehensible: nasilization does not occur spontaneously. He therefore suggests that /m/ in these cases must be interpreted as /w/ (i.e. a consonantal /ų/) and that the nasalization must be due to contact with other nasal vowels or nasal consonants. E.g. °Camuš ( ending of u-stems) is interpreted by Kortlandt as /°Cawųs/ < */°Cawųs/ (see also § under C_š for Kortlandt’s view that the ending ‑uš represents /‑ųs/ < *‑ms); °numēni as /°nuwéni/ in which *e has become nasalized due to *n, and therefore causes *u to yield the nasal-vowel /ų/; etc. Similarly in *CHuV > /ComV/ as treated above: tu-me-e-ni is interpreted by Kortlandt as /t˜wéni/; etc. Although the assumption of nasal vowels would indeed fit other developments as well (especially the development of OH /un/ to /on/, cf. note 83) and explain the rise of a nasal consonant here, it remains problematic why we do not find a nasal consonant in e.g. aniaanzi ‘to work’ = /ʔniauántsi/ (instead of **aniamanzi = “/ʔniawËntsi/”) and why šumariške/a-zi ‘to be pregnant’, which reflects *s(e/o)uH-μ+, shows ‑m- whereas no nasal consonant is present in its preform. Moreover, nasalization as the result of dissimilation has also occurred in ú-ra-a-ni ‘burns’ < *urāri < *urh1óri, where there is no other nasal element that could have caused it.

[197] The form a-ú-ma-ni is MH and replaces OH ú-me-e-ni < *Hu-uéni.

[198] In *élh1st(o) the colouring of *e > /a/ did take place, however: *élh1st(o) > Hitt. /káLɨsta/, kal-li-iš-ta ‘he called’. This implies a scenario *élh1st(o) > */kéLsta/ > */káLsta/ > /káLɨsta/.

[199] This latter form clearly shows that the raising of *e to /i/ between *u and *T predates the weakening of post-tonic *e to /a/ in open syllables.

[200] Melchert 1994a: 83.

[201] Cf. Melchert 1994a: 138.

[202] This /i/ can itself in younger Hittite become subject to the lowering to /e/ as described in §, cf. *nébhes > OH /nébis/, ne-e-pí-iš (OS) > NH /nébes/, ne-pé-eš (NS).

[203] Cf. Melchert 1994a: 139. See s.v. nepiš- ‘sky’ for the demonstration that this weakening of unaccented *e can be dated to the 18th-17th century BC.

[204] This is a powerful explanation for the replacement of the original 3sg.pret.act. ending *‑e of i-inflected verbs by the corresponding ending of the sigmatic aorist, *‑s-t > Hitt. ‑š.

[205] See s.v. nepiš for explanation of the oblique cases nepišV.

[206] Note that I do not follow Garrett (apud Melchert 1994a: 184) who suggests that we are here dealing with a development *=kwe > *=kwə > Hitt. /=kwu/, spelled V=k-ku.

[207] Which must be dated to the 18th-17th century BC, cf. s.v. nepiš- ‘sky’.

[208] With the disappearance of the Hittite examples in favour of ‘Eichner’s Law’, it has in my view become highly unlikely that this law can be upheld for the whole of Indo-European, especially with regard to words like *gwËh3us > Skt. gáus (not **jáus), Gr. βοῦς, *gwËh3um > Skt. gā́m (not **jā́m), Gr. βῶν ‘cow’, *nËh2us > Skt. náus, Gr. ναῦς, *nËh2um > Gr. νᾶν ‘ship’, *sËh2ls > Lat. sāl, Gr. ἅλς, Latv. sãls ‘salt’ (cf. Kortlandt 1985a: 118-9) and possibly *nēh2u- > Lat. nāvis ‘ship’, *h2ē- > Lat. ācer ‘sharp’ (cf. Schrijver 1991: 130-4).

[209] It must be mentioned that the practice of plene spelling is less consistent in younger texts anyway, so that it is possible that in these texts also sequences like /ā́CV/ and /ā́C#/ are spelled without plene ‑a-.

[210] Apparent counter-examples like duggāri, kištāri, etc. must reflect *CC-ór, to which the ‘presentic’ ‑i was attached only after the development of *ó > /ā́/ in final syllables. So *tuk-ór, *œhsd-ór > pre-Hitt. *tukkār, *kištār >> Hitt. tukkāri, kištāri.

[211] Or /‑ā́s/ if we assume that the expected spelling *ku-a-a-aš = /kuā́s/ is unattested only by chance.

[212] Except in 2sg.imp.act. of the dāi/tianzi-class (e.g. da-i ‘place!’ instead of expected **te < *dhh1-ói), where ‑ai was restored on the basis of the stem °Cai- as found in the rest of the paradigm.

[213] See s.v. for the possibility that this word reflects *œhréh1-u.

Etymological Dictionary of the Hittite Inherited Lexicon

Aspects of Historical Morphology

2.1 The Hittite nominal system

Since recently a detailed monograph dealing with the Hittite nominal system has appeared (Rieken’s Untersuchungen zur nominalen Stammbildung des Hethitischen (1999a)), and since almost each noun is in Part Two extensively treated regarding its morphological prehistory, it is not necessary to treat the Hittite nominal system as thoroughly as the verbal system. Nevertheless, I want to make explicit which system of nominal inflection I reconstruct for PIE and in which way this system is still traceable in the Hittite material.

For PIE, I largely follow the system of nominal inflection as described by Beekes (1985 and 1995: 168f.). We must distinguish three basic types: root nouns (i.e. nouns in which the ending is directly added to the root), consonant stems (i.e. nouns in which a suffix of the structure *-(C)eC(C)- is placed between the root and the ending) and thematic stems (i.e. nouns of which the stem ends in *-o-).

Beekes (1985) has shown that for early PIE we must reconstruct three accent types of inflection of consonant stems, from which all attested types can be derived. These three are:

                        hysterodynamic               proterodynamic               static      *CéC-C(-s)                        *CéC-C(-s)                    *CéC-C(-s)        *CC-éC-m                         *CéC-C(-m)                  *CéC-C(-m)        *CC-C-ós                          *CC-éC-s                      *CéC-C-s

The root nouns could show either static or mobile accentuation:[1]


                                    static root nouns                      mobile root nouns                  *Cé/óC(-s)                                *Cé/óC(-s)                    *Cé/óC(-m)                               *Cé/óC(-m)                    *Cé/óC-s                                   *CC-ós

The thematic type was a recent innovation, based on the hysterodynamic form, which originally had the function of ergative. Thematic nouns therefore do not show ablaut or accentual mobility.

In Hittite, the three PIE basic types are attested as well:

thematic stems: a-stem (also adjectives)

consonant stems: i-stem (also adjectives), u-stem (also adjectives), au-stem, ai-stem, t-stem, s-stem, -stem, *m-stem, n-stem, r-stem, r/n-stem, nt-stem (also adjectives), it-stem



2.1.1 Thematic stems

The Hittite a-stem inflection goes back to the PIE o-stem inflection and is known from commune as well as neuter words. A-stem nouns originally do not show ablaut or mobile accentuation. Whenever they do, they must be regarded as recent thematicizations of original root nouns or consonant stems (thematicization is a productive process within Hittite). The Hittite endings are as follows (see s.v. for a detailed account of their origin):

    sg.                                                                           pl.

    nom.c.             -aš       <      *-o-s                            nom.c.         -eš       <      *-ei-es[2]

    acc.c.               -an       <      *-o-m                          acc.c.           -uš       <      *-o-ms

    nom.acc.n.       -an       <      *-o-m                          nom.-acc.n.  -a         <      *-eh2

    gen.                 -aš       <      *-o-s                            gen.             -an       <      *-o-om

    dat.-loc.           -ai, -i   <      *-o-ei,[3] *-i[4]           dat.-loc.       -aš       <      *-os(?)227

    all.                   -a         <      *-o227


                                                abl.          -az   <      *-oti

                                                instr.        -it     <      *-t227



2.1.2 Consonant stems

The Hittite consonant stems show the following endings (see also s.v.):

    sg.                                                                               pl.

    nom.c.         ,[5]              <      *-s,[6] *-Ø                 nom.c.             -eš       <      *-ei-es[7]

    acc.c.           °V-n, °C-an     <      *-m, *-o-m[8]             acc.c.               -uš       <      *-ms

    nom.acc.n.                       <      *-Ø                          nom.-acc.n.      -a,   <      *-eh2, *

    gen.             ,[9] -aš             <      *-(o)s                      gen.                 -an       <      *-om

    dat.-loc.       -i                      <      *-i                            dat.-loc.           -aš       <      *-os(?)

    all.               -a                     <      *-o


                                                abl.          -z, -āz          <      *-(o)ti

                                                instr.        °V-t, °C-it    <      *-t     i-stems and u-stems

The Hittite i-stem and u-stem nouns reflect the PIE proterodynamic i-stem and u-stem inflection. In substantives the ablaut has generally been given up, generalizing the zero grade of the suffix, CVC-i- and CVC-u- (only a few traces remain, e.g. in eši- / ešai- ‘pasture’ and ēu- / ē()a- ‘rain’). In adjectives the original ablaut has been retained, however, albeit that in these nouns root accentuation has been generalized. Due to loss of intervocalic -i- in pre-Hittite times as described in ḱ, the paradigm of the i-stem adjectives has sometimes become muddled. Examples: arki- / argai- ‘white’, tēpu- / tēpa- ‘little, few’.

                            Hitt.                                                                                 PIE         arkiš      <                                                                      *h2ér-i-s           ḫarkin    <                                                                      *h2ér-i-m             argaš    <      *argaaš  <      *h2ér-ei-os    <<       *h2r-éi-s       argai     <      *hargai     <      *h2ér-ei-i       <<       *h2r-éi-i         tēpuš        <                                                                      *dhébh-u-s           tēpun       <                                                                      *dhébh-u-m             tēpaaš   <                                  *dhébh-eu-os    <<       *dhbh-éu-s       tēpai      <                                  *dhébh-eu-i      <<       *dhbh-éu-i


Note that it often is assumed that the word for ‘knee’ reflects a PIE static paradigm *ón-u, *en-u-s (cf. Beekes 1995: 188), whereas the Hittite stems gēnu- and ganu- rather point to an original proterodynamic inflection: *én-u, *n-éu-s.     au-stems and ai-stems

These so-called ‘diphthong-stems’ (cf. Weitenberg 1979) reflect the PIE hysterodynamic i-stem and u-stem inflection. The few substantives that inflect thus clearly show that originally ablaut was still present, although in the course of Hittite the full grade stems in -au- and -ai- have been generalized. In forms with and without are attested (the latter often showing neuter concord). This situation is due to the fact that originally PIE commune nouns did not have a ending at all and that *-s was introduced as the new marker of only after the creation of the o-stem inflection. This process of sigmatization can still be observed in the oldest layers of Hittite (cf. Weitenberg 1995) and the ending eventually has become obligatory for forms of commune words. All forms that did not show this ending automatically were regarded as neuter. Examples: zaḫḫai- / zaḫḫi- ‘battle’, arnāu- / arnu- ‘birthing chair’.


                    Hitt.                                                         PIE   zaḫḫaiš       <                                          *tiéh2-ōi(-s)         (<< *tiéh2-i(-s))     zaḫḫain       <      *tiéh2-oi-m      <<       *tih2-ói-m            zaḫḫi     <      *tiéh2-i-os        <<       *tih2-i-ós   arnauš      <                                          *h3ér-nōu(-s)      (<< *h3ér-nu(-s))     arnaun      <      *h3ér-nou-m    <<       *h3r-nóu-m     arnu    <      *h3ér-nu-os     <<       *h3r-nu-ós     t-stem nouns

In Hittite, only a few t-stems are attested, which do not show synchronic ablaut anymore. Nevertheless, the comparison of Hitt. šīatt- ‘day’ < *diéuot- with CLuw. (d)Tiad- ‘Sun-god’ < *diuot- ˊ shows that ablaut must have existed at the Proto-Anatolian level and that this word ultimately reflects a hysterodynamic paradigm. I therefore reconstruct as follows:


                    Hitt.                                             PAnat.                 PIE   šīaz       <                                                              *diéu-ot-s    (<< *diéu-t(-s))     UD-an     < *diéu-ot-om     <<   *diu-ót-om <<   *diu-ót-m     šīattaš   <  *diéu-ot-os      <<   *diu-ot-ós    <<   *diu-t-ós     s-stems

For Hittite, only two neuter s-stems are attested, nēpiš- ‘heaven’ and aiš / išš- ‘mouth’. The former synchronically does not show ablaut anymore. Nevertheless, the attestation of the deity Nepaš ‘Storm-god’ in OAss. texts, which likely is originally identical to ‘heaven’, shows that at a pre-Hittite stage ablaut was still present: nepaš vs. obl. nepiš-. Furthermore, the existence of CLuw. tappaš- < *nébe/os- besides HLuw. tipas- < *nebés- shows that in Proto-Anatolian accentual mobility still existed in this word, going back to a proterodynamic inflection. Thus the following scenario emerges:


                            Hitt.                                     “pre-Hitt.”           “PAnat.”/PIE    nēpiš        < *nébh-es  <<                               *nébh-os          (<< *nébh-s)             nēpišaš    <                          *nébh-es-os <<   *nebh-és-os     (<< *nbh-és-s)


The noun aiš / išš-, ultimately reflecting PIE *h1eh3-es-, is far less clear regarding its prehistory.     -stem

In Hittite, only one -stem reflecting a PIE *h2-stem has been fully preserved, er- / ara- / ar- ‘line, boundary’, albeit that its paradigm shows much reshuffling of the original ablaut grades. The three stems show that we must assume that this word originally had a hysterodynamic inflection.


                            Hitt.                                                                         PIE           er           <              *h1ér-h2-o-s        <<           *h1ér-h2(-s)             aran          <              *h1r-h2-o-m         <<           *h1r-éh2-m             ar          <                                                          *h1r-h2-ós

    abl.                  araza        <                                                          *h1r-éh2-ti


The paradigms of other *h2-stems have been levelled out, due to which the direct reflex of *h2 was lost. This caused the eventual merging of these stems with the a-stem nouns, cf. e.g. āššā- ‘hearth’.

The outcome of the only known PIE proterodynamic *h2-stem noun, *gwen-h2- ‘woman’, is not fully clear because the reflex of this word in Hittite is written with a sumerogram only. See the discussion s.v. *kuan-.     *m-stem

The only *m-stem attested in Hittite, tēkan / takn- ‘earth’, does not show an -m- anymore. Yet its Anatolian cognates, CLuw. tiamm(i)- ‘earth’ and HLuw. ta-ka-mi-i ‘on the earth’, show that in PAnatolian the -m- must still have been present. Furthermore, CLuw. tiamm- < *dhh-ém- points to a hysterodynamic inflection. See s.v. for a detailed account of the prehistory of tēkan / takn-, which can be schematized thus:


                    Hitt.                                                                                 “PAnat.”/PIE   tēkan        <                                      *dégom   <<           *dhéh-m     tēkan        <                                      *dégom   <<           *dhh-ém-m     taknāš      <<       *takmāš       <      *dgmós   <<           *dhh-m-ós     n-stems

Until quite recently the noun šumanzan- ‘bulrush’ was regarded as denoting ‘cord, binding’ and therefore as cognate to Gr. ὑμήν ‘sinew’, on the basis of which “ šum(m)anza” was reconstructed as *suh1mḗn+s. When this form was compared to āraš ‘eagle’ < *h3ér-ōn+s, it was assumed that PIE forms in *-ōn lost their *-n in PIE already, whereas in *-ēn it was retained. Since “šum(m)anza” now has to be interpreted as the of a neuter noun šumanzan- ‘bulrush’ that has nothing to do with Gr. ὑμήν, the awkward split between *-ēn and *-ōn must be given up.

For commune n-stems, we can now distinguish two types, namely hysterodynamic n-stems with an original suffix vowel *-e- and hysterodynamic n-stems with an original suffix vowel *-o-, both yielding -aš in Examples: imen- ‘string, cord’, āran- ‘eagle’.


                    Hitt.                                                                     PIE   imāš           <                                                  *sh2i-mḗn-s    (<< *sh2éi-mn)     [i]šimenan     <          *sh2i-mén-om      <<       *sh2i-mén-m     unatt.                                                                   *sh2i-mn-ós   āraš              <                                                  *h3ér-ōn-s       (<< *h3ér-n)     āranan          <          *h3ér-on-om        <<       *h3r-ón-m       (<< *h3r-én-m)     āranaš          <          *h3ér-on-os         <<       *h3r-n-ós


The neuter n-stem nouns that are attested in Hittite usually seem to show a hysterodynamic inflection. This must be a rebuilding of an original proterodynamic inflection, however. Example: lāman / lamn- ‘name’.


                                Hitt.                                                                     PIE        lāman      <                                                          *h3néh3-mn                 lamnaš    <      *h3n(e)h3-mn-os              <<       *h3nh3-mén-s     r-stems

In Hittite, only two real r-stem nouns are attested. The oldest attestations of the first, keššar / kiššer- / kišr- ‘hand’, directly reflect a hysterodynamic paradigm:


                        Hitt.                                                                                             PIE       keššar              /kéSr/                   <                                              *hés-r         kiššeran           /k¸Séran/              <          *hs-ér-om      <<       *hs-ér-m         kišraš              /k¸Srás/                <                                              *hs-r-ós


The second one, ašter(a)- ‘star’, probably goes back to a hysterodynamic paradigm as well, but see s.v. for the problems regarding the establishment of its paradigm. Furthermore, it is not clear whether we should analyse the PIE stem as *h2s-ter- or *h2st-er-.


                        Hitt.                                                                 PIE        ašterza      <      *h2stḗr+s              <<           *h2stḗr        (<< *h2éstr?)         ašteran      <      *h2stérom            <<           *h2stér-m         aštiraš       <      *h2stéros(?)         <<           *h2str-ós     r/n-stems

Although in the other IE languages r/n-stems (including stems in -ur/-uen- and -mr/-men-) are rarely attested, they are fully alive in Hittite. We can distinguish two types of inflection, namely a static and a proterodynamic one. Examples: ur / mēun- ‘period, time’, paḫḫur / paḫḫuen- ‘fire’.


    static:                   Hitt.                                                                     PIE        ur             <                                                  *méih2-ur[10]                 unaš         <      *méih2-un-os       <<           *méih2-un-s


    proterodynamic:        paḫḫur            <                                                  *péh2-ur                 paḫḫuenaš      <      *peh2-uén-os       <<           *ph2-uén-s


It is often stated that ātar / itēn- ‘water’ reflects a static paradigm *uód-r, *uéd-n-s. As I have argued in Kloekhorst fthc.b, this is incorrect: ātar, itenaš must be regarded as an inner-Hittite remodelling of an originally proterodynamic paradigm *uód-r, *ud-én-s (cf. s.v. for details).

If per, parnaš ‘house’ really reflects *Pér-r, *Pr-n-ós (see s.v.), then we are here dealing with a hysterodynamic r/n-stem.   nt-stems

In Hittite, many nt-stem nouns are found, especially participles in -ant- and adjectives in -ant- ‘having x’. In these words, no traces of ablaut can be found anymore: we find a stem reflecting *CC-ént- throughout the paradigm. Nevertheless, the fact that in CLuwian the word for ‘Stormgod’ shows  a stem Taruant- besides Tarunt- < *trh2uent- / *trh2unt- indicates that at least in Proto-Anatolian ablaut was still present. Thus, we get the following picture:


                    Hitt.                                                         “PAnat.”                  PIE   uanza      <                                          *h2uh1-ént-s    <<   *h2ueh1-nt(-s)     *uantan  <      *h2uh1-ént-om     <<                                   *h2uh1-ént-m     uandaš    <      *h2uh1-ént-os      <<                                   *h2uh1-nt-ós   it-stems

Only two nouns in Hittite show a stem in -it-, namely militt- / malitt- ‘honey’ and šeppitt- ‘a kind of grain’. The latter shows the stem šeppitt- < *sép-it- throughout, but the former shows ablaut in the root: militt- < *mél-it- vs. malitt- < *ml-it-. It is remarkable that the suffix syllable does not show a full grade form anywhere in the IE languages, which would point to a hysterodynamic inflection: *ml-it-ós. Nevertheless, we would a priori expect a proterodynamic paradigm *mél-it, *ml-iét-s (or *ml-éit-s ?).


                                Hitt.                                                         PIE        milit             <                                          *mél-it                 milittaš        <      *mél-it-os    <<           *ml-it-ós      (<< *ml-iét-s ?)            malitti          <                                          *ml-it-éi       (<< *ml-iét-i ?)


2.1.3 Root nouns


In Hittite only a few root nouns are attested. Often, original root nouns are thematicized (compare e.g. pāt- / pat- ‘foot’ < *pód- / pd-, which eventually is altered to pata-), sometimes in pre-Hittite times already (compare e.g. uḫḫa- ‘grandfather’ < *h2uh2o- that in combination with CLuw. ūa- and Lyc. χuge- < *h2éuh2o- points to a PAnat. ablauting root noun *h2éuh2-s, *h2éuh2-m, *h2uh2-ós). We can distinguish static and mobile root nouns. Examples: itt- (MUKAM) ‘year’, kuan- / kun- ‘dog’, ker / kard(i)- ‘heart’.


    static:                   Hitt.                                                                     PIE               MUKAM-za      <                                                  *uót-s (?)                                                         MUKAM-an      <                                                  *ué/ót-m                                                          izza               <                                                  *uét-s                                                        ītti                  <                                                  *uét-i


    mobile:               kuaš              <          *uÖn-s          <<           *uÖn                 kuanan          <          *uón-om       <<           *uón-m                 kūnaš               <                                                  *un-ós        ker                   <                                                  *ḱḗr                 kardiaš          <                                                  *rd-i-ós

2.2 The Hittite system of personal pronouns

In order to etymologically describe the Hittite personal pronouns ‘I’, ‘you (sg.)’, ‘we’ and ‘you (pl.)’, it is important to first look at the systems of personal pronouns as reconstructed on the basis of the other IE languages (cf. Beekes 1995: 208).



nom.           *h1eH                                                      *tuH

acc.             *h1mé             * (?)                               *tué                *tuē (?)

gen.             *h1méne         *moi                                   *téue              *toi

 “obl.”        *h1me-                                                        *tue-



nom.           *uei                                                            *iuH

acc.             *ns                  *nōs (?)                              *us                  *uōs (?)

“obl.”         *ns-                *nos                                   *us-                *uos


If we compare acc.-obl. *tue to gen. *teue, we seem to be dealing with an ablaut between zero grade *tu-e vs. full grade *teu-e. When applied to gen. *h1men-e, we would expect an acc.-obl. *h1mn-e, with an -n-.[11] I believe that this -n- can explain the words for ‘I, me’ in Tocharian that have an otherwise unexplicable anlaut *ñ- < *nVfront: TochA nom.-obl.m. nā́º, f. ñuk, gen. ñi, TochB nom.-obl. ñā́ś, gen. ñi. This means that we should reconstruct the following basic system:


nom.           *h1eH                                  *tuH

obl.             *h1men- / *h1mn-                 *teu- / tu-

encl.            *moi                                       *toi


nom.           *uei                                        *iuH

obl.             *ns-                                        *us-

encl.            *nos                                       *uos



2.2.1 The Anatolian system: the singular forms


With the above system in mind, let us first look at the words for ‘I, me’ and ‘you (sg.)’. Of the Anatolian languages, the Hittite forms are best attested and probably reflect the most archaic system:


Hitt.            ‘I’                                                        ‘you (sg.)’

nom.           ú-uk                                                    zi-i-ik

acc.             am-mu-uk          =mu                        tu-uk               =t-ta / =d-du

gen.             am-me-el                                            tu-e-el

dat.             am-mu-uk          =mu                        tu-uk               =t-ta / =d-du

abl.             am-me-e-da-az                                  tu-e-da-az


The ending -ēl and the abl. ending -ēdaz are clearly of secondary origin, being taken over from the other pronouns. So the basic Hittite system is as follows:


nom.               ūk                                                                zīk

acc.-dat.        ammuk                                                       tuk

“obl.”             amm-                                                          tu-

encl.                =mu                                                            =tta / =ttu


The other Anatolian languages show the following forms:



nom.                   --                                                             ti-i

acc.-dat.            --                 =mu                                    tu-ú



nom.                   --                                                             ti-i

acc.-dat.            --                 =mu, =mi(?)                      --

gen.adj.              --                                                             tua/i-



nom.                   á-mu                                                       ti

acc.-dat.            á-mu           =mu                                    tu                    =tu

gen.adj.              á-ma/i-                                                   tu-wa/i-



nom.                   amu                                                        --

dat.                     amu                                                        --

gen.adj.              ẽmi-                                                        --



nom.                   ẽmu, emu, amu                                      --

dat.                     emu                                                         --

gen.adj.              ẽmi-                                                        --


Since in none of these languages word-final velars are attested, it is likely that these regularly were lost. As I have argued in Kloekhorst 2004: 39, HLuw. á-mu must be interpreted as /ʔmu/. Since the hieroglyphic script did not distinguish between single and geminate consonants, á-mu can safely be equated with Hitt. ammuk < PAnat. */ʔMug/, which in my view is the preform of Lyd. amu[12] and Lyc. emu[13] as well. It is clear that in these languages the acc.-dat. ‘me’ has ousted the original nom. ‘I’.


I therefore arrive at the following Proto-Anatolian reconstruction:


nom.                   *ʔú                                              *tÏ

acc.-dat.            *ʔ                                          *

“obl.”                 *ʔM-                                              *tu-

encl.                    *=mu                                             *=to(?) / *=tu


Note that I interpret Hitt. ú-uk as /ʔúg/[14] in analogy to e.g. e-eš ‘be!’ = /ʔés/ < *h1és, e-ep ‘take!’ = /ʔép/ < *h1ép, e-et ‘eat!’ /ʔéd/ < *h1éd, i-it ‘go!’ /ʔíd/ < *h1ídhi, etc. There is in my view no indication to assume that ú-uk would have a long ū (contra Melchert 1994a: 84).

If we compare PAnat. *ʔú ‘I’ to the form *h1eH, which is reconstructed on the basis of the other IE languages, we see that it contains an unexpected -u-. It is generally assumed that this -u- in one way or another derives from the paradigm of ‘you’.

Nevertheless, within the PAnat. paradigm of ‘you’, nom. *tÏg is remarkable in the sense that, when compared with *tuH as reconstructed on the basis of the other IE languages, it does not contain an -u-.

In order to explain this situation, several rather ad hoc attempts have been made. For instance, Georgiev (1978) assumes that Hitt. zīk (which he falsely reads as zek) reflects *te-ge, showing a development *t- > Hitt. z-. Apart from the fact that this does not take into account Pal. , CLuw.  and HLuw. ti ‘you’, a development *t- > Hitt. z- is falsified by e.g. tuekk- ‘body’ < *tuek-. Melchert (1994a: 84) assumes a development *tÜ > *tyÜ > *tyÏ > *tÏ but such a development is unparalleled in Anatolian. It is important to note that his argument that “the preform *tÜ is independently required in PA[nat.] as the source of the long ū of the first singular nominative *Ü1 seen in Hitt. ūg” is incorrect since the spelling ú-uk does not necessarily point to a long ū, but just stands for /ʔúg/.

In my view, the form *tÏg, of which the -g can easily be of a secondary origin and the -ī- must reflect *-ih1-,[15] cannot be explained from a pre-form *tuH in any phonetically regular way. Moreover, I do not see how this form could have been a secondary innovation on the basis of analogy: there is no -i- available in the personal pronouns on the basis of which an original *tuH could be altered to PAnat. *tih1. We therefore cannot conclude otherwise than that the Anatolian system *tih1, *tu- is more archaic than the system *tuH, tu- as reflected in the other IE languages[16] and that this latter system therefore must have been an innovation, namely taking over the obl.-stem tu- into the nominative[17] and altering *tih1 to *tuH (which therefore must be identified as *tuh1).[18]

This means that *ʔú ‘I’ cannot have got its -u- from ‘you’ (which was never *tuH, but always *tih1), and therefore must have been influenced by *ʔMúg (again nominative influenced by obl.)


We thus arrive at the following scenario:



nom.                   *h1éH                                          *tíh1

acc.                     *h1mn-                                           *tu-

obl.                     *h1mn-                                           *tu-


stage (1): the -u- of acc. *tu- is taken over to *h1mn-


nom.                   *h1éH                                          *tíh1

acc.                     *h1mnu-                                         *tu-

obl.                     *h1mn-                                           *tu-


stage (2): spread of -u- of acc. *h1mnu- to nom. *h1eH; assimilation of -mn- to -M-; loss of word-final laryngeal


nom.                   *h1ú                                             *tÏ

acc.                     *h1Mu-                                           *tu-

obl.                     *h1M-                                             *tu-


stage (3): either addition of the element *-ge in nom. and acc., or spread of word-final *- of *h1uǵ


nom.                   *h1ú                                             *tÏ-(e)

acc.                     *h1Mú-(e)                                   *tú-(e)

obl.                     *h1M-                                             *tu-


stage (4): loss of word-final -e



nom.                   *ʔú                                              *tÏ

acc.-dat.            *ʔ                                          *

obl.                     *ʔM-                                              *tu-


For a treatment of the enclitic forms, I refer to their own lemmata.


2.2.2 The Anatolian system: the plural forms

In Hitite, the plural forms are as follows:


nom.       ú-e-eš                                                     šu-me-eš

acc.         an-za-a-aš         =(n-)na-aš                 šu-ma-a-aš            =š-ma-aš

gen.         an-ze-el                                                  šu-me-en-za-an

dat.         an-za-a-aš         =(n-)na-aš                 šu-ma-a-aš            =š-ma-aš

abl.         an-ze-da-az                                           šu-me-e-da-az


Again, the endings -ēl, -edaz and -enzan are likely taken over from the other personal pronouns and are irrelevant. So the basic system is


nom.                   ēš                                                      šumeš

acc.-dat.            anzāš                                                  šumāš

obl.                     anz-                                                    šum-                              

encl.                    =(n)naš                                              =šmaš


In the other Anatolian languages, these pronouns are only scarcely attested:



nom.                   --                                                         --

acc.-dat.            an-za, a-an-za, an-za-aš                 u-za-aš



nom.                   a-zuʔ-za                                              u-zuʔ-sa, u-zuʔ-za

acc.-dat.            --                     =nz                             --

abl.-instr.           --                                                         u-za-ri+i

gen.adj.              a-za/i-                                                --


The plural forms of the first person (‘we’) are directly comparable to the PIE system, which had nom. *uei (with Hitt. ēš < *uei-s or *uei-es, compare Goth. weis < *uei-s), obl. *ns- and encl. *nos. The Hitt. acc.-dat. anzāš shows the ending -āš, which is the accented variant of the normal ending -aš. In HLuwian, where a-zuʔ-za and a-za/i- likely stand for /ants-/, the oblique stem *ns- was transferred to the nominative as well (compare ‘I’ above).

The interpretation of the plural forms of the second person (‘you’) is far less clear.[19] The Hitt. stem šum- and the Luwian stem uz- do not seem to fit into one PAnatalion pre-form easily. Often, šumeš has been interpreted as the metathesized outcome of *usme as visible in PIIr. *usmá and Gr. ὔμμε. The element *-me seems to be a Graeco-Indo-Iranian innovation, however, and does not occur in the Hittite paradigm of ‘we, us’ (where we would have expected *asme- or similar). Moreover, this assumption does not explain the enclitic =šmaš.

2.3 The Hittite verbal system

The Hittite verbal system knows many different inflection types, all with their characteristic forms. Each inflection type has its own prehistory. When we look at the Hittite texts diachronically, we see that this verbal system is in decline, however. Some inflection types are disappearing in the course of Hittite, whereas others are rapidly expanding. This creates a situation where a single verb sometimes can show forms that belong to a great number of different inflection classes. Since the historical linguist is mainly interested in the oldest linguistic situation as this provides the best information on the prehistory of a language, it is very important, in the case of the verbal system, to establish the oldest inflection type of each verb, and for each verb to provide a detailed description of the development it shows during the attested period. In this way we can determine which inflection types were productive, which inflection type was usually taken over by another inflection type, etc. With this knowledge, we will be able to gain a better insight into the possible origin of verbs that are not very well attested.

In order to do so, it is important to classify the different inflection types that are available in Hittite. Such a task was taken up by Oettinger in his 1979 masterpiece Die Stammbildung des hethitischen Verbums. Although this book is still very valuable today, I believe it is outdated in certain respects and some views presented in it cannot be upheld anymore. I therefore have chosen to set up my own classification that, although for the largest part based on Oettinger’s work, is in some respects different from it. In the following chapter I will present the classification of the Hittite verbal system that I have used throughout this book. Every inflection type is given its own code and is provided with the following information: original paradigm; prehistory; development during the attested Hittite texts; list of verbs that originally belonged to this type.


2.3.1 Basic division and sub-grouping

The first major division within the Hittite verbal system is between verbs that show an original active inflection and verbs that show an original middle inflection. This presents us with the first problem: some verbs show active as well as middle forms in the oldest texts already. Usually these verbs show a semantic difference between the active and the middle forms (e.g. eš-a(ri) ‘to seat oneself’ vs. eš-zi ‘to sit’), but sometimes such a difference is difficult to grasp (e.g. paš-a(ri) besides paš-i, both ‘to protect, to be loyal to’). Formally, these verbs sometimes use one stem (e.g. eš-a(ri) / aš- besides eš-zi / aš-), but sometimes the stems are different (e.g. uett-tta(ri) besides uttie/a-zi ‘to draw, to pull’, or nē-a(ri) besides nai-i / *ni- ‘to turn’). It must be noted that in the case of originally different stems for the active and the middle, these stems heavily influenced each other in the course of time (see their respective lemmata for the development in the latter two verbs). The active verbs are codified here with the Roman numbers I and II (see below for the difference between I and II), whereas the middle verbs are codified with the Roman number III.


2.3.2 The active verbs

Within the group of verbs that show an active inflection, the number of different inflection types is the largest. We first must make another basic division within the active verbs, namely in verbs that show the mi-inflection and verbs that show the i-inflection. The difference between these two is determined by their verbal endings. In the present tense, for instance, mi-inflected verbs have the endings -mi, -ši, -zi for the singular and -eni, -tteni, -anzi for the plural, whereas i-inflected verbs show -ḫḫi (-ḫḫe), -tti, -i (-e), -eni, -šteni, -anzi. It must be noted that sometimes an ending of the one type spreads at the cost of the ending of the other type (e.g. the mi-ending 2sg.pres.act. -ši is gradually being replaced by the i-ending -tti throughout Hittite, whereas the i-ending 2pl.pres.act. -šteni is being replaced by the mi-ending -tteni; see their respective lemmata for a full treatment of the verbal endings and their rise or fall within the Hittite period), but the basic division between mi-inflection types and i-inflection types is present up to the last Hittite texts. It is important to mention that a particular verbal suffix in principle always takes the same set of endings: e.g. -e/a- (= the -e/a-class) always uses mi-endings, but -ai-/-i- (= the dāi/tianzi-class) always i-endings. It is therefore not correct to say that, for instance, the verb nai-i / *ni- ‘to turn’, which was originally i-conjugated, is becoming mi-conjugated in younger Hittite. We should rather say that the stem nai-i / *ni- (inflecting according to the dāi/tianzi-class, which happens to be i-conjugating) from MH times onwards is being replaced by the stem nie/a-zi (according to the -e/a-class, which happens to be mi-conjugating). The inflection types that use mi-endings are codified with the Roman number I, whereas the i-conjugating inflection types are codified with II.     I = mi-conjugation

Within the mi-conjugated verbs, three types must be distinguished: (a) unextended mi-verbs that show ablaut; (b) mi-verbs that show no ablaut; (c) mi-verbs with a thematic suffix.   Ia = unextended ablauting mi-verbs

The ablauting mi-verbs go back to two PIE verbal categories, namely the root-present and the root-aorist.

For the root-present we can compare the verb ‘to be’.


                       PIE                         Gr.                          Skt.                         Hitt.

      1sg.       *h1és-mi                    εἰμί                         ásmi                        ēšmi

      2sg.       *h1és-si                     εἶ                             ási                           ēšši

      3sg.       *h1és-ti                      ἐστί                        ásti                          ēšzi

      1pl.        *h1s-mé(s)                ἐσμέν                     smás                       *ašuēni

      2pl.        *h1s-th1é                   ἐστέ                        sthá                         *aštēni

      3pl.        *h1s-énti                    εἰσί                         sánti                        ašanzi


For the root-aorist we can compare the verb ‘to put’. Since there is no trace of an augment in Hittite, I have cited here the injunctive forms as attested in Greek (with additional forms out of the paradigm of ἵστημι ‘to stand’) and in Sanskrit (with an additional example of var- ‘to cover’). The Hittite verb tē-zi in fact denotes ‘to speak’ (the plural forms are taken from compound verbs like peute-zi / peut- ‘to bring (away)’ and uate-zi / uat- ‘to bring (here)’).


                      PIE                         Gr.                          Skt.                         Hitt.

      1sg.       *dhéh1-m                  [στῆν]                    *dhā́m                   tēnun

      2sg.       *dhéh1-s                   [στῆς]                    dhā́s                       tēš

      3sg.       *dhéh1-t                   [στῆ]                      dhā́t                       tēt

      1pl.       *dhh1-mé                  θέμεν                     *dhā́ma                 °tumen

      2pl.       *dhh1-té                   θέτε                       *dhā́ta                  °tatten

      3pl.       *dhh1-ént                 θέσαν                    dhúr, [vran]          °tēr, °danzi


Since the formation of the PIE imperfect (ablaut *e/Ø, secondary endings) was identical to the formation of the root-aorist (also *e/Ø-ablaut and secondary endings), the two categories easily merged in Hittite. On the basis of the root-aorist a new inflection with primary endings (= addition of -i) was created which had presentic meaning and was formally identical to the root-present.

As we see, the PIE ablaut was *e (in the singular) vs. *Ø (in the plural). This PIE ablaut-type yielded six different ablaut-types in synchronic Hittite: e/Ø, a/Ø, e/a, a/a, e/¸, a/¸. The verb pai-zi / pai- ‘to go’ has its own class.   Ia1      mi-verbs with e/Ø-ablaut.

This class consists of verbs of the structure CueC-, Cmen- and of the structure *Ceh1-, to which the nasal-infix verbs of the structure *CR-ne-h1- belong as well. Note that in zinni-zi / zinn- and duarni-zi / duarn-, original *-ē- < *-eh1- has been raised to -i-. The verbs of this type most clearly reflect the PIE *e/Ø-ablaut.


aršanē-zi / aršan- ‘to be envious’ < *h1/3rs-ne-h1- / *h1/3rs-n-h1-; uek-zi / uk- ‘to slaughter’ < *h2ueg(h)- / *h2ug(h)-; uek-zi / uk- ‘to conjure’ < *h2uegh- / *h2ugh-; uiš-zi / uš- ‘to live’ *h2ues- / *h2us-; ulle-zi / ull- ‘to smash’ < *h2ul-ne-h1- / *h2ul-n-h1-; kue(n)-zi / kun- ‘to kill’ < *gwhen- / *gwhn-; kuer-zi / kur- ‘to cut’ < *kwer- / *kwr-; peute-zi / peut- ‘to bring (there)’ < *h1poi + *h2ou + *dheh1- / *dhh1-; pee-zi / pe- ‘to send’ < *h1poi + *h1ieh1- / *h1ih1-; šamen-zi / šamn- ‘to pass by’ < *smen- / *smn-; tē-zi ‘to state, to say’ < *dheh1-; duarni-zi / duarn- ‘to break’ < *dhur-ne-h1- / *dhur-n-h1-; ue-zi / u- ‘to send (here)’ < *h2ou + *h1ieh1- / *h1ih1-; uate-zi / uat- ‘to bring (here)’ < *h2ou + x + *dh(e)h1- / *dhh1-; erite-zi / erit- ‘to fear’ < x + *dheh1- / *dhh1-; ete-zi / et- ‘to build’ < x + *dheh1 / *dhh1-; zinni-zi / zinn- ‘to finish’ < *ti-ne-h1- / *ti-n-h1-.   Ia2      mi-verbs with a/Ø-ablaut.

This class consists of verbs in which the *e of the singular forms regularly is coloured to a by a neighbouring laryngeal or due to the development *eRCC > aRCC.


ā-zi / - ‘to believe’ < *h3eH- / *h3H- or *h2eh3- / *h2h3-; arna-zi / arn- ‘to sprinkle’ < *h2r-ne-h2/3- / *h2r-n-h2/3-; kallišš-zi / kališš- /kaL¸S- / kl¸S-/ ‘to call’ < *kelh1s- / *klh1s-, ma-zi / m- ‘to disappear(?)’ < *meh2/3- / *mh2/3-??   Ia3      mi-verbs with e/a-ablaut: the e/a-class.

This class consists of mi-verbs that show a synchronic ablaut e/a. It contains important verbs like eš-zi / aš- ‘to be’ and ed-zi / ad- ‘to eat’. Although it is clear that -e- of the strong stem directly reflects PIE *-e-[20], the origin of -a- of the weak stem has caused some debate.

In some of the verbs of this class, it is quite clear that the -a- as written in the weak stem is not phonologically real: ta-ra-an-zi ‘they speak’ reflects *tr-énti and therefore must represent phonological /trántsi/; ma-ra-an-du ‘they must disappear’ < *mr-éntu must be phonologically interpreted as /mrántu/. In other words, in verbs of the stucture *CeR- the PIE zero grade stem *CR- yielded Hitt. CR- that is spelled CaR-, with an empty -a-.

Regarding the interpretation of the a- as found in the weak stems of the verbs eš-zi / aš- ‘to be’, ed-zi / ad- ‘to eat’, eku-zi / aku- ‘to drink’ and epp-zi / app- ‘to seize’, matters are less clear. These verbs all go back to the structure *h1eC-, and it therefore seems obvious to assume that the weak stems aC- reflect *h1C-. Nevertheless, in view of the loss of initial *h1- before consonants in isolated forms (*h1C- > C-, cf. Kloekhorst 2006b), retention of *h1- in these verbs cannot be phonetically regular.[21] I therefore assume that at the time that *h1 was regularly lost in word-initial position before consonant, the ablaut of stems of the structure /ʔeC- / ʔC-/ still corresponded to the other ablauting mi-verbs, which showed an ablaut *Ce(R)C- / *C(R)C-. In order to avoid an alternation **/ʔeC- / C-/, which would have been fully aberrant in comparison to all other verbs that showed *Ce(R)C- / *C(R)C-, the initial /ʔ-/ of the full grade was restored.

The question now is: what is the relationship between */ʔC-/ and the spelling aC-? It has been claimed that aC- shows a vocalization of the initial *h1- to a-. There is, however, no proof anywhere in Hittite that *h1 would vocalize to -a- in any environment. As I have argued in Kloekhorst 2006b, we should rather assume that a spelling like a-ša-an-zi must be read as ’a-ša-an-zi[22] and therefore is comparable to e.g. ta-ra-an-zi = /trántsi/ in the sense that it stands for /ʔsántsi/ < *h1sénti, where -a- is nothing more than an empty vowel. The same goes for ša-ša-an-zi = /ssántsi/ < *ss-énti ‘they sleep’.

All in all, the synchronic ablaut e/a of class Ia3 is equivalent to the ablaut e/Ø of class Ia1 in the sense that the vowel -a- of the weak stem in the former type is just a graphic device to spell the initial consonant cluster /CC-/ and therefore is identical to phonological /Ø/.


eku-zi / aku- ‘to drink’ < *h1egwh- / *h1gwh-; epp-zi / app- ‘to seize’ < *h1ep- / *h1p-; ed-zi / ad- ‘to eat’ < *h1ed- / *h1d-; eš-zi / aš- ‘to be’ < *h1es- / *h1s-; eš-zi / aš- ‘to sit’ < *h1es- / *h1s-; mer-zi / mar- ‘to disappear’ < *mer- / *mr-; peš(š)-zi / *paš- ‘to rub’ < *pes- / *ps-; šeš-zi / šaš- ‘to sleep’ < *ses- / *ss-; ter-zi / tar- ‘to speak’ < *ter- / *tr-; e-zi / a- ‘to turn’ of secondary origin; en-zi / uan- ‘to copulate’ < *h1/3uenh1- / *h1/3unh1-.   Ia4      mi-verbs with a/a-“ablaut”.

This class consists of verbs of the structure *CeRC-. In the full grade forms, *-e- yielded Hitt. -a- because of the sound law *eRCC > Hitt. aRCC (note that all endings of the singular start in a consonant). In the zero grade forms, however, *CRC- yielded Hitt. /CRC-/, which is phonetically realized as [CəRC-]. In spelling, the full grade stem /CaRC-/ fell together with the zero grade stem /CRC-/: both are spelled CaRC-. This is the reason why these verbs are usually regarded as synchronically non-ablauting. Although I must admit that for the bulk of these verbs it cannot in principle be proven that in Hittite ablaut still existed synchronically, I do believe that in one case this is clear. The verb ārš-zi / arš- ‘to flow’ shows a synchronic ablaut between the strong stem ārš- (a-ar-aš-, a-ar-š°) and the weak stem arš- (ar-aš-, ar-š°). As I have argued s.v., this difference in spelling can only be explained by assuming that ārš- represents /ʔars-/ < *h1ers-, whereas arš- represents /ʔrs-/ < *h1rs-. This means that ārš- / arš- retained its ablaut throughout Hittite. I therefore think that it is likely that at least a part of the verbs that I have gathered under this class show ablaut in Hittite as well. This ablaut unfortunately cannot be seen in spelling, however.


ārku-zi / arku- ‘to chant, to intone’ < *h1erkw- / *h1rkw-; ārš-zi / arš- ‘to flow’ < *h1ers- / *h1rs-; ar(k)-zi / ar(k)- ‘to hold, to keep’ < *h2erk- / *h2rk-; ark-zi / ark- ‘to perish’ < *h3erg- / *h3rg-; arp-zi / arp- ‘to separate oneself and (re)associate oneself esewhere’ < *h3erbh- / *h3rbh-; išpart-zi / išpart- ‘to escape’ *sperdh- / *sprdh-; ištalk-zi / ištalk- ‘to make level, to flatten’ *stelgh- / *stlgh-; ištar(k)-zi / ištar(k)- ‘to ail, afflict’ < *sterḱ- / *strḱ-; karp-zi / karp- ‘to take away, to pick, to pluck’ < *kerp- / *krp-; karš-zi / karš- ‘to cut off’ < *kers- / *krs-; lāpp-zi / lapp- ‘to catch fire’ < *leh2p- / *lh2p-; par-zi / par- ‘to chase’ < *bherh2- / *bhrh2-; parš-zi / parš- ‘to flee’ < *bhers- / *bhrs-; šalk-zi / šalk- ‘to knead’ < *selK- / *slK-; ša(n)-zi / ša(n)- ‘to seek’ < *senh2- / *snh2-; ša(n)u-zi / ša(n)u- ‘to roast’ < *senh2u- / *senh2u-; taru-zi / taru- ‘to siege’ < *terh2u- / *trh2u-; tar(k)u-zi / tar(k)u- ‘to dance’ < *terkw- / *trkw-; al-zi / al- ‘to hit’ < *elh3- / *lh3-; alk-zi / alk- ‘?’ < *elK- / *lK-; arp-zi / arp- ‘to wash’ < *erp- / *rp-.    Ia5      mi-verbs with e/¸-ablaut.

This class consists of two verbs only, namely of ekk-zi ‘to wish’ and terepp-zi ‘to plough’. This class cannot be treated without referring to the other verbs in Hittite that show a vowel -e/i- in their weak stem, namely the verbs of class Ia6 (tamāšš-zi / tame/išš- ‘to (op)press’) and of class IIa3 (karāp-i / kare/ip- ‘to devour’, šarāp-i / šarip- ‘to sip’, ašāš-i / aše/iš- ‘to seat’ and amank-i / ame/ink- ‘to tie’). As I have argued in detail in Kloekhorst fthc.a, the -e/i- in the weak stem tame/išš- must be regarded as an anaptyctic vowel /¸/ that emerged in the cluster *dmh2s-. In my opinion, this vowel /¸/ is the one found in the weak stem forms of these verbs as well.

The case of terepp-zi must be taken together with karāp-i / kare/ip- and šarāp-i / šarip-. It is significant that these verbs are the only ones in Hittite that show a structure *CReC-. In principle, we would expect the zero grade form of these verbs, *CRC-, to regularly yield Hitt. [CəRC-] (compare class Ia4 and IIa2), spelled CaRC-. We therefore would expect that the ablauting pairs would be *CReC- / *CRC- > Hitt. CReC- / CaRC- when mi-conjugated, and *CRóC- / *CRC- > Hitt. CRāC- / CaRC- when i-conjugated. Note that in synchronic Hittite it looks as if the vowel is shifting place: strong stem CRVC- vs. weak stem CVRC-. Since such a Schwebe-ablaut is further absent in Hittite verbs, I believe that it was eliminated here. The zero grade stem CRC- secondarily received the anaptyctic vowel /¸/ on the place of the full grade vowel. In this way, mi-conjugating verbs of the structure *CRéC- / *CRC- yielded synchronic CReC- / CR¸C-, whereas i-conjugating verbs of the structure *CRóC- / *CRC- yielded synchronic CRāC- / CR¸C-. In both cases, the weak stem is spelled CRe/iC-.

With this scenario in mind, we can explain terepp-zi as an ablauting verb terepp-zi / tere/ipp- ‘to plough’, which stands for phonological /trep- / tr¸p-/, the ‘regular’ adaptation of PIE *trep- / *trp-.

The case of ekk-zi is slightly different. As I have argued s.v., here we are dealing with the principle that a PIE ablaut *e/oC- / *uC- is eliminated in Hittite. In analogy to the *- of the full grade, the zero grade *uC- is altered to *C-. This initial cluster then received an anaptyctic vowel, which is /¸/, when the following consonant is a stop. So I interpret ekzi / ekkanzi as /uéktsi / u¸kántsi/ < *eḱ-ti / *uḱ-énti.   Ia6      tamāšš-zi / tame/išš- ‘to (op)press’.

This verb constitutes a class of its own, since it shows a unique synchronic ā/¸-ablaut. As I have argued under its lemma, I regard tame/išš- as the regular outcome of the zero grade stem *dmh2s-, whereas tamāšš- replaced *tamaš-, which would have been the regular outcome of the full grade stem *dmeh2s-.   Ia7      pai-zi / pai- ‘to go’.

This verb, too, has its own class, as it shows a unique inflection. Although in the bulk of the attestations both the strong and the weak stem seems to be pai-, the oldest texts show a strong stem pai-. See s.v. for the discussion of the prehistory of this verb.    Ablaut pattern of the Ia-verbs

In all mi-verbs that show ablaut, this ablaut can be traced back to the PIE ablaut *e/Ø that is inherited from the PIE root present and root aorist. I have presented the distribution of these ablaut-vowels over the verbal paradigms in the following schemes, first giving the attested Hittite forms (the verb kue(n)-zi / kun- ‘to kill’ with additional forms from epp-zi / app- ‘to seize’, tē-zi ‘to state’, eš-zi / aš- ‘to be’ and i-zi ‘to go’), then an abstraction of these Hittite data, followed by the reconstructed PIE forms, exemplified by Sanskrit forms (the verbs han- / ghn- ‘to kill’ and as- / s- ‘to be’), using the present injunctive as the counterpart of the Hittite preterite. Forms between square brackets show the historically unexpected ablaut grade. Forms marked with are in fact unattested.



1sg.         ku-e-mi                             CéC-mi             *CéC-mi        hánmi            ásmi   

2sg.         ku-e-ši                              CéC-si              *CéC-si         há«si             ási

3sg.         ku-e-en-zi                         CéC-zi              *CéC-ti          hánti              ásti


1pl.          ap-pu-e-ni                        CC-éni           *CC-més(i)   hanma¡         smási

2pl.          ap-te-ni                             CC-téni             *CC-th1é       hathá             sthá

3pl.          ku-na-an-zi                       CC-ánzi            *CC-énti        ghnánti          sánti



1sg.         ku-e-nu-un                        CéC-un             *CéC-m         hánam

2sg.         te-e-eš                               CéC-s               *CéC-s          hán

3sg.         ku-en-ta, te-e-et                CéC-t                *CéC-t           hán


1pl.          [ku-e-u-en]                       CC-én[23]          *CC-mé         hanmá

2pl.          [ku-en-ten]                        *CC-tén[24]         *CC-té           hatá

3pl.          [ku-e-ner]                         *CC-ḗr[25]          *CC-ént         ghnan



1sg.         e-eš-li-it                            CéC-lit[26]          

2sg.         e-ep                                  CéC

                i-it                                     CC-t                 *CC-dhí         [jahí]             [edhi]

                ku-e-ni                              CéCi

3sg.         ku-en-du                           CéC-tu              *CéC-tu         hántu             ástu


2pl.          [ku-en-te-en], i-it-te-en     CC-tén[27]           *CC-té           hatá               stá

3pl.          ku-na-an-du                     CC-ándu          *CC-éntu       ghnantu         sántu


part.         ku-na-an-t-                       CC-ánt-            *CC-ént-       ghnánt-          sánt-

v.n.          e-šu-a-ar                        CéC-ar           *CéC-ur        --                  

v.n.          ap-pa-a-tar                      CC-ā́tar          

inf.I         e-ep-pu-a-an-zi              CéC-anzi       

inf.II        ap-pa-an-na                     CC-ā́nna         

impf.        ap-pí-iš-ke/a-                    CC-ské/á-         *CC-sé/ó-   e.g. uchá- < *h2us-sé/ó-    Ib = non-ablauting mi-verbs

Within this class we need to distinguish three types, which I have called Ib1, Ib2 and Ib3.   Ib1      unextended non-ablauting mi-verbs.

This class consists of mi-verbs that have neither a suffix (at least from a synchronic point of view) nor ablaut. This does not mean that they never showed ablaut however: in a few of these verbs it is clear that only one stem of an original ablauting pair was generalized throughout the paradigm: e.g. ane/išš-zi ‘to wipe’ originally belonged to an ablauting verb ānš-i / ane/išš- < *h2omh1s- / *h2mh1s-, of which eventually both stems formed their own paradigm (cf. ānš-i ‘to wipe’); kane/išš-zi ‘to recognize’ originally belonged to an ablauting verb *kanāš-zi / kane/išš- < *neh3s- / *nh3s-, of which the weak stem kane/išš- has been generalized; gulš-zi ‘to carve’ originally belonged to an ablauting verb *kuels- / kuls- < *kwels- / *kwls- in which the weak stem gulš- has been generalized.

In other verbs, the original full grade and zero grade (graphically) merged, e.g.: takš-zi ‘to devise, to unify’ may stand for /taks- / tks-/ < *teks- / *tks-, of which both the stem /taks-/ and /tks-/ are spelled takš-; ūpp-zi ‘to come up (of the sun)’ probably reflects *h1eup- / *h1up-, both of which yield Hitt. upp-; lukk-zi ‘to set fire to’ probably reflects *leuk- / *luk-, both of which yield Hitt. lukk-. Of again other verbs only a few forms are known, which means that it is possible that the second stem is by chance unattested: neku-zi < *negwh- ‘to become evening’ is attested in singular forms only, which is the reason that its weak stem counterpart is unattested (we would expect *ngwh- > Hitt. **naku-?).

Some of these verbs probably never showed ablaut, e.g. ištamašš-zi ‘to hear’, which clearly is of denominative origin (ištaman- ‘ear’ + -s-).


āšš-zi ‘to remain’; ane/išš-zi ‘to wipe’; aššikk-zi ‘to satiate oneself’; i-zi ‘to go’; ištamašš-zi ‘to hear’; kammarš-zi ‘to defecate’; kanen-zi ‘to bow down’; kane/išš-zi ‘to recognize’; kiš-zi ‘to comb’; kukkurš-zi ‘to mutilate’; kukuš-zi ‘to taste’; gulš-zi ‘to carve’; kuašš-zi ‘to kiss’; le/išš-zi ‘to pick, to gather’; lip(p)-zi ‘to lick up’; lukk-zi ‘to set fire to’; neku-zi ‘to become evening’; punušš-zi ‘to ask’; pūš-zi ‘to be eclipsed’; šāi-zi ‘to become sullen’; takš-zi ‘to devise, to unify’; tarupp-zi ‘to collect’; ūpp-zi ‘to come up (of the sun)’; atku-zi ‘to jump’.    Ib2      mi-verbs in -ē-, -ēšš- and -nu-.

This class consists of verbs that show non-ablauting athematic suffixes, namely the ‘stative / fientive’ suffix -ē- < *-éh1-,[28] the ‘fientive’ suffix -ēšš- < *-éh1-sh1- and the ‘causative’ suffix -nu- < *-n(e)u-.[29] For a treatment of the suffix -ē-, cf. Watkins 1973. For a treatment of -ēšš- and -nu-, see s.v.v..


Verbs with -ē-zi:

aššuē-zi ‘to become king’; lalukkē-zi ‘to be or become luminous’; maršē-zi ‘to become corrupt’; miauntē-zi ‘to become old’; nakkē-zi ‘to be honoured’; paprē-zi ‘to be proven guilty’; parkuē-zi ‘to be pure’; šullē-zi ‘to become arrogant’.


Verbs with -ēšš-zi:

aēšš-zi ‘to become hot (?)’; alpuēšš-zi ‘to be sharp’; araēšš-zi ‘to become free’; ašiantēšš-zi ‘to become poor’; ikunēšš-zi ‘to become cold’; anallēšš-zi ‘to become a blood-shedder’; annitalanēšš-zi ‘to become legal adversaries’; appinēšš-zi ‘to become rich’; arkiēšš-zi ‘to become white’; aštalēšš-zi ‘to become brave’; atēšš-zi ‘to become dry’; atkuēšš-zi ‘to become tight’; atukēšš-zi ‘to become terrible’; innaraēšš-zi ‘to become strong’; aššarēšš-zi ‘to become a lord(?)’; idalaēšš-zi ‘to become evil’; karpēšš-zi ‘to become angry’; kartimmiēšš-zi ‘to become angry’; kunnēšš-zi ‘to turn out right’; *lazziēšš-zi ‘to become well’; maantēšš-zi ‘to become a young man’; makkēšš-zi ‘to become numerous’; maleškuēšš-zi ‘to become weak’; maninkuēšš-zi ‘to be short’; marlēšš-zi ‘to become foolish’; maršēšš-zi ‘to become desecrated’; (LÚ)miauntēšš-zi ‘to become an old man’; miēšš-zi ‘to grow; to be born’; mīēšš-zi ‘to be mild’; milit(t)ēšš-zi ‘to be sweet’; mišriēšš-zi ‘to become bright (of the moon)’; nakkēšš-zi ‘to become important’; nakkuššēšš-zi ‘to be(come) a scape-goat’; palēšš-zi ‘to become wide or broad’; pankuēšš-zi ‘to become plentiful(?)’; paprēšš-zi ‘to be found guilty’; parkēšš-zi ‘to become tall’; parkuēšš-zi ‘to become tall’; parkuēšš-zi ‘to be(come) pure’; šaknēšš-zi ‘to be(come) impure’; šallēšš-zi ‘to become large’; šannapilēšš-zi ‘to be emptied’; šan(i)ezziēšš-zi ‘to become pleasant’; šaraziēšš-zi ‘to prevail’; šarkuēšš-zi ‘to become mighty’; šullēšš-zi ‘to become arrogant’; šuppiēšš-zi ‘to become purified’; talliēšš-zi ‘to be pleasant(?)’; dalukēšš-zi ‘to become long’; tameummēšš-zi ‘to become different’; tampuēšš-zi ‘to become blunt’; dankuēšš-zi ‘to become black’; taruēšš-zi ‘to become powerful’; taruilēšš-zi ‘to become powerful’; daššēšš-zi ‘to become heavy’; tekkuššēšš-zi ‘to become visible’; tepaēšš-zi ‘to become little’; tepšaēšš-zi ‘to become tepšu-’; tukkēšš-zi ‘to become important’; ulēšš-zi ‘to hide’; antēšš-zi ‘to become glowing’; ¾arḫuēšš-zi ‘?’; arkēšš-zi ‘to grow fat’; eritēšš-zi ‘to be frightened’; zalukēšš-zi ‘to take long’.


Verbs with -nu-zi:

annanu-zi ‘to train’; arnu-zi ‘to make go, to transport’; aršanu-zi ‘to make flow’; aše/išanu-zi ‘to seat, to settle’; āššianu-zi ‘to make beloved (?)’; aš(ša)nu-zi ‘to take care of’; enu-zi ‘?’; ēšarnu-zi ‘to make bloody’; edrianu-zi ‘to feed (?)’; alinu-zi ‘to make kneel’; arranu-zi ‘to grind’; arknu-zi ‘to ruin’; argnu-zi ‘to make white’; arnu-zi ‘to spray’; aššik(ka)nu-zi ‘to satiate’; aš(ša)nu-zi ‘to bring to birth’; atnu-zi ‘to cause to dry up’; atganu-zi ‘to make tight’; atkešnu-zi ‘to make tight’; atuganu-zi ‘to terrify’; inganu-zi ‘to make bow’; uinu-zi ‘to make run’, uišnu-zi ‘to make recover, to rescue’; untarnu-zi ‘to grunt’; unu-zi ‘to make run’; ušnu-zi ‘to make recover, to rescue’; inu-zi ‘to make hot’; arnu-zi ‘to make bloody’; išparnu-zi ‘to spread’; išpianu-zi ‘to saturate’; ištantanu-zi ‘to delay’; ištappinu-zi ‘to shut’; kanganu-zi ‘to have (something) weighed’; kari(a)nu-zi ‘to silence’; karpanu-zi ‘to pick up’; karšnu-zi ‘to cut off; to cancel’; kardimi(a)nu-zi ‘to make angry’; karūššianu-zi ‘to silence’; genušrinu-zi ‘to make kneel’; kīnu-zi ‘to open up’; kiš(ša)nu-zi ‘?’; kištanu-zi ‘to extinguish’; kuašnu-zi ‘to make kiss’; laknu-zi ‘to fell, to knock over’; laluk(k)e/išnu-zi ‘to illuminate’; lap(pa)nu-zi ‘to kindle’; linganu-zi ‘to make swear’; lukkanu-zi ‘to make it light (?)’; maknu-zi ‘to increase’; mališkunu-zi ‘to make weak’; maninkuanu-zi ‘to bring near (?)’; marnu-zi ‘to cause to disappear’; maršanu-zi ‘to desecrate’; mem(i)anu-zi ‘to make (someone) talk’; mernu-zi ‘to cause to disappear’; mianu-zi ‘to make (branches) fruit-bearing’; mienu-zi ‘?’; mīnu-zi ‘to make mild’; našarnu-zi ‘to make afraid’; ninganu-zi ‘to drench’; nu(n)tarnu-zi ‘to hurry’; pašnu-zi ‘to protect’; paknu-zi ‘to defame’; palanu-zi ‘to broaden’; paranu-zi ‘to make gallop’; parknu-zi ‘to make high’; parkianu-zi ‘to raise’; parkunu-zi ‘to cleanse’; paršnu-zi ‘to make flee’; paršnu-zi ‘to break up’; pattinu-zi ‘to run off with’; pirnu-zi ‘to embezzle (?)’; pukkanu-zi, pukkunu-zi ‘to cause (someone) to be hated’; šaku(a)ntarianu-zi ‘to neglect’; šallanu-zi ‘to melt down’; šallanu-zi ‘to raise, to bring up’; šamenu-zi ‘to bypass’; šaminu-zi ‘to burn (something)’; šamešanu-zi ‘to burn (something) into smoke’; šaš(ša)nu-zi ‘to make sleep (with someone)’; daluknu-zi ‘to lengthen’; tamenganu-zi ‘to make attach (?)’; dammešanu-zi ‘to make punish’; taninu-zi ‘to install’; dankuanu-zi ‘to make black’; danku(a)nu-zi ‘to make black’; darianu-zi ‘to make tired’; taruppianu-zi ‘to bring together’; daš(ša)nu-zi ‘to make strong’; tekkuš(ša)nu-zi ‘to reveal’; tepnu-zi ‘to diminish’; tepšanu-zi ‘to make tepšu-’; titnu-zi ‘to install’; dušganu-zi ‘to make happy’; unu-zi ‘to adorn, to decorate’; anu-zi ‘to make turn’; aggašnu-zi ‘to leave out’; akšianu-zi ‘to deny a person of something’; allanu-zi ‘to erase (?)’; alganu-zi ‘?’; aru(a)nu-zi ‘to plant densely’; argnu-zi ‘to make fat’; arnu-zi ‘to set fire to’; arš(ia)nu-zi ‘to appease (trans.)’; aštanu-zi ‘to regard as an offense’; atkunu-zi ‘to make jump’; eritanu-zi, eritenu-zi ‘to scare’; zanu-zi ‘to cook (trans.)’; zaluknu-zi ‘to postpone’; zapnu-zi ‘to sprinkle’; zinu-zi, zainu-zi ‘to make cross’.  Ib3      non-ablauting mi-verbs with n/Ø-alteration.

This class consists of mi-verbs of which the stems end in °V(n)C-. In the oldest texts, these verbs show a clear distribution between °VnCV and °VCC, i.e. the nasal is lost before two or more consonants. This distribution is nicely preserved in the following paradigm (examples from arni(n)k-, supplemented by forms from šarni(n)k- and ištarni(n)k-):


                  pres.                     pret.                        imp.


1sg.         arnikmi              arninkun                                             

2sg.         arnikši               arnikta                  arnik                     part.      arninkant-

3sg.         arnikzi               arnikta                  arnikdu                 v.n.       arninkuar


1pl.          arninkueni         ištarninkuen                                           inf.I      arninkuanzi

2pl.          arnikteni            --                             arnikten                impf.    arninkiške/a-

3pl.          šarninkanzi          arninker               arninkandu          


I regard the loss of nasal here as a phonetic development[30] that probably took place in recent pre-Hittite times. In young Hittite we come accross forms in which the original distribution between °VnC-V and °VC-C has been given up, probably due to inner-paradigmatic analogy (e.g. lingazi instead of original likzi, linkatta instead of original likta, etc.). For the origin of the nasal-infixed verbs (the verbs in -ni(n)k-zi and tame(n)k-zi ), cf. ḱ 2.3.4 below.

arni(n)k-zi ‘to make disappear’, i(n)k-zi ‘to offer; to bow’, uni(n)k-zi ‘to bash’, išta(n)-zi ‘to taste’, ištarni(n)k-zi ‘to afflict’, li(n)k-zi ‘to swear’, nini(n)k-zi ‘to mobilize’, ni(n)k-zi ‘to quench one’s thirst’, ša(n)-zi ‘to seek’, ša(n)u-zi ‘to roast’, šarni(n)k-zi ‘to compensate’, tame(n)k-zi ‘to attach’, ū(n)-zi ‘to suck (?)’.   Ic = mi-verbs with a thematic suffix[31]

The third group of mi-verbs consists of verbs that show a thematic suffix. I distinguish six types, namely Ic1: mi-verbs in -e/a-; Ic2: mi-verbs in -ae-/-ā-; Ic3: mi-verbs in e/a-; Ic4: mi-verbs in -ue/a-; Ic5: ašše/a-zi; Ic6: mi-verbs in -ške/a-. The first five groups belong together in the sense that they all ultimately reflect the PIE verbal suffix *-e/o-.   Ic1      mi-verbs in -e/a-: the -e/a-class.

This class is one of the most productive verbal classes in Hittite. In NS texts, almost all verbs show at least a few forms that are inflected according to the -e/a-class. The suffix -e/a- clearly goes back to the PIE verbal suffix *-e/o-, which is attested in the IE languages on a large scale. On the one hand, it was a productive suffix to mark imperfectivity, e.g. pres. *gwm-é-ti ‘he is going’ (Gr. βαίνω, Lat. veniō) next to aor. *gwém-t ‘he went’ (within Hittite, this distribution is still clear in e.g. karp(ie/a)-zi, karš(ie/a)-zi, etc.). On the other, it is used to form denominative verbs, e.g. *h3n(e)h3-mn-e/o- ‘to name’ (Gr. ὀνομαίνω, Goth. namnjan, ModDu. noemen, Hitt. lamnie/a-).

The Hittite verbs that belong to the -e/a-class are divided in three groups: (A) original -e/a-inflecting verbs (i.e. attested in OS and MS texts already); (B) denominative -e/a-inflecting verbs; and (C) secondarily -e/a-inflecting verbs (their original stem between brackets).


A: anie/a-zi ‘to work, to carry out’; arie/a-zi ‘to consult an oracle’; arrie/a-zi ‘to be awake’; arkie/a-zi ‘to get lost’; azzie/a-zi ‘to pierce, to prick’; uttie/a-zi ‘to draw, to pull’; e/a-zi ‘to do, to make’; imie/a-zi ‘to mingle’; iškie/a-zi ‘to smear’; karie/a-zi ‘to cover, to hide’; karpie/a-zi ‘to take away, to pick, to pluck’; karšie/a-zi ‘to cut (off)’; lukkie/a-zi ‘to set fire to’; markie/a-zi ‘to disapprove of’; mumie/a-zi ‘to crumble (?)’; parkie/a-zi ‘to raise, to lift’; peššie/a-zi ‘to throw away’; šie/a-zi ‘to shoot’; šapašie/a-zi ‘to scout’; šarie/a-zi ‘to embroider’;

šarie/a-zi ‘to attack’; šūnie/a-zi ‘to dip’; tallie/a-zi ‘to pray’; tekkuššie/a-zi ‘to show’; tie/a-zi ‘to step’; tie/a-zi ‘to bind(?)’; tūrie/a-zi ‘to harness’; tuškie/a-zi ‘to be happy’; ūššie/a-zi ‘to draw open (of curtains)’; emie/a-zi ‘to find’; erie/a-zi ‘to call, to name’.


B: appat(a)rie/a-zi ‘to take in pledge’; armizzie/a-zi ‘to bridge over’; armanie/a-zi ‘to become ill’; aršie/a-zi ‘to plant’; edrie/a-zi ‘to feed’; a(ḫa)rie/a-zi ‘to rake’; alie/a-zi ‘to watch over’; ališšie/a-zi ‘to encase’; antie/a-zi ‘to face’; ap(pa)rie/a-zi ‘to trade’; arnamnie/a-zi ‘to churn’; *aššuezzie/a-zi ‘to become king’; ē(a)anie/a-zi ‘to rain’; ulālie/a-zi ‘to enwrap’; ululie/a-zi ‘to embrace’; untarie/a-zi ‘to fart’; ezzie/a-zi ‘to dominate’; uzzie/a-zi ‘to gird’; ištappullie/a-zi ‘to use as a stopper’; kalelie/a-zi ‘to tie up’; kalutie/a-zi ‘to treat as a group’; kanuššarie/a-zi ‘to kneel’; kardimie/a-zi ‘to be angry’; genuššarie/a-zi ‘to kneel’; gimanie/a-zi ‘to spend the winter’; gimmantarie/a-zi ‘to spend the winter’; kurkurie/a-zi ‘to scare’; kūrurie/a-zi ‘to be hostile’; kuššanie/a-zi ‘to employ’; kuttanie/a-zi ‘to exert force’; laḫḫie/a-zi ‘to go on an expedition’; lam(ma)nie/a-zi ‘to name’; lāzzie/a-zi ‘to set straight’; našarie/a-zi ‘to be afraid’; nekumandarie/a-zi ‘to undress’; nuntar(r)ie/a-zi ‘to hasten’; palašie/a-zi ‘to cover’; parkuantarie/a-zi ‘to become pure (?)’; patallie/a-zi ‘to tie feet, to fetter’; pittulie/a-zi ‘to be anxious’; puntarie/a-zi ‘to be obstinate’; putal(l)ie/a-zi ‘to tie together’; šākie/a-zi ‘to give a sign’; šaknie/a-zi ‘to anoint’; šakunie/a-zi ‘to well up’; šakkurie/a-zi ‘to overpower’; šakuantarie/a-zi ‘to stay, to remain’; šamešie/a-zi ‘to burn for fumigation’; šēurie/a-zi ‘to urinate’; šešarie/a-zi ‘to sieve’; šimišie/a-zi ‘to burn for fumigation’; šišš(i)urie/a-zi ‘to irrigate’; šittarie/a-zi ‘to seal’; šiunie/a-zi ‘?’; šupparie/a-zi ‘to sleep’; tiantie/a-zi ‘to set up’; takšatnie/a-zi ‘to level’; tarkuallie/a-zi ‘to look angrily’; terippie/a-zi ‘to plough’; tešanie/a-zi ‘to appear in a dream’; tuzzie/a-zi ‘to encamp’; ušnie/a-zi ‘to put up for sale’; uddanallie/a-zi ‘to speak about’; uddanie/a-zi ‘to speak about’; akkarie/a-zi ‘to rebel against’; išurie/a-zi ‘to press’; zaḫḫie/a-zi ‘to battle’; zazaḫḫie/a-zi ‘to battle fiercely’.


C: alalamnie/a-zi ‘to cry aloud’; allanie/a-zi ‘to sweat’; ānšie/a-zi ‘to wipe’ (ānš-i); appie/a-zi ‘to be finished’ (āppa-i / āppi-); ārrie/a-zi ‘to wash’ (ārr-i / arr-); aršanie/a-zi ‘to be envious’ (aršanē-zi / aršan-); ar(aš)šie/a-zi ‘to flow’ (ārš-zi / arš-); alzie/a-zi ‘to cry out’ (alzai-i / alzi-); ānie/a-zi ‘to draw (water)’ (ān-i / an-); arnie/a-zi ‘to sprinkle’ (arna-zi / arn-); arpie/a-zi ‘to change allegiance’ (arp-zi); aršie/a-zi ‘to till (the soil)’ (ārš-i); ullie/a-zi ‘to smash’ (ulle-zi / ull-); amie/a-zi ‘to sing’ (amai-i / ami-); ie/a-zi ‘to bind’ (ai-i / i-); išparrie/a-zi ‘to spread out’ (išpār-i / išpar-); išpartie/a-zi ‘to escape’ (išpart-zi); išpie/a-zi ‘to be satiated’ (išpai-i / išpi-); ištalkie/a-zi ‘to level’ (ištalk-zi); ištarkie/a-zi ‘to ail’ (ištar(k)-zi); kanenie/a-zi ‘to bow down’ (kanen-zi); mallie/a-zi ‘to mill, to grind’ (malla-i / mall-); memie/a-zi ‘to speak’ (mēma-i / mēmi-); nana(n)kuššie/a-zi ‘to be(come) dark’ (nana(n)kušš-zi); paškie/a-zi ‘to plant’ (pāšk-i / pašk-); pee/a-zi ‘to send’ (pee-zi / pe-); pennie/a-zi ‘to drive (there)’ (penna-i / penni-); pie/a-zi ‘to give’ (pai-i / pi-); šie/a-zi ‘to impress’ (šai-i / ši-); šalikie/a-zi ‘to have contact with’ (šalÌk-a(ri)); šalkie/a-zi ‘to knead’ (šalk-zi); šamnie/a-zi ‘to create’ (šamnae-zi); šartie/a-zi ‘to rub’ (šarta-i / šart-); šullie/a-zi ‘to be arrogant’ (šullē-zi); šunnie/a-zi ‘to fill’ (šunna-i / šunn-); šuppie/a-zi ‘to sleep’ (šupp-(tt)a(ri)); tie/a-zi ‘to put, to place’ (dai-i / ti-); taištie/a-zi ‘to load’ (tāišta-i / tāišti-); dālie/a-zi ‘to leave’ (dāla-i / dāli-); taparie/a-zi ‘to rule’ (CLuw. tapar-); tarie/a-zi ‘to exert oneself’ (tarai-i / tari-); taruppie/a-zi ‘to collect’ (tarupp-zi); tetie/a-zi ‘to thunder’ (tit-a); tutuḫḫie/a-zi ‘to brandish (?)’; duarnie/a-zi ‘to break’ (duarni-zi / duarn-); ue/a-zi ‘to send here’ (ue-zi / u-); ie/a-zi ‘to cry out’ (ai-i / i-); ak(ki)šie/a-zi ‘to be lacking’; alkie/a-zi ‘?’; appie/a-zi ‘to bark’; arpie/a-zi ‘to bathe’ (arp-zi); aššie/a-zi and eššie/a-zi ‘to wear’ (ašše/a-zi).


The distribution of the ablaut vowels *e/o in the suffix *-e/o- changes throughout the Hittite period, as we can see by the following overview of attested spellings:


                  PIE              OS                              MH/MS                                 NH


1sg.         *-ó              -a-mi, -i-e-mi           -a-mi                                 -a-mi

2sg.         *-é-             -i-e-ši                          -a-ši                                  -a-ši

3sg.         *-é-             -i-e-ez-zi                     -i-e-ez-zi / -a-az-zi            -a-az-zi


1pl.          *-ó-             -a-u-e-ni[32]                -a-u-e-ni                            -a-u-e-ni

2pl.          *-é-             --                                -i-et-ta-ni, -a-at-te-ni         -a-at-te-ni

3pl.          *-ó-             -a-an-zi                    -a-an-zi                             -a-an-zi



1sg.         *-ó-             -i-e-nu-un[33]                -a-nu-un                            -a-nu-un        

2sg.         *-é-             --[34]                              -a-aš                                 -a-aš / -a-at

3sg.         *-é-             -i-e-et                          -i-e-et / -a-at                      -a-at


1pl.          *-ó-             -a-u-en                     -a-u-en                              -a-u-en

2pl.          *-é-             --[35]                              -a-at-ten                            -a-at-ten

3pl.                                -i-er                            -i-e-er, -a-er                      -i-er, -a-er



2sg.                               --                                -i[36], -a                               -a

3sg.                               --                                -i-e-ed-du / -a-ad-du         -a-ad-du


2pl.                                --                                -a-at-ten                            -a-at-ten

3pl.                                --                                -a-an-du                            -a-an-du


part.                               --                                -a-an-t-

v.n.                                --                                -a-u-a-ar                        -a-u-a-ar

inf.I                               --                                -a-u-a-an-zi                    -a-u-a-an-zi

inf.II                              --                                --                                         -a-an-na     Ic2      mi-verbs in -ae-/-ā-: the atrae-class

This class is very large and very productive: in NS texts virtually every verb shows forms that are inflected according to the atrae-class. Usually, the verbs that originally belong to this class are derivatives in *-e/o- of o-stem nouns, e.g. atrae-zi ‘to write’ < *h2etro- + -e/o-.

These verbs show the following inflection:


               PIE                 OS                         MH/MS                         NH


1sg.      *-Co-ó-          °Ca(-a)-e-mi          °Ca(-a)-mi                      °Ca(-a)-mi                 

2sg.      *-Co-é-           --                            °Ca(-a)-ši                        °Ca(-a)-ši                   

3sg.      *-Co-é-           °Ca(-a)-ez-zi          °Ca(-a)-ez-zi                   °Ca(-a)-ez-zi              


1pl.       *-Co-ó-          --                            --                                     °Ca(-a)-u-e-ni            

2pl.       *-Co-é-           --                            °Ca(-a)-at-te-ni               °Ca-at-te-ni                

3pl.       *-Co-ó-          °Ca-an-zi               °Ca(-a)-an-zi                  °Ca(-a)-an-zi             



1sg.      *-Co-ó-          --                            °Ca(-a)-nu-un                 °Ca(-a)-nu-un

2sg.      *-Co-é-           --                            °Ca-a-eš                          °Ca-a-eš, °Ca(-a)-iš

3sg.      *-Co-é-           °Ca(-a)-et               °Ca-a-et                          °Ca(-a)-et


1pl.       *-Co-ó-          --                            °Ca(-a)-u-en                   °Ca(-a)-u-en

2pl.       *-Co-é-           --                            °Ca(-a)-at-ten                 °Ca-at-ten

3pl.                                --                            °Ca(-a)-er                       °Ca(-a)-er

It is not fully clear how these forms were accented: *-ó-e/o- or *-o-é/ó-. The forms with *-o-o- do not give information because both *-ó-o- and *-o-ó- would yield Hitt. -ā-. So we should look at the forms that show *-o-e-. On the one hand, the frequent plene spelling of -a- in e.g. 3sg.pres.act. °Ca-a-IZ-zi and 2sg.pret.act. °Ca-a-eš seems to point to *-ó-e-. On the other hand, the fact that the thematic vowel *-e- remains -e- up to the NH period (cf. spellings like 1sg.pres.act. °Ca-e-mi, 2sg.pret.act. °Ca-a-eš, which indicate that °Ca(-a)-IZ-zi and °Ca(-a)-IT have to be read as °Ca(-a)-ez-zi and °Ca(-a)-et, at least in the OH and MH period), seems to point to *-o-é-. Since the long -ā- can also be explained as the result of compensatory lenghtening after loss of intervocalic *i (cf., we have to assume that the accentuation in fact was *-o-é/ó-. So *CoóCi > */CaiáCi/[37] > */CaáCi/ > Hitt. /Cā́Ci/ and *CoéCi > */CaiéCi/ > OH/MH */CāéCi/. The NH forms that are spelled °Ca-a-iš may show that OH/MH /CāéC/ eventually yielded a diphthong /Cā́iC/ in NH times. Throughout this book, forms with the spelling °Ca(-a)-IZ-zi and °Ca(-a)-IT therefore have been transliterated as °Ca(-a)-ez-zi and °Ca(-a)-et in OS and MS texts, but as °Ca(-a)-iz-zi and °Ca(-a)-it in NS texts.


Verbs that originally belong to this class:

āppalae-zi ‘to entrap’; arae-zi ‘to rein in’; arkuae-zi ‘to pray’; arae-zi ‘to go down the line’; armae-zi ‘to be pregnant’; armualae-zi ‘to shine (of the moon)’; aruae-zi ‘to bow’; ašandulae-zi ‘to be on garrison duty’; egae-zi ‘to cool down’; ēšarnumae-zi ‘to make bloody’; aluganae-zi ‘to bring news’; antae-zi ‘to arrange (together)’; andandae-zi ‘(+ parā) to show providence’; antiae-zi ‘to support’; appešnae-zi ‘(+ ara) to dismember’; apae-zi ‘to wet’; ap(pa)rae-zi; appirae-zi ‘to trade’; arnae-zi ‘to stir, to churn’; arpae-zi ‘to heap up’; aru(a)nae-zi ‘to dawn’; *aššueznae-zi ‘to be king’; attaluae-zi ‘to bolt’; attarae-zi ‘to prick’; atrae-zi ‘to write’; ilae-zi ‘to have a halo’; uantalae-zi ‘to spare’; ultalae-zi ‘to spare’; igae-zi ‘to cool down’; irae-zi ‘to go down the line’; arnumae-zi ‘to make bloody’; galaktarae-zi ‘to make drowsy’; kappae-zi ‘to diminish’; kappilae-zi ‘to pick a fight’; kartae-zi ‘to cut off’; kattuae-zi ‘to be aggrieved’; genzuae-zi ‘to treat gently’; kinae-zi ‘to (as)sort’; TÚGkurešnae-zi ‘to provide with head-dress’; kururae-zi ‘to be hostile’; kutruae-zi ‘to bear witness’; lappinae-zi ‘to insert a wick (?)’; lelae-zi ‘to conciliate’; leluntae-zi ‘to use a pitcher’; lipae-zi ‘to lick up’; līšae-zi ‘?’; mÀlae-zi ‘to approve of’; maniaḫḫae-zi ‘to be in charge of’; markištae-zi ‘to take someone by surprise’; marlae-zi ‘to become mad’; marzae-zi ‘to crumble (?)’; mītae-zi ‘to tie with red wool (?)’; mūgae-zi ‘to invoke’; munnae-zi ‘to hide’; mūtae-zi ‘to root’; nekumantae-zi ‘to undress oneself’; palašae-zi ‘to cover’; palae-zi ‘to cry out’; palzaae-zi ‘to stretch out’; paršae-zi ‘to crumble’; paršnae-zi ‘to squat (?), to crouch (?)’; paršul(l)ae-zi ‘to crumble’; pašiae-zi ‘to rub’; patalae-zi ‘to fetter’; peanae-zi ‘to reward (someone)’; pittae-zi ‘to bring; to carry’; pittalae-zi ‘to abandon’; ae-zi ‘to pound’; puruttae-zi ‘to cover with mud’; šaešnae-zi ‘to fortify (?)’; šallakartae-zi ‘to offend someone through arrogance’; šamnae-zi ‘to create’; šarlae-zi ‘to exelt; to praise’; šaruae-zi ‘to loot’; šaitištae-zi ‘to wean’; šēurae-zi ‘to urinate’; šiptamae-zi ‘to seven (?)’; takšulae-zi ‘to agree’; damme/išae-zi ‘to damage’; tarmae-zi ‘to nail’; taruae-zi ‘to fix (magically)’; tattarae-zi ‘?’; tuḫḫae-zi ‘to produce smoke’; tuušiae-zi ‘to await’; ulae-zi ‘to hide’; uranae-zi ‘to bring a fire-offering’; urkiae-zi ‘to track down’; ašdulae-zi ‘to offend’; arrae-zi ‘to come to help’; arpae-zi ‘to suppress’; arpa/ilae-zi ‘to surround(??)’; ešuae-zi ‘?’; edae-zi ‘to bring (here)’; zaḫḫurae-zi ‘to break, to crush’; zammurae-zi ‘to insult’.


Stems that secondarily inflect according to the atrae-class (in brackets their original stem):

appae-zi ‘to be finished’ (āppa-i / āppi-); arae-zi ‘to (a)rise’ (arai-i / ari-); alae-zi ‘to set in motion’ (alai-i / ali-); arrae-zi ‘to grind’; (arra-i / arr-); arkiae-zi ‘to get lost’ (ark-zi); uae-zi ‘to run’ (uai-i / ui-); uae-zi ‘to scatter’ (uai-i / ui-); išgae-zi ‘to smear’ (iškie/a-zi); išgarae-zi ‘to stab’ (iškār-i / iškar-); išpae-zi ‘to be satiated’ (išpai-i / išpi-); išparzae-zi ‘to escape’ (išpart-zi); ištalgae-zi ‘to level’ (ištalk-zi); ištantae-zi ‘to stay put, to linger’ (ištantāe/a-zi); kappuae-zi ‘to count’ (kappue/a-zi); karšae-zi ‘to cut (off)’ (karš(ie/a)-zi); kišae-zi ‘to comb’ (kiš-zi); lae-zi ‘to loosen’ (lā-i / l-); lauae-zi ‘to pour’ (u-i / lau-); lukkae-zi ‘to set fire to’ (lukk(ie/a)-zi); pae-zi ‘to go’ (pai-zi / pai-); paškae-zi ‘to stick in’ (pāšk-i / pašk-); paškuae-zi ‘to reject’ (pašku-zi); pattae-zi ‘to run’ (pattai-i / patti-); peššiae-zi ‘to throw away’ (peššie/a-zi); šae-zi ‘to become sullen’ (šāi-zi); šakuae-zi ‘to see, to look’ (šakuāe/a-zi); šarae-zi ‘to embroider’ (šar-(tt)a(ri); šarie/a-zi); šartae-zi ‘to wipe’ (šarta-i / šart-); šešae-zi ‘to decide’ (šiša-i / šiš-); šullae-zi ‘to become arrogant’ (šullē-zi); šuae-zi ‘to push away’ (šue/a-zi); šuae-zi ‘to spy’ (šuāe/a-zi); tae-zi ‘to steal’ (e/a-zi); daištiae-zi ‘to load’ (tāišta-i / tāišti-); dālae-zi ‘to leave in peace’ (dāla-i / dāli-); tapariae-zi ‘to lead, to decide’ (taparie/a-zi); tarnae-zi ‘to allow’ (tarna-i / tarn-); taruppae-zi ‘to collect’ (tarupp-zi); tekkuššiae-zi and tekkuššae-zi ‘to show’ (tekkuššie/a-zi); duarnae-zi and duarniae-zi ‘to break’ (duarni-zi / duarn-); unuae-zi ‘to decorate’ (unu-zi); iae-zi ‘to cry out’ (ai-i / i-); aššae-zi and aššiae-zi ‘to wear’ (ašše/a-zi); aštae-zi ‘to sin’ (ašta-i / ašt-); zae-zi ‘to cross’ (zai-i / zi-); zankilae-zi ‘to fine’ (zankila-i / zankil-).   Ic3      mi-verbs in e/a-: the e/a-class.

This class consists of four verbs only that go back to *-eh2-é/ó-. Because of the fact that an OH intervocalic -- is lost within the Hittite period (cf., these verbs in principle regularly develop into atrae-class verbs in NH times (e.g. ištantāe/a- > ištantae-, OS šu-a-i-ez-zi > NS šu-a-a-iz-zi), although we often find NH forms in which -- has been restored (NH ta-a-i-e-ez-zi). See s.v.v. for these verb’s inflections.


ištantāe/a-zi ‘to stay put, to linger’; šakuāe/a-zi ‘to see, to look’; šuāe/a-zi ‘to spy’; e/a-zi ‘to steal’.    Ic4      mi-verbs in -ue/a-.

The sequence -ue/a-, as found in the verbs of this class, clearly goes back to *-u-é/ó-. On the one hand, these are denominative verbs that are derived from u-stem nouns and, on the other, verbs of the structure *Cu-é/ó-. For the prehistory of e-zi / ua- ‘to come’, see s.v.


e/a-zi, uiše/a-zi ‘to stay alive’; kappue/a-zi ‘to count, to calculate’; genzue/a-zi ‘to treat gently’; šakurue/a-zi to water (animals)’; šarkue/a-zi ‘to put on footwear’; šarue/a-zi ‘to loot’; šue/a-zi ‘to fill’; šue/a-zi ‘to push (away)’; e-zi / ua- ‘to come’.    Ic5      ašše/a-zi ‘to dress’.

This verb consitutes a class of its own, because it is the only verb that shows the sound law *VsV > Hitt. VššV. See s.v. for an elaborate treatment, in which I argue that ašše/a-zi ultimately goes back to *us-é/ó-.    Ic6      imperfectives in -ške/a-.

See at its own lemma for an elaborate treatment of the suffix -ške/a- < *-sé/ó-. The distribution between the thematic vowels -e- and -a- are changing throughout the Hittite periode, compare the following overview:


               PIE                 OS                         MH/MS                                     NH


1sg.      *-só-               °š-ke-e-mi               °š-ke-mi                                      °š-ke-mi          

2sg.      *-sé-               °š-ke-e-ši                °š-ke-ši                                        °š-ke-ši           

3sg.      *-sé-               °š-ke-ez-zi               °š-ke-ez-zi                                   °š-ke-ez-zi      


1pl.       *-só-               °š-ke-e-a-ni          °š-ke-u-e-ni, °š-ga-u-e-ni            °š-ke-u-e-ni    

2pl.       *-sé-               °š-ket9-te-ni             °š-ke-et-ta-ni, °š-ke-et-te-ni         °š-ke-et-te-ni   

3pl.       *-só-               °š-kán-zi                 °š-kán-zi                                     °š-kán-zi         



1sg.      *-só-[38]            --                             °š-ke-nu-un                                 °š-ke-nu-un

2sg.      *-sé-               --                             °š-ke-eš                                       °š-ke-eš

3sg.      *-sé-               °š-ke-e-et                °š-ke-et                                        °š-ke-et


1pl.       *-só-               --                             °š-ga-u-en                                   °š-ke-u-en

2pl.       *-sé-               --                             °š-ke-et-ten                                  °š-ke-et-ten

3pl.       --                       °š-ke-e-er                --                                                 °š-ke-er



2sg.      *-sé                °š-ki-i                     °š-ki                                             °š-ki

3sg.      *-sé-               --                            °š-ke-ed-du                                  °š-ke-ed-du


2pl.       *-sé-               --                            °š-ke-et-tén                                  °š-ke-et-tén

3pl.       *-so-              --                            °š-kán-du                                    °š-kán-du


part.      *-só-              °š-kán-t-                 --                                                 --

sup.      --                      °š-ke-a-an            °š-ke-u-a-an                              °š-ke-u-a-an


Often, this suffix is transliterated with the vowel -i-: °š-ki-mi, °š-ki-ši, °š-ki-iz-zi, etc. This is incorrect, however, as is clearly shown by cases like 2sg.pret.act. °š-ke-eš (never **°š-ki-iš) and plene spellings like °š-ke-e-mi, °š-ke-e-et. It must be admitted that plene spellings predominantly occur in OS texts, but occasional NH cases like a-ri-iš-ke-e-nu-un (KUB 14.13 i 53 (NH)) and me-mi-eš-ke-e-zi (KUB 23.93 rev. 21 (NH)) indicate that in NH times, too, the suffix contained the vowel -e-. The case of 2sg.imp.act. is different, however. Here we find several plene spellings °š-ki-i,[39] including an OS one, which indicate that we are dealing with /-skí/. Apparently, absolute word-final *-é# > Hitt. /-í/.


Since the number of imperfectives in -ške/a- is very large, I did not find it useful to list them all here.     II = i-conjugation

Within the i-conjugated verbs only two classes can be distinguished: (a) i-verbs that show ablaut; (b) i-verbs that do not show ablaut. It should be noted that, unlike under the mi-verbs, there are no i-verbs that show a thematic suffix. The only suffixes that can be found within the i-conjugation are the ablauting *-oi-/-i-suffix (see IIa4 and IIa5), the imperfective suffixes -šša-/-šš- (see under IIa1β) and -ānna- / -ānni- (see under IIa5), and the nasal-infix (mainly under IIa1γ, but cf. also ḱ 2.3.4).   IIa = ablauting i-verbs

The origin of the Hittite i-conjugation has been fiercely debated. Nevertheless, I think that it is clear that formally the i-conjugation can be compared to the PIE perfect, with the only difference that in Hittite there are no traces of a reduplication syllable. Structurally, the Hittite i-verbs are best compared to the isolated PIE verb *ueid- ‘to know’, which was unreduplicated but inflected as a perfect. For the matter of comparison, I have given here the paradigm of the perfect of the Skt. verb dā- ‘to give’, the preterite paradigm of Hitt. dā-i / d- ‘to take’ (with the plural forms taken from the compound verbs peda-i / ped- and uda-i / ud-) as well as the Greek paradigm of οἶδα ‘to know’:


            PIE                       Skt.             Hitt.                                cf. Gr.


1sg.   *de-dóh3-h2e           dadáu          ḫḫun                           (ϝ)οῖδα       <    *uóid-h2e

2sg.   *de-dóh3-th2e          dadā́tha       dātta                                (ϝ)οῖσθα    <    *uóid-th2e

3sg.   *de-dóh3-e              dadáu          dāš (cf. pres. dāi)            (ϝ)οῖδε        <    *uóid-e


1pl.    *de-dh3-mé             dadimá        °tumen                             (ϝ)ίδμεν     <    *uid-mé

2pl.    *de-dh3-+Õ             dadá            °tišten                              (ϝ)ίστε       << *uid-+Õ

3pl.    *de-dh3-ḗr              dadúr          °ter                                  (ϝ)ίσᾱσι     << *uid-ḗ́r


As I will explain below, I believe that the PIE ablaut *o/Ø underlies all the ablauting i-verbs as attested in Hittite. Within the ablauting i-verbs I distinguish five types, namely verbs that show an ablaut ā/Ø (with subtypes), ā/a, ā/¸, -ai-/-i- and -a-/-i-.   IIa1    i-verbs with ā/Ø-ablaut

For the sake of convenience, I have subdivided this class into three groups.

The first group (IIa1α) consists of verbs that show a structure *CueC- and *Ceu(C)-. Note that the two verbs that show the structure *Ceu(C)-, au-i / u- ‘to see’ and mau-i / mu- ‘to fall’, use a secondary, mi-inflecting stem in their 3sg. forms: aušš- and maušš- . This is probably due to the fact that expected *āi and *i were not transparent enough and therefore were changed to aušzi and maušzi on the basis of 3sg.pret.act. *auš and *mauš, which afterwards received mi-endings themselves as well: aušta and maušta.


au-i / u- ‘to see’ < *h2ou- / *h2u-; uapp-i / upp- ‘to hurl; to do evil’ < *h2uoph1- / *h2uph1-; uart-i / urt- ‘to curse’ < *h2uort- / *h2urt-; mau-i / mu- ‘to fall’ < *mouh1- / *muh1-.


It should be mentioned that in none of these verbs the plene vowel ā < *ó is attested as such. In the verbs au-/u- and mau-/mu- this is regular since *-óu- yields Hitt. /áu/ and not **/ā́u/ (cf. *-ói- > /ái/ and not **/ā́i/ in class IIa4 below), but in the verbs uapp-/upp- and uart-/urt- we would have expected *u-a-a-ap-p° and *u-a-a-ar-t° in the oldest texts. The absence of these spellings must be ascribed to the fact that both verbs are unattested in OS texts whereas an OH sequence /ā́CCV/ develops into NH /áCCV/ (so shortening of OH /ā́/ in non-final closed syllable within the Hittite periode, cf. ḱ  

The second group (IIa1β) consists of two monosyllabic verbs that end in *-eh1/3-, namely dā-i / d- ‘to take’ < *doh3- / *dh3- and lā-i / l- ‘to loosen, to release’ < *loh1- / *lh1-. The original inflection of these verbs is as follows:


                pres.                     pret.                      imp.


1sg.         Ca-a-a-é         Ca-a-a-u-un                               

2sg.         Ca-a-at-ti             Ca-a-at-ta            Ca-a                    part.     Ca-an-t-

3sg.         Ca-a-i                  Ca-a-aš                Ca-a-ú                 v.n.      Ca-a-u-a-ar


1pl.          Cu-me-e-ni          Ca-a-u-en                                        inf.I     Ca(-a)-u-a-an-zi

2pl.          Ca-at-te-e-ni        Ca-a-at-te-en       Ca-a-at-te-en       inf.II    Ca-a-an-na

3pl.          Ca-an-zi              Ca-a-er                Ca-an-du             impf.    Ca-aš-ke/a-


Note that lā-i / l- from MH times onwards is being replaced by lae-zi, according to the productive atrae-class inflection. In the verb dā-i / d-, the development OH /ā́CCV/ > NH /áCCV/ gives rise to a slightly different NH paradigm:


                pres.                     pret.                      imp.


1sg.         da-a-i              da-a-u-un                                   

2sg.         --                          da-at-ta                da-a                     part.     da-an-t-

3sg.         da-a-i                   da-a-aš                da-a-ú                 v.n.      --


1pl.          tu4-me-e-ni           --                                                      inf.I     da-a-u-a-an-zi

2pl.          da-at-te-e-ni         da-at-te-en           da-at-te-en           inf.II    da-an-na

3pl.          da-an-zi               da-a-er                 da-an-du             impf.    da-aš-ke/a-  

The third group (IIa1γ), which is called the tarn(a)-class, consists of verbs that show a stem CVCa- besides CVC-, e.g. tarna-i / tarn-. Some of these verbs also go back to roots that end in *-eh1/3-, but the difference with group IIa1β (dā-i / d- and lā-i / l-) is that the latter are monosyllabic, whereas the tarn(a)-class verbs are polysyllabic (e.g. peda-i / ped-). This makes the inflection of the tarn(a)-class quite different:



1sg.      °Ca-a-é                                 tar-na-a-é (OS), pé-e-ta-a-é (OS)

2sg.      °Ca-at-ti                                    pé-e-da-at-ti (OS)                                 

3sg.      °Ca-i                                          tar-na-i (OS), al-zi-iš-ša-i (OS), pé-e-ta-i (OS)


1pl.       °Cu-me-e-ni                               pé-e-tu-me-e-ni (OS), ú-du-me-e-ni (OS)

2pl.       °C-štēni > °Ca-at-te-ni              i-iš-te-e-ni (OS); tar-na-at-te-ni (MS)                            

3pl.       °Ca-an-zi                                   tar-na-an-zi (OS), al-zi-iš-ša-an-zi (OS)



1sg.      °Ca-a-u-un                           tar-na-a-u-un (OS), ú-da-u-un (OS)

2sg.      °Ca-at-ta > °Ca-aš, °Ci-iš-ta    a-aš-ta-at-ta (OH/NS); pé-e-da-aš (MS);

                                                                                 ḫal-zi-iš-ši-iš-ta (NH)

3sg.      °Ca-aš > °Ci-iš-ta                     tar-na-aš (OS), pé-e-da-aš (OS); tar-ni-iš-ta (NH)      


1pl.       °Cu-me-en                                 pé-e-tu-mé-en (OS)                               

2pl.       *°C-šten > °Ca-at-te-en            a-aš-ta-at-ten (MS), tar-na-at-ten (NS)

3pl..      °Ce-er                                        i-iš-še-er (OS), pé-e-te-er (OS), ú-ter (OS)



2sg.      °Ca                                            tar-na (OS), al-zi-iš-ša (OS), pé-e-da (MS)

3sg.      °Ca-ú                                         tar-na-ú (OS)                                        


2pl.       °Ci-iš-te-en > °Ca-at-te-en        i-iš-te-en (OS), pé-ti-iš-te-en (OS), tar-na-at-ten (MS)

3pl.       °Ca-an-du                                 pé-e-ta-an-tu (OS)


part.      °Ca-an-t-                                   tar-na-an-t- (OS)                                              

v.n.       °Cu-mar                                    tar-nu-mar (NS)

inf.I      °Cu-ma-an-zi                            pé-e-tu4-ma-an-zi (MS)


Note that the vowel of the strong stem is long in the monosyllabic verbs, whereas it is short in the polysyllabic verbs (da-a-i vs. pé-e-da-i), according to the phonetic developments as described in ḱ and ḱ Furthermore, we find full grade forms in the preterite plural of the monosyllabic verbs, but zero grade in these forms of the polysyllabic verbs (da-a-u-en vs. pé-e-tu-mé-en and da-a-er vs. pé-e-te-er).

In the older literature, the tarn(a)-class is often called ‘thematic’ but this is incorrect. Although the strong stem CVCa- at first sight resembles the structure of some thematic mi-verbs, and although this stem spreads into the 2pl. forms (where the original form *CVC-stēni often yielded awkward clusters), the fact that we find forms like 1pl. CVC-umēni, CVC-umen, verb.noun CVC-umar and inf.I CVC-umanzi throughout Hittite shows that the basic ablaut distinction between the strong stem CVCa- and the weak stem CVC- remains intact.


The tarn(a)-class verbs go back to two groups. On the one hand, we find verbs with the structure *CR-no-h1/3- (nasal-infixed verbs, see ḱ 2.3.4), *Ce-C(R)oh1/3- (reduplicated verbs) and verbs in -šša-i / -šš- (imperfective suffix) < *-soh1- / *-sh1-:


alzišša-i / alzišš- ‘to cry out, to call’ < *h2lt-i-soh1- / *h2lt-i-sh1-; anna-i / ann- ‘to sue, to judge’ < *h3e-h3noh3- / *h3e-h3nh3-; īšša-i / īšš- ‘to do, to make’ < *HH-i-soh1- / *HH-i-sh1-; mimma-i / mimm- ‘to refuse’ < *mi-moh1- / *mi-mh1-; peda-i / ped- ‘to take (away)’ < *h1poi-doh3- / *h1poi-dh3-; pippa-i / pipp- ‘to tear down’ < *pi-poh1/3- / *pi-ph1/3-; šanna-i / šann- ‘to hide’ < sn-no-h1- / *sn-n-h1-;

šišša-i / šišš- ‘to impress’ < *sh1-i-soh1- / *sh1-i-sh1-; šunna-i / šunn- ‘to fill’ < *su-no-h1/3- / *su-n-h1/3-; tarna-i / tarn- ‘to let (go)’ < *tr(k)-no-h1/3- / *tr(k)-n-h1/3-; uda-i / ud- ‘to bring (here)’ < *h2ou-doh3- / *h2ou-dh3-; arrišša-i / arrišš- ‘to help’.


On the other hand, we find verbs with the structure *CoCh2/3-. We would have expected that these verbs, which showed an original ablaut *CoCh2/3- / *CCh2/3-, ended up in the class IIa1α or IIa2. That this is not the case is in my view caused by the 3sg.pres.act. form. In verbs of the structure *CoCh2/3-, the 3sg.pres.act. *CoCh2/3-e-i yielded Pre-Hittite *CoCai, whereas in verbs of class IIa1α and IIa2, 3sg.pres.act. *CC-oh1-e-i and *CoC-e-i yielded Pre-Hittite *CC-ae and *CoCe, respectively. At the time when the ending *-e was replaced by -i (in analogy to the mi-endings that all ended in -i), the ending -ai of CoCai merged with the ending of CCai of the tarn(a)-class, and not with CoCi of class IIa1α and IIa2. On the basis of this 3sg. form, the whole paradigm was transferred into the tarn(a)-class.[40] This scenario explains the inflection of the following verbs:


arra-i / arr- ‘to grind’ < *h2orh3-ei / *h2rh3-enti; iškalla-i / iškall- ‘to split’ < *skolh2/3-ei / *sklh2/3-enti; išparra-i / išparr- ‘to trample’ < *sporh2/3-ei / *sprh2/3-enti; malla-i / mall- ‘to mill’ < *molh2-ei / *mlh2-enti; padda-i / padd- ‘to dig’ < *bhodhh2-ei / *bhdhh2-enti; šarta-i / šart- ‘to wipe, to rub’ < *sordhh2/3-ei / *srdhh2/3-enti.


From MH times onwards, the tarn(a)-class inflection is becoming productive, and we therefore find several verbs that sometimes show forms that secondarily inflect according to the tarn(a)-class (their original stem between brackets):


Àrra-i / Àrr- ‘to wash’ (ārr-i / arr-), ezza-i / ezz- ‘to eat’ (ed-zi / ad-), alila-i / alil- ‘to genuflect’ (alila-i / alili-), atta-i / att- ‘to pierce’ (att-a(ri)), u(a)rta-i / u(a)rt- ‘to curse’ (uart-i / urt-), ua-i / u- (uai-i / ui-), išgara-i / išgar- ‘to stab’ (iškār-i / iškar-), išparra-i / išparr- ‘to spread out’ (išpār-i / išpar-), išparza-i / išparz- ‘to escape’ (išpart-zi), ganga-i / gang- ‘to hang’ (kānk-i / kank-), karša-i / karš- ‘to cut’ (karš(ie/a)-zi), kuenna-i / kuenn- ‘to kill’ (kue(n)-zi / kun-), laua-i / lau- ‘to pour’ (u-i / lau-), lilua-i / lilu- ‘to pour’ (lilua-i / lilui-), malda-i / mald- ‘to recite’ (māld-i / mald-), mēma-i / mēm- ‘to speak’ (mēma-i / mēmi-), nanna-i / nann- ‘to drive’ (nanna-i / nanni-), para-i / par- ‘to chase’ (par-zi), parip(p)ara-i / parip(p)ar- ‘to blow (a horn)’ (parip(p)ara-i / parip(p)ari-), penna-i / penn- ‘to drive (there)’ (penna-i / penni-), šalika-i / šalik- ‘to touch’ (šalÌk-a(ri)), šarra-i / šarr- ‘to divide up’ (šārr-i / šarr-), šiša-i / šiš- ‘to decide, to appoint’ (šiša-i / *šiši-), taišta-i / taišt- ‘to load’ (taišta-i / taišti-), dāla-i /dāl- ‘to let, to leave’ (dāla-i / dāli-), teta-i / tet- ‘to thunder’ (tit-a), titta-i / titt- ‘to install, to assign’ (titta-i / titti-), duarna-i / duarn- ‘to break’ (duarni-zi / duarn-), ūnna-i / ūnn- ‘to send (here)’ (ūnna-i / ūnni-), uppa-i / upp- ‘to send (here)’ (uppa-i / uppi-), zinna-i / zinn- ‘to finish’ (zinni-zi / zinn-).


This phenomenon is often called ‘thematization’, but as we have seen above, this term should be avoided since the tarn(a)-class has nothing to do with thematic inflection.   IIa2    i-verbs with ā/a-ablaut

This class consists of verbs with a root structure *HeC-, *Ceh2(C)-, *CeR(C)-, *TeT-, *seT- and *Ces-. In these verbs, the full grade vowel *ó yielded Hitt. -ā-, whereas in the initial cluster of the zero grade stem an empty vowel -a- is used in spelling (e.g. akkanzi = /ʔkántsi/, ananzi = /Hnántsi/, išparanzi = /¸sprántsi/, maldanzi = /mldántsi/, pašanzi = /pHsántsi/, dakkanzi = /tkántsi/, etc.). The frequent distribution of single stop in the full grade stem vs. geminate stop in the zero grade stem (e.g. āki : akkanzi) is due to lenition of consonants after *ó. Note that a plene spelling -ā- is attested not in all verbs, but this is due to the inner-Hittite shortening of OH /ā́/ to NH /á/ in non-final closed syllables (OH /ā́CCV/ > NH /áCCV/, cf. ḱ so if a certain verb of the structure CaCC- is attested in NH texts only, it will not show a long -ā- anymore. Nevertheless, I assume that in all these verbs we have to reckon with original -ā- in the full grade stem.


āk-i / akk- ‘to die’ < *h1/3ok- / *h1/3k-; ār-i / ar- ‘to come’ < *h1or- / *h1r-; ārr-i / arr- ‘to wash’ < *h1orh1- / *h1rh1-; ārk-i / ark- ‘to cut off’ < *h1or()- / *h1r()-; ārk-i / ark- ‘to mount, to copulate’ < *h3orgh- / *h3rgh-; ān-i / an- ‘to draw (water)’ < *h2on- / *h2n-; āš-i / ašš- ‘to give birth’ < *h2oms- / *h2ms-; āt-i / at- ‘to dry up’ < *h2od- / *h2d-; atk-i / atk- ‘to shut, to close’ < *h2odhgh- / *h2dhgh-; iškār-i / iškar- ‘to stab’ < *skor- / *skr-; išpānt-i / išpant- ‘to libate’ < *spond- / *spnd-; išpār-i / išpar- ‘to spread’ < *spor- / *spr-; ištāp-i / ištapp- ‘to shut’ < *stop- / *stp-; kānk-i / kank- ‘to hang’ < *ḱonk- / *ḱnk-; kalank-i / kalank- ‘to soothe’ < *glongh- / *glngh-; u-i / lau- ‘to pour’ < *loh2u- / *lh2u-; lāk-i / lak- ‘to make lie down’ < *logh- / *lgh-; mālk-i / malk- ‘to spin’ < *molK- / *mlK-; māld-i / mald- ‘to recite’ *moldh- / *mldh-; mārk-i / mark- ‘to divide, to separate’ < *mork- / *mrk-; -i / naḫḫ- ‘to fear’ < *noh2- / *nh2-; paš-i / paš- ‘to protect’ < *poh2s- / *ph2s-; palā-i / palaḫḫ- ‘to call (?)’ < *Ploh2- / *Plh2-?; papparš-i / papparš- ‘to sprinkle’ < *-pors- / *-prs-; pāš-i / paš- ‘to swallow, to gulp down’ < *poh3-s- / *ph3-s-; šākk-i / šakk- ‘to know’ < *sokh1- / *skh1-; dākk-i / dakk- ‘to resemble’ *do(k)h1- / *d(k)h1-; āk-i / akk- ‘to bite’ < *oh2g- / *h2g-; arš-i / arš- ‘to harvest, to wipe’ < *ors- / *rs-; āš-i / *aš- ‘to sell’ < * os- / * s-; -i / zaḫḫ- ‘to beat’ < *tioh2- / *tih2-.    IIa3    i-verbs with ā/¸-ablaut

This class consists of the verbs ašāš-i / aše/iš- ‘to sit’, amank-i / ame/ink- ‘to tie’, karāp-i / kare/ip- ‘to devour’ and šarāp-i / šarip- ‘to sip’. These verbs show the vowel -ā- in the strong stem forms and the vowel -e- or -i- in the weak stem forms. It should be noted that the verb that is usually cited as major representative of this class, “šākk-/šekk-” ‘to know’, in fact does not belong here. As I have argued s.v., its oldest forms show that this verb is in fact šākk-i / šakk- and belongs with class IIa2 (ā/a-ablaut).

The origin of the ablaut -ā-/-e/i- has always been problematic. For instance, Oettinger (1979a: 114) assumes that the source of the -ā-/-e/i- ablaut is the verb "šākk- / šekk-": in his opinion, the vowel -e- is the regular outcome of a reduplication syllable *se-sg-. Apart from the fact that this etymological interpretation has proven to be incorrect, the verb actually is šākk-i / šakk- and does not originally belong to the -ā-/-e/i-ablauting type at all (see s.v. for a detailed description of its prehistory).

A view that has gained much support was initiated by Jasanoff (1979: 86) who assumes that the Hittite ablaut “ā/e” can only be explained by assuming that it reflects a PIE ablaut *o/e. The simplicity of this solution has appealed to many scholars (e.g. Melchert (1994: 81), who reconstructs šākk-/šekk- < *sók-/sék- and karāp-/garip- < *grób-/gréb-). Nevertheless, it is in my view quite problematic that no instances of verbal *o/e-ablaut from any other IE language are known. Either this means that the Hittite -ā-/-e/i-ablaut is very archaic, or it means that Jasanoff’s idea is incorrect.

As I have explained under class Ia5 (cf. ḱ, I believe that the weak stem vowel e/i that we find in the mi-inflecting verbs of classes Ia5 and Ia6 must be compared to the vowel e/i as attested in the i-verbs of the present class, and that this vowel must be identified as the anaptyctic vowel /¸/. This vowel was used in the zero grade forms of these verbs for different reasons.

The verbs karāp-i / kare/ip- and šarāp-i / šarip- must be taken together with terepp-zi / tere/ipp- ‘to plough’ (from class Ia5). As I have stated in ḱ more elaborately, it is significant that these are the only three verbs in Hittite that show a structure *CReC-. I therefore assume that the phonetically expected reflexes of the ablauting pair *CReC- / *CRC- > Hitt. CReC- / CaRC- (when mi-conjugated) and *CRóC- / *CRC- > Hitt. CRāC- / CaRC- (when i-conjugated) were too aberrant (with the stem vowel shifting place) and have therefore been eliminated: in the zero grade form the anaptyctic vowel /¸/ was secondarily placed on the place of the vowel of the full grade form. In this way, a mi-conjugating verb *CRéC- / *CRC- was adjusted to synchronic CReC- / CR¸C-, whereas the i-conjugating *CRóC- / *CRC- was adjusted to synchronic CRāC- / CR¸C-. In both cases, the weak stem is spelled CRe/iC-.

With this scenario in mind, we can explain karāp-i / kare/ip- ‘to devour’ as phonological /krāb- / kr¸b-/, the ‘regular’ secondary outcome of *ghróbh1- / *ghrbh1-, and šarāp-i / šarip- ‘to sip’ as phonological /srāb- / sr¸b-/, the eventual outcome of *srobh- / *srbh-.

The interpretation of ašāš-i / aše/iš- ‘to seat’ and amank-i / ame/ink- ‘to tie’ is less clear. Nevertheless, in the case of ašāš- / aše/iš-, which reflects either *h1sh1os- / *h1sh1s- or *h1h1sh1oh1s- / *h1h1sh1h1s- (thus LIV2, in analogy to Oettinger 1979a: 431), the assumption of an anaptyctic vowel to release the cluster *h1sh1s- or *h1h1sh1h1s- seems unproblematic to me (cf. the phonetically regular development of *CRHsV > Hitt. CR¸ššV (Kloekhorst fthc.a and ḱ This means that ašāš-i / aše/iš- ‘to seat’ must be phonologically interpreted as /ʔsās- / ʔs¸s-/ < *h1(h1)sh1ó(h1)s- / *h1(h1)sh1(h1)s-.

In the case of amank-i / ame/ink- we are dealing with a zero grade stem *h2mngh-. As we will see in ḱ 2.3.4, a pre-Hittite cluster *CNNC regularly develops into /CN¸NC/. Here, *h2mnh- yields Hitt. /Hm¸ng-/, spelled ame/ink-. Consequently, amank-i / ame/ink- ‘to tie’ must be phonologically interpreted as /Hmāng- / Hm¸ng-/ < *h2mongh- / *h2mngh-.   IIa4    i-verbs in -ai-/-i-: the dāi/tianzi-class.

The formal interpretation of this class has been elaborately treated in Kloekhorst 2006a. As I have explained there, almost all these verbs go back to a structure *CC-oi- / *CC-i-, i.e. the zero grade of a verbal root followed by an ablauting suffix *-oi-/-i-:


arai-i / ari- ‘to (a)rise’ < *h3r-oi- / *h3r-i-; alai-i / ali- ‘to set in motion’ < *h2l-oi- / *h2l-i-; alzai-i / alzi- ‘to call out’ < *h2lt-oi- / *h2lt-i-; uai-i / ui- ‘to run’ < *h2uh1-oi- / *h2uh1-i-; ai-i / i- ‘to bind’ < *sh2-oi- / *sh2-i-; amai-i / ami- ‘to sing’ < *sh2m-oi- / *sh2m-i-; uai-i / ui- ‘to throw, to scatter’ *sh2u-oi- / *sh2u-i-; išpai-i / išpi- ‘to be satiated’ < *sph1-oi- / *sph1-i-; mai-i / mi- ‘to grow’ < *mh2-oi- / *mh2-i-; nai-i / *ni- ‘to turn’ < *noih1/3- / *nih1/3-; pai-i / pi- ‘to give’ < *h1p-oi- / *h1p-i-; parai-i / pari- ‘to blow’ < *prh1-oi- / *prh1-i-; paddai-i / patti- ‘to run, to flee’ < *pth1-oi- / *pth1-i-; šai-i / ši- ‘to press’ < *sh1-oi- / *sh1-i-: šalai-i / šali- ‘?’ < *sl-oi- / *sl-i-?; dai-i / ti- ‘to put, to place’ < *dhh1-oi- / *dhh1-i-; tarai-i / tari- ‘to exert oneself’ < *Tr-oi- / *Tr-i-; ai-i / i- ‘to cry’ < *oi- / *i-; zai-i / zi- ‘to cross’ < *h1t-oi- / *h1t-i-.


The original inflection of these verbs was as follows:

pres.                                                                    imp.

1sg.   *CC-ói-h2e-i        °Ce-e-e-é                 

2sg.   *CC-ói-th2e-i       °Ca-it-ti                         2sg.      *CC-ói                 °Ca-i

3sg.   *CC-ói-e-i           °Ca-a-i                          3sg.      *CC-ói-u             °Ca-a-ú


1pl.    *CC-i-éni[41]        °Ci-u-e-ni                     

2pl.    *CC-i-sténi[42]       °Ci-iš-te-e-ni                 2pl.       *CC-i-stén[43]        °Ci-iš-te-en

3pl.    *CC-i-énti            °Ci-an-zi                       3pl.       *CC-i-éntu[44]        °Ci-an-du



1sg.   *CC-ói-h2e          °Ce-e-e-u-un            part.      *CC-i-ént-[45]         °Ci-an-t-

2sg.   *CC-ói-th2e         °Ca-it-ta                        v.n.       *CC-ói-r           °Ca-u-a-ar

3sg.   *CC-ói-s              °Ca-iš                           v.n.       *CC-i-ā́tar        °Ci-a-a-tar


1pl.    *CC-i-én[46]         °Ci-u-en                        inf.I      *CC-i-anzi[47]      °Ci-u-a-an-zi

2pl.    *CC-i-stén[48]                                              inf.II     *CC-i-anna         °Ci-an-na

3pl.    *CC-i-ḗr[49]           °Ci-e-er                         impf.    *CC-ske/a-[50]


This is the paradigm as found in OH texts. From the MH period onwards, we find generalization of the long -ā- as found in 3sg.pres.act. °Ca-a-i, yielding forms like 2sg.pres.act. °Ca-a-it-ti, 2sg.pret.act. °Ca-a-it-ta, 3sg.pret.act. °Ca-a-iš, 2sg.imp.act. °Ca-a-i, and 2pl.imp.act. °Ca-a-iš-tén (with introduction of strong stem). Moreover, from MH times onwards, the dāi/tianzi-class inflection is gradually being replaced by the -e/a-class (Ic1) and, to a lesser extent, by the atrae-class (Ic2). The transfer to the -e/a-class is triggered by forms like 3pl.pres.act. °Cianzi, 3pl.pret.act. °Cier, 3pl.imp.act. °Ciandu, part. °Ciant-, etc., whereas the transfer to the atrae-class is based on the trivial replacement of 3sg.pres.act. °Cāi by °Cāizzi and on the fact that e.g. 2sg.imp.act. °Cai and verb.noun °Caar are identical in both classes.   IIa5    i-verbs in -a-/-i-: the mēma/i-class.

In the oldest texts, the verbs of this class show two stems: we find a strong stem ending in -a- besides a weak stem ending in -i-. The original inflection can be schematically represented as follows:


                  pres.                     pret.                        imp.


1sg.         °Caḫḫe                °Caḫḫun                                               

2sg.         °Catti                   --                             °Ci

3sg.         °Cai                     °Cišʔ                       °Cau


1pl.          °Cieni                °Cien                                                    part.      °Ciant-

2pl.          °Cišteni                °Cišten                    °Cišten                    inf.I      °Cianzi

3pl.          °Cianzi                °Cier                       °Ciandu                  impf.    °Ciške/a-


The ablaut -a-/-i- cannot reflect a PIE situation however, and therefore it is likely that the mēma/i-class is of a secondary origin. There are a few clues that shed some light on the prehistory of this class.

First, some of the verbs that belong to this class are derived from dāi/tianzi-inflected verbs: penna-i / penni-, ūnna-i / ūnni- and nanna-i / nanni- derive from nai-i / *ni- ‘to lead’ (the first two showing the preverbs pe- and u-, the third one showing reduplication) and uppa-i / uppi- derives from pai-i / pi- ‘to give’ (with the preverb u-).

Secondly, the mēma/i-class is not a very stable inflection type. If we look at a diachronic overview of attestations, we see that from MH times onwards, tarn(a)-class inflected forms (printed in bold) and, to a lesser extent, -e/a-class inflected forms (printed underlined) are spreading throughout the paradigm,.


             OS                       MH                                               NH


1sg.      °Ca-a-é           °Ca-a-i                                     °Ca-a-i

2sg.      --                          °Ca-at-ti                                        °Ca-at-ti

3sg.      °Ca-i                    °Ca-i, °Ca-a-i                               °Ca-a-i


1pl.       °Ci-u-e-ni            °Ci-u-e-ni, °Ca-u-e-ni                  °Ci-a-u-e-ni

2pl.       --                          °Ci-iš-te-ni                                    °Ca-at-te-ni

3pl.       °Ci-an-zi              °Ci-a-an-zi, °Ca-an-zi               °Ci-a-an-zi, °Ca-an-zi



1sg.      °Ca-a-u-un     °Ca-a-u-un                               °Ca-a-u-un

2sg.      --                          --                                                   °Ci-iš-ta, Ci-eš-ta, Ci-eš

3sg.      --                          °Ci-iš, °Ci-iš-ta                             °Ci-iš, °Ci-iš-ta, °Ca-aš, °Ci-it,



1pl.       --                          °Ci-u-en, °Cu-me-en, °Ca-u-en   °Ca-u-en, °Ci-a-u-en

2pl.       --                          --                                                   °Ci-iš-ten, °Ci-a-at-ten

3pl.       °Ci-er                  °Ci-e-er                                         °Ci-er, °Ce-er



1sg.      --                          --                                                   °Ca-al-lu

2sg.      --                          °Ci, °Ca                                        °Ci, °Ca

3sg.      --                          °Ca-a-ú                                         °Ca-a-ú, °Ca-at-tu4, °Ci-iš-du


2pl.       --                          °Ci-iš-te-en, °Ci-eš-te-en              °Ci-iš-ten, °Ci-eš-ten, °Ca-at-ten

3pl.       --                          °Ci-an-du                                      °Ci-an-du, °Ca-an-du


part.      --                          °Ci-an-t-                                       °Ci-a-an-t-, °Ca-an-t-

v.n.       --                          --                                                   °Ci-a-a-ar, °Cu-mar

inf.I      --                          °Ci-a-an-zi, °Ci-a-a-an-zi     °Ci-a-a-an-zi, °Ca-a-an-zi,


impf.    °Ci-iš-ke/a-          °Ci-iš-ke/a-                                   °Ci-iš-ke/a-, °Ci-eš-ke/a-


In my view, these two facts clearly indicate that the mēma/i-class consists of verbs that belonged to the dāi/tianzi-class originally, but were gradually transferred to the tarn(a)-class from pre-Hittite times onwards. This replacement first took place in the singular forms, which yielded the OH situation as attested: stems in -a- in the present en preterite singular, stems in -i- elsewhere. From MH times onwards, the replacement is taking place in the plural and infinite forms as well. The occasional -e/a-inflected forms can be explained by the fact that dāi/tianzi-class verbs in NH times as a rule are being transferred to the -e/a-class.

The question is, of course, why not all dāi/tianzi-class verbs have been transferred to the tarn(a)-class. Why do e.g. penna/i-, ūnna/i-, nanna/i- and uppa/i- belong to the mēma/i-class, and their basic verbs nai- / *ni- and pai-/pi- do not? In my view, the answer lies in the fact that penna/i-, ūnna/i-, nanna/i- and uppa/i- are polysyllabic whereas nai-/ni- and pai-/pi- are not. This has consequences for the ending of the 3sg.pres.act. form. If we take the original (reconstructed) paradigms of pai-/pi-, uppa/i- and tarn(a)- we get the following picture:


1sg.                 peḫḫi                       *uppeḫḫi                         tarnaḫḫi

2sg.                 paitti                        *uppaitti                          tarnatti

3sg.                 pāi                           uppai                               tarnai


1pl.                  pieni                      uppieni                          tarnumeni

2pl.                  pišteni                      uppišteni                          tarništeni

3pl.                  pianzi                      uppianzi                          tarnanzi


Although the paradigms of pai-/pi- and *uppai-/uppi- are almost identical, they differed in the 3sg. form, where pai-/pi- has a long vowel, pāi, and *uppai-/uppi- a short vowel, uppai. The 3sg. ending of *uppai-/uppi- was identical to the ending of the tarn(a)-class, however, which had a short vowel as well: tarnai. In my view, this form therefore triggered a secondary refurnishing of the polysyllabic dāi/tianzi-verbs in analogy to the tarn(a)-verbs. First, the singular forms were changed (*uppeḫḫi > uppaḫḫi, *uppaitti > uppatti), and then the other forms (uppieni > uppaeni, uppišteni > uppatteni, uppianzi > uppanzi, etc.).


The verbs that inflect according to the mēma/i-class are: āppa-i / āppi- to be finished’; alila-i / alili- ‘to genuflect’ < *h2li-h2loi- / *h2li-h2li-; lilua-i / lilui- ‘to pour’ < *li-lh2u-oi- / *li-lh2u-i-; mēma-i / mēmi- ‘to speak’ < *me-h1m-oi- / *me-h1m-i-?; nanna-i / nanni- ‘to drive’ < *no-noiH- / *no-niH-; parip(p)ara-i / parip(p)ari- ‘to blow (a horn)’ < *pri-prh1-oi- / *pri-prh1-i-; penna-i / penni- ‘to drive (there)’ < *h1poi+noiH- / *h1poi+niH-; šiša-i / *šiši- ‘to decide, to appoint’ < *si-sh2-oi- / *si-sh2-i-; taišta-i / taišti- ‘to load’ < *dhoh1-es+dhh1-oi- / *dhoh1-es+dhh1-i-; dāla-i / dāli- ‘to let, to leave’ < *+lh1-oi- / *dā+lh1-i-; titta-i / titti- ‘to install, to assign’ < *dhi-dhh1-oi- / *dhi-dhh1-i-; ūnna-i / ūnni- ‘to send (here)’ < *h2ou+noiH- / *h2ou+niH-; uppa-i / uppi- ‘to send (here)’ < *h2ou+h1p-oi- / *h2ou+h1p-i-.


Verbs showing the imperfective suffix -anna-/-anni- also belong to this class:

allanna-i / allanni- ‘to trample down’; aluganna-i / aluganni- ‘to make an anouncement’; attanna-i / attanni- ‘to pierce’; uganna-i / uganni- ‘to conjure’; uttianna-i / uttianni-; uittianna-i / uittianni- ‘to draw’; ianna-i / ianni- ‘to march’; uanna-i / uanni- ‘to throw’; iškaranna-i / iškaranni- ‘to sting’; šippandanna-i / šippandanni- ‘to libate’; išparanna-i / išparanni- ‘to spread’; laḫḫianna-i / laḫḫianni- ‘to go on an expedition’; pianna-i / pianni- ‘to give’; paranna-i / paranni- ‘to chase’; paršianna-i / paršianni- ‘to break’; peššianna-i / peššianni- ‘to throw away’; piddanna-i / piddanni- ‘to bring away’; šallanna-i / šallanni- ‘to pull’; takšanna-i / takšanni- ‘to level’; tianna-i / tianni- ‘to lay down’; tušanna-i / tušanni- ‘to cut off’; alanna-i / alanni- ‘to hit’; erianna-i / erianni- ‘to call’.    Ablaut pattern of the IIa-verbs

In all i-verbs that show ablaut, this ablaut can be traced back to the PIE ablaut *o/Ø. Therewith it is likely that the i-conjugation is etymologically connected with the PIE perfect, albeit that in Hittite no reduplication is found. The best comparandum therefore is the PIE root *ueid- ‘to know’.

In the following scheme I have indicated the distribution of the ablaut-vowels over the verbal paradigms, first giving the attested Hittite forms (the verb au-i / u- ‘to see’, supplemented with forms from ištāp-i / ištapp- ‘to plug up’, āk-i / akk- ‘to die’, pai-i / pi- ‘to give’, ārr-i / arr- ‘to bathe’ and mēma-i / mēmi- ‘to speak’), then an abstraction of these Hittite data, followed by the reconstructed PIE forms, exemplified by the Sanskrit paradigm of ved- ‘to know’.



1sg.         u-u-i                                      CóC-i                      

2sg.         a-ut-ti                                         CóC-ti                       

3sg.         iš-ta-a-pí                                    CóC-i                        


1pl.          ú-me-(e-)ni                                CC-éni                    

2pl.          uš-te-ni                                      CC-sténi[51]                 

3pl.          ú-a-an-zi                                 CC-ánzi                    



1sg.         u-u-u-un                                CóC-un                        *CóC-h2e            véda

2sg.         a-uš-ta                                       CóC-ta                           *CóC-th2e           véttha

3sg.         a-ak-ki-iš                                   CóC-s                            *CóC-e[52]             véda


1pl.          pí-u-en                                       CC-én                          *CC-mé               vidmá

2pl.          --                                                *CC-stén                        *CC-sÕ ?[53]          vidá

3pl.          pí-i-e-er                                     CC-ḗr                            *CC-ḗr               vidú¡



1sg.         ú-a-al-lu, ú-e5-el-lu-ut           CC-(e)llu(t)               

2sg.         a-ú, iš-ta-a-pí                            CóC(-i)

3sg.         iš-ta-a-pu                                   CóC-u


2pl.          pí-iš-te-en                                  CC-stén                     

3pl.          ú-a-an-du                                CC-ándu


part.         ú-a-an-t-                                  CC-ant-

v.n.          ú-a-tar[54]                                  CC-ā́tar                    

v.n. a-ar-ru-a-aš               CóC-ar

inf.I         me-mi-a-an-zi                         CC-ánzi                  

inf.II        ú-a-an-na                                CC-ā́nna

impf.        ú-uš-ke/a-                                  CC-ské/á-              IIb = non-ablauting i-verbs

The only verbs that belong to this class are, on the one hand, those which originally showed ablaut but have abandoned it:

ānš-i ‘to wipe’ originally belonged to an ablauting verb ānš-i / ane/išš- < *h2omh1s- / *h2mh1s-. Both stems formed their own paradigm: ānš-i and ane/išš-zi.

ārš-i ‘to till the soil’ originally belonged to an ablauting verb ārš-i / **are/išš- < *h2órh3-s- / *h2rh3-s-. After the initial - of the weak stem spread to the strong stem (which regularly should have yielded **ārš-i), the stem ārš- was generalized.

šā-i ‘to stuff up’ reflects *soh2- / *sh2-. Here the strong stem *soh2- > šā- was generalized because the weak stem *sh2- phonetically yielded **-, which was too aberrant.

āš-i ‘to buy’ is only attested with strong stem forms, so its weak stem cannot be determined. It reflects *uos- / *us-, the weak stem of which should regularly have yielded **uš- as attested in ušnie/a-zi ‘to put up for sale’. Within the paradigm āš- / **uš- it is likely that the latter stem was eliminated and replaced by āš-, or that it received an anaptyctic vowel, resulting in aš- (cf. ašše/a-zi for a similar scenario).

eakk-i ‘to demand’ goes back to *ue-uo- / *ue-u-. Here, too, the strong stem was generalized because the weak stem *ue-u- phonetically yielded **ūk-, which was too aberrant.


On the other hand, this class consists of the factitives in -aḫḫ-i (see s.v. for a treatment of this suffix):


alanzaḫḫ-i ‘to bewitch’; araaḫḫ-i ‘to make free’; armaḫḫ-i ‘to make pregnant’; ēšaraḫḫ-i ‘to make blood-red’; alaḫḫ-i ‘to make yellow(green)’; antezziaḫḫ-i ‘to make foremost’; appinaḫḫ-i ‘to enrich’; attaḫḫ-i ‘to instruct’; ikunaḫḫ-i ‘to make cold’; in(n)araḫḫ-i ‘to make strong’; innarauaḫḫ-i ‘to make strong’; aššaraḫḫ-i ‘to make lordly’; iulaḫḫ-i ‘to bind by treaty’; idālaaḫḫ-i ‘to treat badly’; kallaraḫḫ-i ‘to make inauspicious’; kappilaḫḫ-i ‘to get in a fight’; kardimiaḫḫ-i ‘to make angry’; katteraḫḫ-i ‘to lower’; kunnaḫḫ-i ‘to set aright’; kūruriaḫḫ-i ‘to wage war on’; kutruaḫḫ-i ‘to summon as witness’; *lazziaḫḫ-i ‘to make right’; leliaḫḫ-i ‘to make haste’; lūriaḫḫ-i ‘to humiliate’; maantaḫḫ-i ‘to rejuvenate’; maniaḫḫ-i ‘to distribute’; man(n)i(n)kuaḫḫ-i ‘to approach’; manikuandaḫḫ-i ‘to make short’; markištaḫḫ-i ‘to take someone by surprise (?)’; marlaḫḫ-i ‘to make foolish (?)’; maršaḫḫ-i ‘to desecrate’; meuandaḫḫ-i; miauantaḫḫ-i ‘to make old’; mišriaḫḫ-i ‘to make mišriant-’; nakkiaḫḫ-i ‘to become a concern to someone’; *neknaḫḫ-i ‘to regard someone as a brother’; aḫḫ-i ‘to renew’; *palšiaḫḫ-i ‘to set on the road’; papraḫḫ-i ‘to defile’; pararaḫḫ-i ‘to chase’; pedaššaḫḫ-i ‘to install’; šakiaḫḫ-i ‘to give a sign’; šallakartaḫḫ-i ‘to offend someone through arrogance’; šannapilaḫḫ-i ‘to empty’; šan(i)ezziaḫḫ-i ‘to make pleasant’; šarāzziaḫḫ-i ‘to make win’; šumumaḫḫ-i ‘to braid together (?)’; šuppiaḫḫ-i ‘to purify’; dankuaḫḫ-i ‘to make black’; taruppaḫḫ-i ‘?’; taruppiaḫḫ-i ‘?’; dašuaḫḫ-i ‘to make blind’; tatraḫḫ-i ‘to incite’; tepaaḫḫ-i ‘to make little’; (a)lkiššaraḫḫ-i ‘to make perfectly’; aštaḫḫ-i ‘to sin’; ātarnaḫḫ-i ‘to order; to instruct’.


2.3.3 The middle verbs

It is commonly known that two separate endings can be used to express the 3sg.pres.midd. form, namely -tta(ri) and -a(ri). Although synchronically they do not express any difference in meaning, a given verb in principle always uses the same ending. It must be noted that in many verbs that originally use the ending -a(ri), in younger texts the ending -tta(ri) is also used. Of the few verbs that are attested with both -a(ri) and -tta(ri) and of which too little forms are found to set up a chronological overview of attestations, we may therefore assume that -a(ri) is the original ending.

In the literature we sometimes come across the habit to interpret the ending -tta(ri) as “mi-conjugated” and -a(ri) as “i-conjugated”. This should be abandoned, however: the choice of a verb to use either -tta(ri) or -a(ri) has nothing to do with the inflection that it uses in active forms (cf. also their respective lemmata).

Although synchronically no difference in meaning between -tta(ri) and -a(ri) can be determined, it is remarkable that the impersonal verbs of the type tukkāri ‘is visible, is important’ (IIIf) all use the ending -āri. This fits in well with the fact that Kortlandt (1981: 126-7) on the basis of the distribution of the Sanskrit endings -e and -te assumes a semantic diffirence between these endings, namely *-o = ‘deponent’ and *-to = ‘transitive’.[55] It is therefore important to distinguish between the use of the ending -a(ri) and -tta(ri) when classifying the middle verbs. Another criterion is whether or not the verbal root originally showed full grade or zero grade. Furthermore, the impersonally used middles show some remarkable formal features, on the basis of which they must have had a special position.

Taking into account all these criteria, I arrive at the following six basic categories: (a) middle verbs of the structure *CéC-o; (b) middle verbs of the structure *CéC-to; (c) middle verbs of the structure *CC-ó; (d) middle verbs of the structure CC-tó; (e) impersonally used middle verbs of the structure *CéC-o; (f) impersonally used middle verbs of the structure *CC-ó. To these must be added (g) middle verbs in -e/a-tta(ri) and (h) middle verbs of other structures, which are probably of secondary origins.     IIIa     middle verbs of the structure *CéC-o

ā(i)-a(ri) ‘to be hot’ < *h1éh3i-o? (>> -tta(ri) (NH)); eš-a(ri) ‘to sit down’ < *h1éh1s-o (>> eš-tta(ri) (NH)); app-a(ri) ‘to work out’ < *h2ép-o (>> app-tta(ri) (MH)); att-a(ri) ‘to pierce, to prick’ < *h2ét-o (>> azzie/a-tta(ri) (NH)); uetti-a(ri) ‘to draw, to pull’ (>> uettie/a-tta(ri)); kīš-a(ri) / kiš- ‘to happen, to occur’ < *éis-o; *kikkiš-a(ri ‘to happen, to occur’ < *éis-o (>> kikkiš-tta(ri) ) (NH)); nē-a(ri) ‘to turn’ < *néih1/3-o; paš-a(ri) ‘to protect’ < *péh2s-o (>> paš-tta(ri) (NH)); šalīk-a(ri) ‘to touch’ if from *sléi-o; e-a(ri) ‘to turn oneself’ < *uéih2-o (>> e-tta(ri) and ea-tta(ri) (MH)); zē-a(ri) ‘to cook (intr.)’ < *tiéh1-o.     IIIb     middle verbs of the structure *CéC-to

arp-tta(ri) ‘to change allegiance’ < *h3érbh-to; uett-tta(ri) ‘to draw, to pull’ < *h2uéTH-(t)o (>> uettie/a-a(ri) (OH)); ki-tta(ri) ‘to lie’ < *éi-to; lukk-tta ‘to get light’ < *léuk-to (>> lukka-tta ?(OH)); park-tta(ri) ‘to rise’ (>> parkie/a-tta(ri)); arš-tta(ri) ‘to lift oneself’; ešš-tta ‘to be dressed’ < *ués-to (>> aššie/a-tta(ri) (NH)).     IIIc     middle verbs of the structure *CC-ó

ark-a(ri) ‘to mount, to copulate’ < *h3rh (>> ark-tta(ri) (MH)); alzi-a(ri) ‘to cry out’ < *h2lt-i-ó (>>alzie/a-tta(ri) (NH)); parši-a(ri), parš-a(ri) ‘to break’ < *bhrs(-i)-ó (>> paršie/a-tta(ri) (NH)); tuš-a(ri) ‘to (be) cut off’.     IIId     middle verbs of the structure *CC-tó

ar-tta(ri) ‘to stand’ < *h3r-tó; karp-tta(ri) ‘to be angry’ < *rp-tó (?) (>> karpie/a- (NH)); tarupp-tta(ri) ‘to collect oneself’ < *trup-tó.     IIIc/d     middle verbs whose original structure cannot be determined

pukk-(tt)a(ri) ‘to be hateful’; šar-(tt)a(ri) ‘to embroider’< *sr-(t)ó; šupp-(tt)a(ri) ‘to sleep’ < *sup-(t)ó.     IIIe     impersonally used middle verbs of the structure *CéC-o

tit-a ‘to thunder’ (preform unclear).     IIIf     impersonally used middle verbs of the structure *CC-ó

ištu-āri ‘to be exposed’ < *stu-ór-i; kišt-āri ‘to perish’ < *(h)sd-ór-i; lag-āri ‘to fall’ < *lgh-ór-i; mi-āri ‘to be born’ < *mh2i-ór-i (?); tukk-āri ‘to be visible’ < *tuk-ór-i; ur-āri ‘to burn’ < *urh1-ór-i; akk-āri ‘to be lacking’ < *uh2g-ór-i.     IIIg     middle verbs in -e/a-tta(ri)

āššie/a-tta(ri) ‘to be loved’; armanie/a-tta(ri) ‘to become ill’; armalie/a-tta(ri) ‘to become ill’; ermanie/a-tta(ri) ‘to become ill’; allie/a-a(ri) ‘to kneel down’; andae-tta(ri) < *andae/a- ‘to get fixed’; *aššuezzie/a-tta(ri) ‘to become king’; e/a-tta(ri) ‘to go’; imie/a-tta(ri) ‘to mingle’; aru¾e/a-tta(ri) < *arue/a- ‘to weep’; karie/a-(tt)a(ri) ‘to be gracious towards’; karpie/a-tta(ri) ‘to be angry’; kardimie/a-tta(ri) ‘to be angry’; kištanzie/a-tta(ri) ‘to suffer famine’; lāzzie/a-tta(ri) ‘to be good’; lelanie/a-tta(ri) ‘to infuriate’; marrie/a-tta(ri) ‘to melt down’; našarie/a-tta(ri) ‘to show respect’; pangarie/a-tta(ri) ‘to become widespread’; šallie/a-tta(ri) ‘to melt down’; šarrie/a-tta(ri) ‘to be divided’; ešie/a-tta(ri) ‘to pasture’; išurie/a-tta(ri) to suffocate’.     IIIh     other middle verbs

This group consists of the middle verbs that cannot be classified as belonging to one of the classes described above (sometimes beacuse the etymology is unknown). Note that these also include verbs like anna-a(ri), marra-tta(ri) and tarra-tta(ri), which are sometimes called ‘thematic’. In my view, it is possible that in these verbs the -a of the 3sg.pres.midd. ending has spread throughout the paradigm (cf. e.g. s.v. tarra-tta(ri) for an elaboration of this idea).


āšš-a(ri) ‘to be loved’; ai(n)k-tta(ri), ink-a(ri) ‘to bow’; anna-a(ri) ‘to sue’; inik-tta(ri) ‘to pour (?)’; marra-tta(ri) ‘to melt down’; šalla-tta(ri) ‘to melt down’; šarra-tta(ri) ‘to be divided’; šuppiaḫḫ-a(ri) ‘to purify’; damiummaḫḫ-tta(ri) ‘to change’; tarra-tta(ri) ‘to be able’.


2.3.4 Excursus: The Prehistory of the Nasal-infixed verbs

In Hittite, we find a number of verbs that can be regarded as containing a nasal infix. Although most of these verbs inflect according to the mi-conjugation, there are a few i-inflected nasal infixed verbs: amank-i / ame/ink- ‘to tie’, šanna-i / šann- ‘to hide’, šunna-i / šunn- ‘to fill’ and tarna-i / tarn- ‘to let (go)’. Since these i-verbs can hardly have been created secondarily (there is no model in analogy to which they could have been formed), we must assume that they are archaic.


Within the group of Hittite nasal infixed verbs we must distinguish three types:

(1) Verbs with an infix -ni(n)-: arnikzi / arninkanzi ‘to make disappear’ < *h3erg-; unikzi / uninkanzi ‘to bash’ < *h2ueg(h)-; ištarnikzi / ištarninkanzi ‘to afflict’ < *ster-; ninikzi / nininkanzi ‘to mobilize’ < *neik-; šarnikzi / šarninkanzi ‘to compensate’ < *ser-.

(2) Verbs with an infix -Vn-: amanki / ame/inkanzi ‘to tie’ < *h2emh-; tamekzi / tame/inkanzi ‘to attach’ < *temk-.

(3) Verbs with an infix -nV-:[56] aršanezzi / aršananzi ‘to be envious’ < *h1/3ersh1- or *h1/3resh1-; *arnazi / arnanzi ‘to sprinkle’ < *h2erh2/3- or *h2reh2/3-; ullezi / ullanzi ‘to smash’ < *h2uelh1-; kinae-zi ‘to (as)sort’ < *kieh2-, munnae-zi ‘to hide’ < *meuh2/3-, šannai / šannanzi ‘to hide’ < *senh1-; šunnai / šunnanzi ‘to fill’ < seuh1/3-; tarnai / tarnanzi ‘to let (go)’ < *terh1/3-; duarnizzi / duarnanzi ‘to break’ < *dhuerh1-; zinnizzi / zinnanzi ‘to finish’ < *tieh1-.


It should first be noted that the verbs of type (1) and (2) show the development *e/inCC > Hitt. e/iCC (whereas the -n- is preserved in e/inCV), which is also known from e.g. likzi / linkanzi < *h1lengh- (see also ḱ This means that e.g. arnikzi goes back to *arninkzi, unikzi < *uninkzi, and, in type (2), tamekzi < *tamenkzi.


On the basis of the nasal-infixed verbs in the other IE languages, scholars usually assume that the PIE nasal presents inflected according to the structure *CR-né-C-ti / *CR-n-C-énti (e.g. Skt. bhinátti / bhindánti < *bhi-né-d-ti / *bhi-n-d-énti from the root *bheid-). Indeed, this structure seems to underly the Hittite type (3), e.g. duarnizzi / duarnanzi < *dhur-né-h1-ti / *dhur-n-h1-énti and šannai / šannanzi < *sn-nó-h1-ei / *sn-n-h1-énti.

The other two types are less clear regarding their interpretation, however. Type (1) seems to reflect the structure *CR-nÕn-C-ti / *CR-nVn-C-énti. Despite attempts by several scholars to derive this type out of the classical model, I know of no convincing solution for this type. Type (2) seems to reflect the structures *CR-én-C-ti / *CR-n-C-énti and *CR-ón-C-ei / *CR-n-C-énti, respectively. To my knowledge, no attempts have been made to explain this type.

In my view, the three types cannot be treated without reference to each other. Moreover, it is significant that each type of nasal infix corresponds to a specific type of verbal root: type (1), -nin-, is formed of roots of the structure *CeRK- and *CReK- where R-m- and K = any velar; type (2), -Vn-, is formed of roots of the structure *CemK-; and type (3), -nV-, is formed of roots that end in a laryngeal. This must be a starting point for treating the prehistory of the Hittite nasal infixed verbs.


Typologically speaking, infixation is a rare phenomenon and always the result of epenthesis. It is therefore attractive to assume that the nasal infix as attested in the IE languages derives from an earlier n-suffix.[57] In view of the athematic i-presents *t-éi-ti / *t-i-énti[58] and *dhh1-ói-e / *dhh1-i-ḗr or the original form of the s-presents, *CC-és-ti / *CC-s-énti,[59] it is likely that in (pre-)PIE, the structure of the n-suffixed verbs was *CRC-én-ti / *CRC-n-énti and *CRC-ón-e / *CRC-n-ḗr, respectively. In order to derive the structures as attested in Hittite and the other IE languages, we can envisage the following scenario (exemplified by the roots *h3erg-, *temk- and *dhuerh1-).


(1) Original situation:


   *h3rg-én-ti, *h3rg-n-énti            *tmk-én-ti, *tmk-n-énti           *tih1-én-ti, tih1-n-énti



(2) In the forms with *CRC-n-, n-epenthesis occurs: the stops preceding -n- become prenasalized:[60]


   *h3rgénti, *h3rngnénti                *tmkénti, *tmnknénti               *tih1énti, *tinh1nénti



(3) The prenasalized stop of the plural spreads throughout the paradigm:


   *h3rngénti, *h3rngnénti               *tmnkénti, *tmnknénti              *tinh1énti, *tinh1nénti



(4) The cluster *-nCn- is simplified to -nC-:


   *h3rngénti, *h3rngénti                *tmnkénti, *tmnkénti               *tinh1énti, *tinh1énti



(5) Under pressure of the plural forms, which seem to contain a root *CRnC-, the singular stem *CRnCen- metathesizes to *CRnenC-:


   *h3rnéngti, *h3rngénti                *tmnénkti, *tmnkénti               *tinénh1ti, *tinh1énti



At this stage, the Anatolian branch splits off from Proto-Indo-European. In the latter group only one further development takes place:


(6a) The nasalized consonants lose their nasalization, which leads to the classical model *CR-né-C-ti / *CR-n-C-énti:


   *h3rnégti, *h3rngénti                 *tmnékti, *tmnkénti                *tinéh1ti, *tinh1énti



In Anatolian, the following developments take place:



(6b) The laryngeals lose their nasalization, and the cluster *Cmne- is assimilated to *Cme-:


   *h3rnéngti, *h3rngénti                *tménkti, *tmnkénti                 *tinéh1ti, *tinh1énti



(7) The nasalization of the velars develops into a real nasal consonant:


   *h3rnéngti, *h3rngénti               *tménkti, *tmnkénti                *tinéh1ti, *tinh1énti



(8) On the basis of the full grade stem *CRnenK- in type (1), the zero grade stem *CRnK- is changed to *CRnnK-


   *h3rnéngti, *h3rnngénti             *tménkti, *tmnkénti                *tinéh1ti, *tinh1énti



(9) In the sequence *CNNC an anaptyctic /¸/ develops:


   *h3rnéngti, *h3rn¸ngénti            *tménkti, *tm¸nkénti               *tinéh1ti, *tinh1énti



(10) *-nenK- > -ninK-


   *h2rníngti, *h2rn¸ngénti             *tménkti, *tm¸nkénti               *tinéh1ti, *tinh1énti



(11) *e/inCC > *e/iCC


   arnikzi, arninkanzi                tamekzi, tame/inkanzi             zinnizzi, zinnanzi

   /Hrníktsi, Hrn¸nkántsi/                /tméktsi, tm¸nkántsi/                /tsiNítsi, tsiNántsi/


[1] Different words show different root-vowels, so at this point it is unclear if there originally was one type only, and how it must have looked like.

[2] Taken over from i-stem nouns.

[3] If -ai really is an old ending, and not of secondary origin. See s.v. -i ( ending) for discussion.

[4] Taken over from consonant stems.

[5] In stems in -t-, this ending is written -z = /-ts/.

[6] See Weitenberg 1995 for a detailed account of the sigmatization of original asigmatic commune forms.

[7] Taken over from i-stem nouns.

[8] Taken over from the thematic nouns.

[9] The proterodynamic ending is very rare: it has virtually everywhere been supplanted by hysterodynamic -aš < *-os.

[10] Note that contra Eichner 1973 I do not see any reason to reconstruct *-ē- in the form.

[11] The fact that the cluster *-mn- does not seem to have left traces in the other IE languages cited above points to a late-PIE assimilation of *-mn- to *-m- as is visible in the Ved. of -man-stems: e.g. raśmán- has raśmā́ and drāghmán- has drāghmā́, both from *-mn-oh1. When the preceding root contained a labial consonant, the cluster -mn- was assimilated to -n-: Ved. prathinā́ from prathimán-, pre°ā́ from premán-, bhūnā́ from bhūmán-, mahinā́ from mahimán- and vari°ā́ from varmán-; but also Skt. budhná- ‘bottom’ < *bhudh-mno- as visible in Gr. πυθμήν ‘bottom’; Av. raoγna- ‘butter’ < *Hre/ough-mno- as visible in Icel. rjómi ‘cream’ < *reugman- and MHG rōme < *raugman-; PGerm. *bragna- ‘brain’ < *mrogh-mno- as visible in Gr. βρέχμα ‘skull’ (last examples taken from Kroonen 2006).

[12] The ẽ- of Lyd. gen.adj. ẽmi- is the regular outcome of raising of *a- due to the following -i-.

[13] Which has a variant amu due to u-umlaut.

[14] With /g/ on the basis of ú-ke-el, ú-ki-la ‘I (emph.)’.

[15] Note that **-ih2 would have yielded **-e (cf. ke-e ‘these’ < *ih2).

[16] Including Tocharian where TochA tu, TochB tu, tuwe reflect *tuH(om).

[17] Which is a very common development, compare e.g. the Luwian languages where PAnat. nom. *ʔú ‘I’was replaced by acc.-dat. *ʔMúg ‘me’.

[18] Thus already Cowgill 1965: 16956. The fact that the Anatolian branch retained the older situation, *tiH, *tu-, whereas all the other IE languages (including Tocharian) show the innovated system *tuH, tu-, is an argument in favour of the view that the Anatolian branch was the first one to split off from PIE, cf. ḱ 0.6.

[19] The views on the Anatolian 2pl. pronouns as expressed by Katz (1998a) did not convince me.

[20] In the literature, we often find the view that the plene spelling of e in the singular forms (e.g. e-eš-mi ‘I am’, e-et-mi ‘I eat’ or še-e-eš-mi ‘I sleep’) of some of these verbs indicates original length and points to acrostatic (i.e. ‘Narten’) inflection (e.g. Oettinger 1979a: 87, but also still LIV2, where e.g. e-et-mi is given as ēdmi, reflecting **h1ēd-mi). This view must be abandoned. The plene spelling only indicates the fact that *e is accented. For each verb, cf. its respective lemma for my view that all e/a-ablauting verbs go back to normal root-presents with *e/Ø-ablaut.

[21] A common other view is that these verbs introduced the a- in the weak stem in analogy to šeš-zi / šaš- ‘to sleep’ (e.g. Melchert 1994a: 66-7, Kimball 1999: 390). This, however, is highly improbable: it is hard to believe that in Hittite a wide-scale leveling within the paradigm of verbs like ‘to be’, ‘to eat’ and ‘to drink’ took place in analogy to one less frequent verb only. Moreover, the -a- of šaš- probably is an empty vowel as well.

[22] Taking the sign A as having the value ’ax as is known from Boğazköy Akkadian (cf. Durham 1976: 117).

[23] Synchronically in Hittite, the normal form of 1pl.pret.act. is CeC-en: e-ep-pu-en, e-šu-en, e-du-en, e-ku-en, še-eš-u-en etc. Nevertheless, the original form probably was CC-en, as still seen in ap-pu-en ‘we seized’ (KUB 34.77 obv. 2 (OH or MH/NS)), ú-e-tu4-me-en ‘we built’ (KBo 4.1 i 28 (NH)), u-ul-lu-mé-en (KUB 23.21 obv. 29 (MH/NS)), u-ul-lu-um-me-[en] (KBo 3.15, 6 (NS)) ‘we smashed’, and possibly u-u-ga-u-en ‘we conjured’ (KUB 18.12 obv. 13 (NH)). This CC-en corresponds to the zero grade form that we find in the i-conjugated verbs.

[24] Synchronically in Hittite the normal form of 2pl.pret.act. is CeC-ten, cf. e-ep-te-en, e-eš-te-en, ku-en-ten, etc. Nevertheless, on the basis of the original zero grade in 1pl.pret.act. and on the zero grade forms in the preterite plural of i-conjugated verbs I assume that originally this form was CC-ten.

[25] Synchronically in Hittite, the normal form of 3pl.pret.act. is CéC-er, however: e-ep-pér, e-ku-er, e-še-er, e-te-er, ku-e-re-er, še-e-š[e-er]. Nevertheless, on the basis of the original zero grade in 1pl.pret.act. and on the zero grade forms in the preterite plural of i-conjugated verbs I assume that originally this form was *CC-ḗr as well. This *CC-ḗr is possibly attested in ú-a-te-er, ú-e-te-er and pé-e-u-te-er although these verbs in principle could reflect both *dhh1-ḗr as well as *dhéh1-ēr.

[26] See the lemma -llu, -lit for a detailed treatment of the formation of the 1sg.imp. form.

[27] The archaic i-it-te-en ‘you must go’ < *h1i-té (Gr. ἴτε, Skt. itá) clearly shows that the original form was CC-tén. Synchronically in Hittite, the normal form of 2pl.imp.act. is CeC-ten, however: ku-en-te-en, e-ep-te-en, etc.

[28] The reconstruction *-eh1- goes back to Watkins 1973a. Recently, Jasanoff (2002-03: 147) has stated that a reconstruction *-eh1-e/o- is possible as well, assuming that *-eh1-ti and *-eh1-e-ti both would yield Hitt. -ezzi. In view of the development *VHiV > OH /ViV/ > NH /VV/ as described in ḱ, this is incorrect, however. A paradigm *CC-eh1-ie-ti / *CC-eh1-io-nti would regularly have yielded OH **/CCeietsi / CCeiantsi/, spelled °Ce(-e)-i(-e)-ez-zi / °Ce(-e)-a-an-zi, which further developed into NH **/CCetsi / CCeantsi/, spelled °Ce(-e)-ez-zi / °Ce(-e)-a-an-zi. Since a spelling with -i- does not occur in any of these verbs (only in 3pl.pres.act. na-ak-ke-a-a[n-zi], which must represent /nakeantsi/ < virtual nakkē- + -anzi), we must stick to Watkins’ reconstruction with -eh1-.

[29] As we will see s.v., the suffix -nu- does show traces of original ablaut, however: the forms a-a-nu-ú-mi (KBo 17.1 + 25.3 ii 18 (OS)) and u-e‹-eš›-nu-ú-ut (KBo 3.28 ii 19 (OH/NS)) show that originally the strong stems showed /-nÜ-/ vs. /-nu-/ of the weak stem, reflecting *CC-néu-ti / *CC-nu-énti.

[30] Note that loss of nasal does not occur in verbs like kānk-i or amank-i / ame/ink- (e.g. ga-a-an-ga-a-i /kā́nkHi/, a-ma-an-ga-a-i /Hmā́ngHi/), but this is in my view due to the fact that we are here dealing with a preceding /ā/. The fact that in /ānCC/ the nasal is retained whereas in e.g. */inCC/ it is lost in my view is comparable to the fact that in Lycian the vowel a has a nasalized variant ã, but the vowel i does not: low vowels apparently were better in retaining a following nasal element than high vowels.

[31] Oettinger 1979a: 259f. also distinguishes an “einfach thematische Klasse”, but all verbs that he regards as belonging here should be interpreted otherwise: “ulle-” = ulle-zi / ull- (Ia1); “išparre-” =, on the one hand, išpār-i / išpar- ‘to spread out’ (IIa2) and, on the other, išparra-i / išparr- ‘to trample’ (IIa1γ); “lukke-” = lukk(ie/a)-zi (Ib1 / Ic1); “malle-” = malla-i / mall- (IIa1γ); “papre-” = paprē-zi (Ib2); “šarre-” = šārr-i / šarr- (IIa2); “šulle-” = šullē-zi (Ib2); “šue-” = šue/a-zi (Ic4); “ašše-” = ašše/a-zi (Ic5).

[32] Thus in pé-eš-ši-a-u-e-ni (KUB 35.164 obv. 6 (OS)), although we would expect -aani as attested in i-a-u-a-ni (KBo 3.8 ii 24 (OH/NS)).

[33] Thus in a-ap-pa-ri-e-nu-un (OS), a-ni-e[-nu-un] (OS), but compare pé-eš-ši-a-nu-un (OH/MS).

[34] Unattested in OS texts, but compare 2sg.pret.act. i-e-eš (OH/NS).

[35] We would expect **-etten, but this ending is not attested. The 2pl.pret.act. ending is attested in NS texts only as -a-at-te-en and -a-at-ten (both OH/NS).

[36] Compare u-it-ti (OH/MS). This ending is replaced by -a in MH times already (e.g. i-a ‘do!’ (MH/MS), ti-a ‘step!’ (MH/MS)).

[37] Note that *ó > Hitt. short /á/ in internal syllables.

[38] Perhaps reflected in the one attestation da-aš-ga-nu-un (KUB 13.35+ i 40, 44 (NS)).

[39] E.g. ak-ku-uš-ki-i ‘drink!’ (KBo 7.28 obv. 23 (OH/MS)), az-zi-ik-ki-i ‘eat!’ (KBo 7.28 obv. 23 (OH/MS), KBo 21.60 rev. 15 (OH/NS)), uš-ki-i ‘see!’ (KBo 25.123, 10 (OS)).

[40] Note that this did not happen in verbs of the structure *Ceh2-, which ended up in class IIa2 (-i / naḫḫ-,-i / zaḫḫ-). The different outcome of *CóCh2ei > CaCai, on the one hand, and *Cóh2ei > i, on the other, is due to the fact that in the former verb the regular loss of *h2 after consonant caused phonologization of the ending /-ai/, whereas in *Cóh2ei, which despite its phonetic pronunciation [Cā́hai] phonologically remained /Cā́hei/ due to the presence of --, we are dealing with the ending /-ei/. This ending then regularly developed into /-e/ and later on was replaced by -i, so *Cóh2ei > */Cā́he/ >> i.

[41] Thus in pí-u-e-ni. All other verbs show CC-a-eni according to class Ic1.

[42] Thus in pí-iš-te-ni and zi-iš-te-e-ni. Secondary CC-oi-teni (with the mi-ending -tteni) in ta-a-it-te-ni.

[43] Thus in iš-pí-iš-te-en, al-zi-iš-te-en and pí-iš-te-en. Secondary CC-oi-sten in da-iš-ten, and pa-iš-ten.

[44] But compare 3pl.imp.act. a-ra-an-du from arai-i / ari- ‘to (a)rise’. Does this form and the ones mentioned in notes 268 and 273 point to a situation in which non-finite forms of this paradigm were -i-less?

[45] But compare part. a-ra-an-t- from arai-i / ari- ‘to (a)rise’, cf. note 267.

[46] Thus al-zi-u-en and pí-u-en. Secondary CC-oi-en in da-i-u-en.

[47] Thus ti-u-a-an-zi. All other verbs have CC-a-anzi according to class Ic1.

[48] This form on the basis of 2pl.imp.act. *CC-i-stén. Secondary *CC-oi-sten in iš-a-iš-te-en.

[49] Thus in al-zi-i-e-er, u-i-e-er, iš-i-i-e-er, iš-pí-i-e-er and pí-i-e-er. Secondary CC-oi-ēr in da-i-e-er and u-a-a-er.

[50] *CC-ské/ó- is the only possible reconstruction for zaške/a- /tske/a-/, zikke/a- /ts¸ke/a-/ < *dhh1-ské/ó-, but also alziške/a- besides alzeške/a- and piške/a- besides peške/a- point more to an interpretation *Cəske/a-, reflecting *h2lt-ské/ó- and *h1p-ské/ó-, than to *C-i-ske/a- from *h2lt-i-ské/ó- and *h1p-i-ské/ó-. Cf. note 267.

[51] See s.v. for a treatment of the 2pl. ending -šten(i).

[52] The form *CóC-e is still seen in 3sg.pres.act. CóC-i < *CóC-e-i. For a treatment of the 3sg.pret.act. ending (*CóC-s), see s.v.

[53] Usually, this ending is reconstructed as *-é on the basis of Skt. . See the lemma -šten(i) and Kloekhorst fthc.c, however, for the possibility that Hitt. -šteni together with the PToch. 2pl.pret. ending *- points to a PIE ending with an element -s-.

[54] Note that the spelling ú-a-tar phonologically stands for /ʔuā́dr/ < *Hu-ó-tr, which contrasts with the spelling a(-a)-tar ‘water’ that denotes /uā́dr/ < *uódr without initial glottal stop.

[55] See also Oettinger 1976b, who states that the ending *-o originally belonged to ‘statives’ and the ending *-to to ‘middles’ (pace e.g. Jasanoff 2003: 51, who merely sees “*-to(r) as a modernized form of *-o(r)”).

[56] The verbs kinae-zi < *ki-né-h2-ti / *ki-n-h2-énti and munnae-zi < *mu-né-h2/3-ti / *mu-n-h2/3-énti form their own sub-category. See at their lemmata for further treatment.

[57] Cf. Thurneysen 1883: 301-2.

[58] Cf. Kortlandt 1989a: 109.

[59] Cf. Pedersen 1921: 26.

[60] Cf. Thurneysen 1883 for the Latin phenomena that can be explained by prenasalization (the mediae found in Lat. pandō from *peth2-, pingō from *pei-, ē-mungō from *meuk-, mandō from *meth2-, etc.), and cf. Kortlandt 1979: 61 for prenasalization in e.g. the BSl. stem *undn- ‘water’ < *ud-n-.