1. Vowels

9) a) It is not always possible to read with certainty the vowel e (as in Akkadian). The signs of the syllables me, ne, el, eš differ from those of the syllables mi, ni, il, iš, but the signs for re, le, ez, etc..., are also used for ri, li, iz, etc...

b) The existence of a vowel o differentiated from u in writing is unclear.

10) Even when Hittite can differentiate in writing between e and i, both vowels are often found. Next to e-eš-har "blood", one also find written iš-har ; next to pé-eš-ta "he gave", pí-iš-ta ; next to pé-eš-ši-ia-mi "I threw", pé-eš-ši-ia-zi "he threw", one also finds pé-eš-še-ia-mi, pé-ši-ia-az-zi ; next to -ši "to him", also -še ; next to iš-hi-i "to the lord", also eš-hé ; next to u-un-nu-me-en "we pushed her", also u-un-nu-um-mi-in, etc... It is probable that the pronunciation of the Hittite e was very closed, close to that of i.

11) There exists an alternation between e (i) and a in the Hittite paradigms : from sak- "to know" are derived the forms saggai "I know", sakti and sekti "you know (Sg.)", sakki "he knows", sekteni "you know (Pl.)", sekkanzi "they know", sakta and sekta "he knew" ; from ak- "to die", aki "he dies", akkanzi "they die", but akir and ekir "they died" ; and from asas- "to sit", asāsi "he sits", but asesanzi "they sit", asasta and asesta "he sat", asesir "they sat".

watar "water" has a Gen. Sg. wetenas and a Nom.-Acc. Pl. widār, tekan "earth" a Gen. Sg. taknas. Next to esmi "I am" and eszi "he is", one finds asanzi "they are" ; next to ekuzi "he drinks", akuwanzi "they drink" and the iterative akkusk- "to drink copiously" ; next to mekki- "much", a verb makkeszi "he does much". Instead of paiweni "we go", paitteni "you go (Pl.)", one can find paiwani and paittani ; instead of daskitteni "you regularly take (Pl.)", daskatteni ; instead of piskir "they regularly gave", piskar. The reason for this alternation is not well understood.

12) There are sometimes variations between u and ú : a-pu-u-un and a-pu-ú-un "this" (Acc. Sg. of apā- "this"), u-i-ia-at-tin and ú-e-ia-at-tin "send! (Pl.)", da-a-ú and da-a-u "he should take".

13) a) The diphtongue -ai- can be contracted to -e- (-i-) : paista and pesta "he gave", naisut and nesut "turn! (Sg.)", kappuwāit and kappuet "he checked", kappuwāizzi and kappuizzi "he checks".

b) A reverse diphtongation sometimes occurs by analogical correction where e is written ai : instead of epta "he grabs", one finds a-ip-ta ; instead of meggaus "much", ma-iq-qa-us.

c) An alternation between -āi- and -a- sometimes occurs : pāisi and occasionally pāsi "you go (Sg.)".

14) a) 1. -(i)ya- can be reduced to -e- (-i-) : memiyani and memini "for the word", tiezzi and tizzi "he enters", wemiyat and wemit "he found", tiyantes "those who are sitting" (from dāi- "to sit") and tintes.

2. From sankuwai- "nail", one finds the Gen. Sg. sankuis next to the neighboring form sankuwayas (and sankuwas).

b) Next to iskiyazi and iskizzi "he anoints", one finds the poorly understood form iskiyāizzi.

15) a) -aya- can be contracted to -a- : Gen. Sg. of salli- "tall" sallayas and sallas, Abl. Sg. of suppi- "pure" suppayaz(a) and suppaz(a).

16) The diphtongs ue- (ui-) and -ue- (-ui-) can be contracted to u- and -u- : ueter and uter "they brought ", uinut and sometimes unut "let leave! (Sg.)", kuera- and kura- "entrance", karuili- and karuli- "old", atraweni and atrauni "we write", parkueszi and sometimes parkuszi "he becomes pure".

17) a) In the same way, (u)wa- and -(u)wa- can be reduced to u- and -u- : antuwas and antuas "man", awari- and auri- "border post", lauwatin and lautin "pour! (Pl.)" (also lauwai and laui "he pours"), uwartas and urtas "he cursed", waranu and uranu "he should burn down", sanuwanzi and sanunzi "they roast" (Part. sanuwant- and sanunt- "roasted").

b) More rarely, -uwa- (uwa-) is contracted to -ue- (ue-) : kappuwanzi and kappuenzi "they check", uwanzi and uenzi "they come".

c) At the opposite of a), the initial u- can be lengthened to uw- : uwarkant for warkant- "fat", uwasta- for wasta- "to sin", uwitar for widār (N.-A. of watar "water").

18) The variations of the previous forms should not be confused with the regular I.E. ablaut that connects kuénzi "he strikes" to kunánzi "they strike" and kuerzi "he cuts" to kuranzi "they cut". The ablaut also connects dāi "he sits" to tiyanzi "they sit", the Nom. Sg. zahhāis "battle" to Gen. Sg. zahhiyas (§69), ais "mouth" to Dat.-Loc. Sg. issi (§87), as well as tekan "earth" to Gen. Sg. taknas (§78), hannessar "business" to Gen. Sg. hannesnas (§84), asawar "enclosure" to Dat.-Loc. asauni (§85).

2. Consonants

a) Consonants groups

19) a) There is no rule to discriminate in the writing between simple and doubled consonants. Next to the forms pessiyazi "he throws", iyattari "he goes", istamasti "you hear (Sg.)", memiyani "word (Dat.-Loc. Sg.)", innarawanni "vigor (?) (Dat.-Loc. Sg.)", tarnatti "you let (Sg.)", one also finds pisiyazzi, iyatari, isdammasti, memiyanni, innarawani, tarnati. It is probable that the scribes were inclined to omit complex cuneiform signs when they felt that they were not necessary.

b) The difference between simple consonant and doubled consonnant corresponds to an opposition between fortis consonant and lenis consonant. This opposition probably results for stops in an opposition unvoiced / voiced. For example, attas "father" is pronounced /atas/, whereas apēz "consequently" is pronounced /abēz/.

Stops are supposed to be always fortis in initial position and lenis in final position : genu "knee" = /kenu/ and sipant "libation" = /spand/. The "law of Sturtevant" explains that in median position, the sequence of cuneiform signs V-CV points out a lenis consonant, whereas a sequence VC-CV points out a fortis consonant. E.g. ap-pa-tar = /apadar/.

20) a) The choice between a voiced or unvoiced cuneiform sign seems completely arbitrary : da, di, du can be found instead of ta, ti, tu ; ga (qa), gi, gu instead of ka, ki, ku ; ba instead of pa. Some examples : damai- and tamai- "other", atta- and adda- "father", -ti- and -di- "your (Sg.)", esdu and estu "he must be", kanes- and ganes- "to discover", kinu- and ginu- "to leave", taggasta and takkista "he gathers", KUŠkursa- and KUŠgursa- "skin", daskatten, dasqaten and daskiten "take! (Pl.)", Gen. Sg. of kuiski "someone" kuelka, kuelga and kuelqa, patili- and batili- (a priest).

Some words are however always written in the same manner. For example, one finds written gi(-e)-nu "knee" (and not *ki-e-nu-), dāi "he sits", but tiyanzi "they sit". In the lexicon, the voiced b, d, g will not be separated from the unvoiced p, t, k.

b) The Hittites apparently borrowed the cuneiform writing from the Hurrians who themselves had borrowed it from the Old-Akkadians. This could explain why the Hittite writing system did not differentiate between the voiced and the unvoiced consonants, along with the usage of š for the phoneme /s/, matching the Old-Akkadian writing of the sibilants. The Hittite writing system was thus different from the one in usage at the same time in Mesopotamia. As an example of the difference in the writing of the stops between Hittite and Akkadian, the Akkadian-Hittite lexical list KBo I,45 contains at line 11 : ṣa-pa-du (for the Akkadian verb ṣabātu) = ap-pa-tar.

21) A limitation of the cuneiform writing comes from the fact that cuneiform signs can only represent syllables of the kind consonant + vowel (CV), vowel + consonant (VC) and consonant + vowel + consonant (CVC). Hence, it is neither possible to write a group of two initial or final consonants, nor to write a group of three consonants, for example *tri- "three" (initial), *link "swear! (Sg.)" (final), *karp-zi "he lifts" (median). The Hittite scribes overcame this limitation by inserting an extra vowel, especially at positions normally forbidden by the rules of cuneiform writing : te-ri-, li-in-ik and kar-ap-zi. Indeed, in classical Akkadian cuneiform writing, the last phoneme of a phonetic sign must be in harmony with the first phoneme of the following phonetic sign in such a way that a (non-initial) sign starting with a vowel must be preceded by a sign ending with a vowel (CV-VC but not CV-CV). One should always keep in memory that the writing always exhibits more vowels than the spoken language. However, it is not always easy to know whether a vowel is real or not.

22) a) The interpretation in median position is generally easy : forms such as ša-an-ah-zi "he searches" have an incorrect spelling in "usual" cuneiform writing (with a final consonant and an initial vowel following), which indicates that the pronunciation is not *sanahzi which would be written *ša-na-ah-zi. Furthermore, a form such as ša-an-hu-un "I searched" points out a root sanh-, which brings to a spoken form *sanhzi "he searches". Many verbs are similar to sanh-, for example parh- "to hunt" (pár-ah-zi "he hunts"), karp- "to lift" (kar-ap-zi "he lifts"), tarh- "to hold in check" (tar-ah-zi), walh- "to strike" (wa-al-ah-zi), warp- "to bathe" (wa-ar-ap-zi), etc...

However, unclear forms such as ša-an-ha-zi "he searches", wa-ar-pa-zi "he bathes", etc... can be found. Moreover, for hinkzi "he presents" (root hink-), one finds hi-in-ik-zi, hi-in-ga-zi and hi-ik-zi (with a reduced n ; §31a), for linkt(a) "he swore" (root link-) li-in-ik-ta, li-in-kat-ta and li-ik-ta (§157), iterative forms such as a-ar-aš-ki-it for *ar-sk-it "he came several times" (root ar-).

b) Initial forms are less clear like az-zi-ik-kán-zi "they adore" for *at-sk-anzi (iterative of ed- "to eat" ; §141b), ši-pa-an-za-ki-iz-zi "he makes several sacrifices" for *spant-sk-izzi. One finds even less clear forms : from *(i)spart- "to escape", iš-pár-za-zi "he escapes", (i.e. *(i)spart-s-zi with z = /ts/, §27a), iš-pár-za-aš-ta "he escaped" (i.e. *(i)spart-s-t) next to iš-pár-te-er "they escaped" (i.e. *(i)spart-er). From hat- "to dry" (hāti "it dries", hāter "they dried"), one finds ha-az-ta and ha(-az)-za-aš-ta "it dried" (i.e. *hat-s-t(a)). One also finds unclear iterative forms of tar- "to say" : tar-aš-ši-ki-iz-zi "he repeated" (i.e. *tar-sk-izzi) and tar-aš-ša-aš-ki-id-du "he must repeat" (i.e. *tar-sk-iddu !)

23) a) These extra vowels sometimes occur even though they are not necessary in the writing : e.g. ša-an-hu-un and ša-an-ah-hu-un "I searched" (root sanh-), wa-al-hi-ir and wa-al-ah-hi-ir "they stroke" (root walh-), kar-pa-an-zi and kar-ap-pa-an-zi "they lift" (root karp- ; §157). The reason for this phenomenon is unclear ; it can be analogical to forms such that ša-an-ah-ta "he searched", wa-al-ah-zi "he stroke", etc.., or really express a phonetically double consonant *sanhhun, *walhher, or indicate a mute vowel *san(Ə)hun, *wal(Ə)her, or in the case of the alternation h/hh two different consonants (§28). For arhun "I arrived" (root ar-), the transcription a-ar-ah-hu-un next to a-ar-hu-un can be found.

b) An incorrect spelling of syllables is not always caused by an extra vowel ; the vowel can be real with a simplified spelling of the word. The form kiš-an "as follows" is often found next to ki-iš-ša-an, more rarely ma-a-ah-an next to ma-ah-ha-an "as", hi-ip-pár-aš ("prisoner"), ši-iš-at-ti for *ši-iš-ša-at-ti "you destroy (Sg.)", šu-up-ia-ah "clean! (Sg.)" next to šu-up-pí-ia-ah,dIa-ar-iš next to dI-ia-ar-ri-iš, mar-mar-aš "bush (??)" next to Dat.-Loc. Sg. mar-mar-ri, etc...

c) Finally, one can find correct transcriptions according to syllable splitting, even though the vowel does not actually exist : ša-na-ah-ti "you searches (Sg.)" (for *sanh-ti) and wa-la-ah-ši "you strikes (Sg.)" (for *walh-si).

24) The interpretation of the writing at the initial position is mainly based on assumptions. For zi-ik-kán-zi "they put", the analogy with az-zi-ik-kán-zi "they adore" (= *at-sk-anzi ; §22b) allows to posit the form *t-sk-anzi (with a reduction t- of the root dāi- "to put, to place"). For pa-ra-a "ahead, forward", the etymological link with I.E. *pro allows to assume a pronunciation *pra, whereas the form te-ri-ia-al-la (a liquid) has a variant 3-ia-al-la which lets assume a pronunciation *triyalla (§129b3).

It is probable that the very frequent (written) initials isp- and ist- (e.g. ispāi- "to be content", ispant "night", ispart- "to escape", istap- "to lock", istamas- "to hear", istandāi- "to hesitate", istark- "to become ill", etc...) are in fact pronounced sp- and st-, even if it is also possible that it could be a prothetic vowel (cf. lat. scalas > fr. échelle).

25) a) 1. The interpretation of the finals is not always as clear as the imperatives li-in-ik "swear! (Sg.)" (next to li-in-ki, root link-), wa-al-ah "strike! (Sg.)" (root walh-), ša-an-ha "search! (Sg.)" (next to ša-a-ha, root sanh- ; §157). For example, the analogy brings to postulate from the ending of Pret. 3 Sg. in -t of verbs with vocalic ending such as iya- "to do ", hatrāi- "to write" (iyat "he did", hatrāit "he wrote") that the written ending -ta of Pret. 3 Sg. of verbs with consonant ending such as es- "to be", istamas- "to hear", walh- "to strike" (e-eš-ta "he was", iš-ta-ma-aš-ta "he heard", wa-al-ah-ta "he stroke") represents in fact a (spoken) ending -t (i.e. *est, *(i)stamast, *walht).

2. The same phenomenon occurs for the substantive : from aniyat(t)- "vigor" (Acc. Sg. aniyattan), the Nom. Sg. a-ni-ia-az (§76a) is interpreted as a form *aniyat-s (with z = /ts/ ; §27a). This allows to interpret the Nom. Sg. ka-aš-za "hunger" (root kast- ; Acc. Sg. kastan) as a form *kast-s, likewise for ša-ú-i-ti-iš-za "baby" interpreted as *sawitist-s. The written forms Nom. Sg. hu-u-ma-an-za "whole" and participles a-da-an-za "eaten" (roots humant-, adant-) are interpreted as spoken forms *hūmant-s, *adant-s.

b) However, when the particle -a "and, also" is added to the Nom. Sg. of a participle in -an-za = *-ant-s, the ending, spoken as *-ants-a, is not written simply -an-za, but rather -an-za-ša, e.g. ir-ma-la-an-za "ill", ir-ma-la-an-za-ša "ill too". In fact, to -an-za = *-ants is added the smallest phonetic unit ending the group *-antsa, that is the sign ša = *sa (§27b). The ending -an-za-aš-ša is also used : ap-pa-an-za "the prisoner", ap-pa-an-za-aš-ša "and the prisoner".

26) It seems that the groups of consonants could be split by inserting vowels really pronounced. Thus, one finds close written forms like gimra- and gimmara- "field", kussani and kusni "as a reward" (also kussansit, kussanissit and kussasset "his reward" §31a), assanu- and asnu- "to prepare", kar-ša-nu-, kar-aš-nu- and kar-aš-ša-nu- "to miss", tuhs- and tuhhus- "to carve", nasma and nassuma "or". The distinction with the case of unspoken vowels (§22ff.) is not always clear : from taks- "to link" is derived the Part. ták-ša-an-za "linked", but ták-ke-e-eš-ša-an-zi "they link" ; e-eš-har-šum-mi-it "their blood".

b) Isolated consonants

27) From the four sibilants z, s, š, ṣ of the Akkadian cuneiform writing, Hittite only uses š and z. The letter š is used, as in Assyrian, to denote the sound /s/, while the letter z denotes the sound /ts/.

a) z = /ts/ can be demonstrated by comparing forms like da-šk-izzi "he takes several times" (iterative of dā- "to take") and azzikkizzi = *at-sk-izzi "he adores" (iterative of ed- "to eat"), as well as the Nom. Sg. aniyaz = *aniyat(t)-s "vigor" compared with the Acc. Sg. aniyattan (root aniyat(t)-).

b) š = /s/ can be demonstrated by Egyptian incriptions like Mrsr for mMuršili, Htsr for mHattušili, etc... (the Egyptian writing differenciates s and š), as well as the comparison of the forms da-šk-izzi and azzikkizzi = *at-sk-izzi.

c) On the other hand, the extent of the difference between š and z is not well known for Proto-Hatti (e.g. the name of the town URULihšina next to URULihzina) and Palaic, even for Hittite. Indeed, one finds close forms such as šakkar and zakkar "excrements", zamangur "beard" and šamankurwant- "bearded", zašhi- and zazhi- "dream".

28) It is possible that the Hittite h had two different pronunciations :

a) a (less attested) stronger pronunciation, close to k, e.g. the isolated forms tetkissar, hameskanza instead of the neighboring forms tethessar "storm", hameshanza "spring", and conversely UZUishisa- instead of UZUiskisa- "back". The woman's Hurrian name fGiluhepa is written Krgp in Egyptian.

b) a (well attested) weaker pronunciation (maybe only a breath), as indicated by the variants eshar "blood" (Gen. Sg. eshanas) next to the rarer essar (Gen. Sg. esnas), as well as the isolated forms idalawatti and tannattauwanzi for the neighboring form idalawahti "you act badly (Sg.)", danattahhuwanzi "to devastate".

c) The Hittite h comes from Proto-Indo-European phonemes called "laryngeals" that vanished in all I.E. language families, except the Anatolian family. Many variants of the theory of laryngeals exist that vary on the quality and the number of P.I.E. laryngeals. These laryngealist theories have been built to explain some particular phenomena in I.E., but the later discovery of Hittite has proved their validity.

In the most common theory, P.I.E. had three laryngeals, noted H1, H2 and H3 that could "color" a neighboring vowel 'e'. The laryngeal H1 had no coloration effect, the laryngeal H2 colored in 'a' and the laryngeal H3 colored in 'o'. In Hittite, the laryngeal H1 vanished and the laryngeal H3 was retained only in initial position. In median position, the fricative resulting from a laryngeal can be lenis (written between two vowels by h') or fortis (written between two vowels by 'hh'). For example, eshar "blood" < *esH2er, tar-ah-ha-an (root tarh- "to defeat") < *terH2-, hant "face" < *H2ent, happ-in-ant "rich" < *H3ep-. It should be noted that the theory described here is incomplete : it does not explain cases where Hittite displays a 'h' where there is no laryngeal, and conversely cases where Hittite does not display a 'h' where a laryngeal occured.

According to Kortlandt, for example, initial H2 and H3 remain in front of an 'e' but vanish in front of an 'o' : *H3erbh- > harp- "to separate" but *H3orgh-ey- > ark- "to mount".

29) a) The -w- of the group -uw- often becomes m, especially with verbs in -nu- (§169) and verbs in -u(m)- (§174). From hatrāi- "to write" hatraweni "we write" are constructed the Inf. I hatrawanzi and the verbal substantive hatrawar, but from arnu- "to bring" arnummeni, arnummanzi and arnummar, from tarna- "to let" tarnummeni, tarnummanzi and tarnummar.

b) -mu- is sometimes found instead of -wu- : from idālu- "nasty", one finds the Nom. Pl. Com. idālawes but the Acc. Pl. Com. idālamus. Likewise, one finds from zashāi- "dream" the Acc. Pl. zashimus.

30) a) Hittite has no initial r.

b) In median and final position, r is weakly pronounced, to such a point that it is sometimes omitted in writing. Examples in final position : paprāta instead of paprātar "impurity", miyata instead of miyatar "prosperity", hatressa instead of hatressar "sending" ; in median position : waggantes instead of wargantes (Nom. Pl.) "fat", artati- instead of artarti- "mushroom (?)", pian instead of piran "in front of", TÚGkuessar instead of TÚGkuressar "scarf".

31) In the same way, n in median position is weakly pronounced and is often omitted : hu-u-ma-da-az next to hūmandaz (Abl. Sg. of hūmant- "each"), ú-e-eš-ša-ta for wessanta "they dress", iš-ta-ta-a-it next to istandāit "he stayed", kar-pa-zi next to karpanzi "they lift", ne-e-a-za next to neyanza "driven", me-mi-ia-u-a-zi next to memiyawanzi "to speak", li-ik-ta next to li-in-kat-ta (both = *linkt) "he swore" (root link-). It is possible that the vowel was nasalized.

b) Conversely, an extra n letter is sometimes written though it does not exist : nepisanza instead of nepisaz "from the sky" (Abl. Sg.), hassannanza for hassannaz "out of the family", an isolated form li-in-kán-ta for *linkt "he swore" can be found.

32) a) 1. The group -tn- is regularly transformed to -nn-, especially in the declension of abstracts in -ātar (§83 ; Gen. Sg. -annas < -atnas) : haddulātar "health", Gen. Sg. haddulannas, idālawātar "nastiness", Gen. Sg. idālawannas.

2. The group -tn- remains without change in the following form : huitar "animal life", Gen. Sg. huitnas, very rarely with abstracts in -ātar : harātar "scandal", Dat.-Loc. Sg. haratni.

b) 1. The group -mn- can be assimilated to -m- : next to the ethnic names URUHattusumna- "inhabitant of Hattusa", URULuiumna "Luwian", URUPalāumna- "Palaian", one also finds forms such as LÚ URUZalpūma- "inhabitant of Zalpa", LÚ URUHalpūma- "inhabitant of Alep", Acc. Sg. mSuppiuman next to Dat.-Loc. mSuppiumni.

2. The transformation of -mn- into -nn- is less clear, and it is not sure whether the form hilannas is the Gen. Sg. of hilamnar "gate".

33) The group -nunu- can apparently be shortened to -nu- : from kistanu- "to delete", one finds 1. Sg. Pret. kistanun (for *kistanunun), from *mernu- "to make disappear", 1. Sg. Pret. mernun, from *menunu- "to fail (?)", Imp. 3. Sg. menuddu.

34) The group -nza sometimes changes for unknown reason to -nzan : hanza ep- and more rarely hanzan ep- "to give a friendly welcome (?)", nanza (i.e. nu "and" + -an "him" + -za "self") and nanzan.

3. Liaisons

35) Hittite words are normally written separately, and liaisons are not visible in writing. A variant with liaison such as hal-ki-im pí-an-zi for halkin pianzi "give (Pl.) grain" is a rare exception.

36) a) 1. A final -n- is generally assimilated with the initial consonant of a following enclitic particle (with or without redoubling of this consonant) : *istamanan-san "his ear" (Acc. Sg.) > istamanassan, *halugatallan-tin "your (Sg.) envoy" > halugatallat-tin, *tuzzin-man "my army" (Acc.) > tuzziman, appizziyan-ma-at "but him after" > appizziyamat, hūman "all" + particle -san (§300) > hūmassan, ŠUM-an-smit "their name" (i.e. lāman-smit) > ŠUM-asmit, mān "if" + particle -wa of quotation (§289) > māwa.

2. For example, kussan-set "his salary" can be assimilated to kussasset and stay like that, or instead use an extra vowel (§26) kussanissit.

b) Phenomena of false cut sometimes occur : nassan (nu "and" + -as "he" + particle -san, §300) can be written nansan, as if the Nom. -as "he" was replaced by the Acc. -an "him".

c) A similar assimilation occurs more rarely with t : ta "and" + -at "it" + -si "to him" usually becomes tatsi, but sometimes also tasse, nu "and" + -at "it" + particle -san usually becomes natsan, but also sometimes nassan.

37) Phenomena of liaison especially occur at the beginning of enclitic pronouns such as -mu "to me", -ta "to you (Sg.)", -si "to him", -as "he", -an "him (Acc.)", -at "it (Neut.)" and enclitic particles such as -a (-ya) "and", -ma "but", -asta "then", -wa(r) (quotation), -za (reflexive), -kan and -san (location) after an accentuated word or a particle nu, ta, etc...

38) a) When the particle nu "and" is followed by the pronoun -as "is (ea)", -an "eum (eam)", -at "id", -e "ei, eae, ea", -us (-as) "eos, eas" or the particle -asta or -apa (both = "then (?)"), the u of nu is replaced by the following vowel : *nu-as > nas, *nu-an > nan, *nu-at > nat, *nu-e > ne, *nu-us > nus (*nu-as > nas), *nu-asta > nasta, *nu-apa > napa.

b) The same phenomenon occurs for the older particle ta "and" : *ta-as > tas, *ta-an > tan, *ta-at > tat, *ta-us > tus, *ta-asta > tasta, and for the rarer and older particle su "and" : *su-as > sas, *su-an > san, *su-us > sus.

c) In the same conditions, the particle of quotation -wa(r)- takes the whole form -war- : -war-as, -war-an, -war-at, -wari (i.e. *-war-e), -war-us, -war-asta.

39) When a simple consonant stands between two vowels between a word and an enclitic (the consonant being either the final of the word or the initial of the enclitic), it is often (but not always) doubled : nu + -san > nu(s)-san, sumās "to him" + -an "him" > sumāssan, nu "and" + -wa (quotation) + -nas "us" > nuwa(n)nas, mān "if" + -a "also" > mānna, apās "this" + -a "and" > apā(s)sa (also the Acc. Sg. apūn + -a > apū(n)na), wastul "sin" + -ma "but" + -za (reflexive) + -kan (§295) > wastulma(z)zakan, UL (negation) + -wa (quotation) + -ta "to you (Sg.)" + -kan > UL-wa(t)ta(k)kan.

40) The enclitic pronoun -ta "to you (Sg.)" becomes, when placed before the reflexive particle -za, the form -tu (-du) : nu "and" + -wa (quotation) + -ta + -za + -kan > nuwaduzakan.

41) a) 1. The particle -a/-ya "and" that connects two isolated words takes the form -a after a consonant and -ya after a vowel : dTelipinusa (< dTelipinus-a) "and Telipinu", apāssa (< apās-a ; §39) "and this", but kā-ya "and here", apē-ya "and these".

2. It is generally written -ya after ideograms, Akkadian words and foreign names : ENMEŠ-ya "and the lords", UL-ya "and not", URUKargamis-ya "and Kargamis".

b) After words ending in -z and after all words in the ablative, the form -ya is preceded by a liaison vowel i : tamedazziya (< tamedaz-ya) "and from another", kēz kēzziya "from here and there".

42) Phenomena of simplification of doubled syllables can be observed :

a) -ma "but" + -smas "to you (Pl.), to them" > -masmas which can be shortened to -mas.

b) 1. When the particle -za follows the particle -san, they become simplified in -zan (pronounced *-ts-san).

2. Following §34, nanza (< nu "and" + -an "him" + -za reflexive) can also be written nanzan.

c) 1. When the particle -(a)sta "then (?)" follows the syllables -as, -is, -us, the initial -(a)s can vanish and the particle is reduced to -ta : nu "and" + -wa (quotation) + -smas "to them" + -(a)sta > nuwasmasta, ĜIŠTUKULHI.A-us-sus "your (Pl.) weapons" (Acc. Pl.) + -(a)sta > ĜIŠTUKULHI.A-ussusta.

2. Also after -z : nu "and" + -za (reflexive) + -(a)sta > nuzata (pronounced *nu-ts-sta), kēz "here" + -(a)sta > kez-sta (written ke-e-ez-ta, ke-e-ez-za-at-ta, ke-e-ez-za-aš-ta).