| Proto-Hitatc |
|Period||c. -2800 YP|
|Spoken in||Ici forest|
|Classification|| Hitatc |
|Basic word order||VSO w/ topic fronting|
|Created by||cedh audmanh|
Due to the scarce attestation of early Hitatc languages and the wide divergence of those languages of the family that survived into post-classical times, Proto-Hitatc has proven extremely difficult to reconstruct. A summary of the most securely ascertained features will be given in the following sketch.
The probable consonant inventory of Proto-Hitatc is reconstructed as follows:
|plosives||p · b||t · d||c · ɟ||k · ɡ||q · ɢ||ʔ|
|nasals||m̥ · m||n̥ · n||ŋ̊ · ŋ||ɴ̥ · ɴ|
|laterals||l̥ · l|
|trills||r̥ · r||ʀ|
Several issues require comment:
- The most peculiar feature of the Proto-Hitatc consonant system is the series of voiceless nasals, inferred from the fact that the clearly reconstructible Western Hitatc fricative series /f θ x χ/ regularly corresponds to /h/ or /ş/ preceded by a nasalized vowel in Pirikõsu, and to the nasal consonants /m n ñ ŋ/ in Necine. Evidence from morphophonemic alternations supports this assumption. However, the feature [+nasal] in the eastern branch appears to be absent where alternation relationships would predict a voiceless uvular [ɴ̥] to occur. It is therefore hypothesized that the latter was the first of the series to become denasalized, before the breakup of Proto-Hitatc. This is reflected by listing the corresponding segment as *χ in the above table.
- Similarly, the uvular trill *ʀ exhibits the same morphophonemic alternation with *q that was found between voiced and unvoiced plosives at other POA. It thus clearly corresponds to the missing voiced uvular plosive [ɢ], and is sometimes reconstructed as such. As with *χ, the actually attested reflexes - [ʀ ʁ ɑ] and zero - and their distribution indicate that lenition of this segment most likely predates the Proto-Hitatc stage.
- The voicing contrast for plosives was apparently neutralized in word-initial position, where only unvoiced segments can be reconstructed. However, morpheme-initial plosives preceded by another morpheme within the same phonological word did not exhibit said neutralization.
- The segment traditionally reconstructed as *z based on reflexes in the Wan-Mlir subgroup is listed as *j here, and most instances of traditional *j have been reassigned to the new phoneme *ɰ. This rearrangement is grounded on three pieces of evidence: Firstly, traditional *z does not alternate with *s like all other voiced/unvoiced consonant pairs do. Secondly, the factors determining the apparently unmotivated loss of some *j in Eastern Hitatc have not yet been satisfactorily explained. Relegating so-called "weak j" to a separate phoneme solves this problem. And thirdly, some seemingly irregular vowel correspondences make sense if one assumes three semivowels *ʋ *j *ɰ which were capable of transferring the features [+labial], [+palatal], and [+velar] to preceding reduced vowels in the eastern branch of the family.
- Last but not least, it should be mentioned that a second laryngeal segment *ʔ has been proposed by some scholars on the basis of phonotactical considerations. Most significantly, a few cases of unexpected vowel epenthesis in compounds might be explainable by positing a phonemic glottal stop. However, as the phonetic reflex of *ʔ would generally be identical to that of *h, the evidence for its existence is so far unconclusive.
The situation with vowels is more straightforward. All reconstructions agree on the five cardinal vowels *i *e *a *o *u. In addition, almost all seemingly irregular vowel correspondences can be resolved by positing a reduced vowel *ə, which would then assimilate to a following approximant or to the vowel of the next syllable in Eastern Hitatc.
Various suprasegmental vowel features have been posited, such as length, nasality, and/or devoicing. As it turns out, none of these are necessary if a) voiceless nasals, b) phonemic *ə, and c) underlying morpheme-final consonants are accepted.
Vowels in unstressed morpheme-final syllables harmonized with the next syllable within the same phonological word. This was most evident in prefixes, but it also applied to root-final syllables in compounds. Stem-final vowels before a suffix were usually not affected, nor was *ə, curiously (see below). The assimilation pattern seems to hint at an earlier three- or four-vowel-system in which *e *o were /ai au/ respectively. It operated in such a way that adjacent vowels of the same backness could only be one height level apart, and vowels of different backness had to be of the same height, with *a being unspecified for backness. The rules can be summarized as follows:
- *a → *e / _Ci, _Cəj
- *a → *o / _Cu, _Cəʋ
- *i → *e / _Ca, _Co, _Cəɰ
- *u → *o / _Ca, _Ce, _Cəɰ
- *e → *i / _Cu, _Cəʋ
- *o → *u / _Ci, _Cəj
The behaviour of *ə is notable in two regards. Firstly, it did not participate in vowel harmony, which is rather unusual for a vowel presumably underspecified for all relevant quality features. Secondly, *ə appeared to be transparent to the POA features of immediately following semivowels: The sequences *əj *əɰ *əʋ acted exactly like *i *a *u in setting the environment for vowel assimilation. A possible explanation might be that *ə had the allophones [ɪ ɐ ʊ] before the corresponding semivowels, which were sufficiently specified to cause harmony, but still lax enough to remain separate from *i *a *u in Western Hitatc.
Almost all surface syllables in Proto-Hitatc were of the form CV, that is, exactly one onset consonant followed by exactly one vowel. The only exception is that word-initial consonants appear to have been optional; some linguists have argued that this would be a strong argument for including the phoneme *ʔ as distinct from *h, so a pure CV structure could be maintained.
However, a more detailed analysis of morphophonemic alternations and certain other irregularities shows that underlying morphemes could actually both begin with a vowel and/or end with a consonant. The underlying syllable structure must therefore be analyzed as (C)V(C).
Basic roots usually consisted of two or three syllables. Monosyllabic morphemes are also reconstructed; these were mostly affixes or grammatical function words such as pronouns or adpositions. A small number of affixes were subsyllabic C or V or antisyllabic VC in their underlying form; sandhi would make these conform to the canonical syllable structure where necessary.
The general morpheme structure can be summarized as follows:
- roots: (C)VCV(CV)(C)
- particles: CV(C)
- affixes: -(C)V(C)- or rarely -C-
In order to preserve a strict CV surface structure, a number of sandhi rules were applied when two vowels or two consonants came in contact with each other at morpheme boundaries, when a vowel ended up in word-initial position, or when a consonant ended up in word-final position. The rules should be applied in the following order, starting from the beginning of the word:
- Before *i *a *u, the glides *j *ɰ *ʋ are inserted.
- *e *o mutate into *ɰi *ɰu.
- *ə is dropped.
- *ij *ej *əj *uj become *i, and *aj *oj become *e.
- *uʋ *oʋ *əʋ *iʋ become *u, and *eʋ *aʋ become *o.
- *aɰ *eɰ *əɰ become *a, *uɰ *oɰ become *o, and *iɰ become *e.
- Word-final *h is dropped.
- After the palatal stops *c *ɟ and after the laterals *l̥ *l, an epenthetic *i is inserted.
- After the uvular consonants *q *χ *ɴ *ʀ, an epenthetic *a is inserted.
- After all other consonants, an epenthetic *ə is inserted.
- *ə adjacent to another vowel is dropped.
- *ie *ee are simplified to *e, and *uo *oo are simplified to *o.
- *i *a *u after another vowel are treated like the semivowels *j *ɰ *ʋ.
- Remaining *e *o after another vowel are treated like the semivowels *j *ʋ.
- *s *h in clusters with a consonant other than *j *ɰ *ʋ drops out, with compensatory devoicing of the remaining consonant where applicable. The clusters *sh *hs both simplify to *s.
- Preconsonantal *l is treated like the semivowel *j.
- Preconsonantal *r is dropped.†
- *k *g become palatal *c *ɟ when preceded by *j.
- If the first consonant of a cluster is a semivowel (including reinterpreted underlying vowels), the same changes as listed in the word-final section apply.
- If the first consonant of a cluster is a voiceless sonorant, it drops out with compensatory devoicing of the remaining consonant where applicable.
- If the first consonant of a cluster is a voiced nasal, it drops out with compensatory voicing of the remaining consonant where applicable.
- If the second consonant of a cluster is a voiceless liquid, it drops out with compensatory devoicing of the remaining consonant where applicable.
- Two plosives of the same POA simplify to a single plosive, which is unvoiced unless both underlying plosives are voiced.
- Two contiguous dorsal plosives (velar and uvular) likewise simplify to a single plosive, taking on the POA of the second onderlying segment.
- Postconsonantal voiced liquids *l *r are treated like the vowels *i *ə.
- Postconsonantal semivowels *j *ɰ *ʋ are treated like the vowels *i *a *u.
- All remaining consonant clusters are resolved by insertion of an epenthetic vowel, which surfaces as *a when adjacent to an uvular consonant, as *i when adjacent to a palatal consonant, as *e when adjacent to both an uvular and a palatal consonant, and as *ə elsewhere.
- All remaining vowel clusters are resolved as described at the beginning of this section.
For the purposes of all sandhi rules, *ʀ counts as the voiced equivalent of *q (so to say, *ɢ), and *χ counts as the unvoiced equivalent of *ɴ (so to say, *ɴ̥).
Proto-Hitatc seems to have had a fairly strong dynamic stress accent, leading to significant vowel reduction especially in the western daughter languages. The location of the accent was phonemic; any syllable of a root could be stressed.
In some daughter languages, most prominently in the western ones, affixes could cause word stress to move. All such affixes, however, were only grammaticalized after the breakup of Proto-Hitatc. With the current state of research, Necine (of the eastern branch) seems to be the only post-Zārakātiāran Hitatc language that preserved the original accent location for all basic morphemes.
- Proto-Hitatc phonemes will be designated using mostly IPA notation, with the exception that the ring diacritic for voiceless sonorants is replaced by a following <h>: *mh *nh *ŋh *lh *rh.
- Proto-Hitatc items will generally be written in italics and prefixed by an asterisk.
- In glossed examples, the Proto-Hitatc form will be written in bold type instead, without the asterisk.
- Stress is transcribed with an acute accent.