Diachronic Phonology

In this paper we argue that the suffixes *-nt- and *-mh1no- were grammaticalized as active and middle participles after Anatolian left the family. Anatolian inherited *-nt-, which originally functioned as a denominal possessive suffix. In Hittite, the *-nt-suffix is used to form verbal adjectives that express a state and are primarily object oriented (similar to *-to- in the “inner-Indo-European” languages). We propose that the reinterpretation of stative-intransitive verbal adjectives as processual (Hitt. ānt- ‘hot’ → ‘being hot’) and the subsequent usage of the nt-suffix with non-stative intransitive verbs and transitive verbs paved the way for its development to an active, subject-oriented participle in inner-Indo-European languages. This process independently took place in Hittite with certain verbs (e. g., adānt- ‘eating’/‘eaten’).
We show that the usage of *-nt- in Tocharian does not differ qualitatively from that of the inner-Indo-European languages and that *-mh1no-, albeit synchronically grammaticalized as gerund-like form, originally had the same function in Tocharian as its reflexes in the inner-Indo-European languages. With respect to their participial morphology, Tocharian is therefore less archaic as previously thought and patterns with the inner-Indo-European languages, while the Anatolian branch remains the most archaic one in the family.

Extended Summary:
Since word-initial vowel deletion (or elision) is not a frequent phonological
phenomenon, and it requires, in some cases, etymological knowledge, it has been
studied on Turkology in a very limited way. The most exhaustive work carried on
the phenomenon so far has been the paper of Vahit Türk, “Türkçede Ön Seste Ünlü
Düşmesi (Aphaeresis, Aphesis) Örnekleri” [Word-initial vowel deletion (Aphaeresis,
Aphesis) examples in Turkish], which was published in 2003. The paper, however, lacks
earlier literature concerning the phenomenon. Therefore, our paper aims to complete the
gap in terms of literature around word-initial vowel deletion and share new observations
through whole materials including the ones that escaped Türk’s work.
The scope of this paper is restricted to diachronic (lexicalized) word-initial vowel
deletion (henceforth WIVD) instances in Turkic literary languages. In cases where the
references include dialectical materials, we’ve also considered them, otherwise we
have not, in order to find further instances, resorted to dialectological dictionaries of
the languages in question. The sandhi cases (e.g., cuma ertesi ‘Saturday’ > cumartesi),
whose phonological consideration is controversial in literature of Turkology, have
been ruled out. Early cases are shown only when we seek to indicate the diachronic
traces of modern examples. If we’ve determined a word base with WIVD, we have
not taken derivatives from the base into consideration. The word definitions elicited
from dictionaries have been curtailed to the extend to give basic meaning.
WIVD in Turkic Studies
Turkologists who have examined WIVD cases in Turkic languages or a given
Turkic language since XIXth century all come to the conclusion that this phonological
phenomenon not often occurs. Yet, Deny disagrees with them in case of Turkish, saying
“it occurs quite often than we think” (1995, p. 101). There are some explanations on
why this phenomenon takes place. Räsänen claims that stress on final syllable of
word causes reduction on the first syllable vowel, even deletion of it (1949, p. 44). In
Deny’s opinion, aphérèse principally happens on an open syllable and causes the loss
of that syllable (1995, p. 101). Tenishev et al. propose that vowels in initial syllables in
polysyllabic words transform according to following conditions: 1) quality of vowel,
2) stress, 3) character of syllable, 4) adjoining consonants, 5) influence of following
vowels (1984, p. 78). All the observations on WIVD correspond on following three
facts:
a) It occurs on unstressed and high vowels,
b) It occurs on open syllables,
c) It occurs before liquid or fricative consonants.
Nevertheless, we have also those instances violating first fact. Türk underscores
the fact that WIVD cases on low vowels go through raising before vowel deletion
(2003, p. 228). Our paper, involves, though a few in number, those which violate the
second and third facts.
As mentioned above the only mono-graphic work on WIVD in Turkic languages
belongs to Vahit Türk. Since this paper has unique position on the Turkology literature, it is worth mentioning. As he puts, he determined 89 words (some are nominal
phrases) with WIVD, yet it is 97 according to our calculation. A small part of the
instances are taken from early Turkic texts and Anatolian Turkish dialects, whereas
majority are elicited from modern Turkic languages. Some examples are but phonetic
variants in different languages, e.g., fat- ‘to crush, crumble’ (Gagauz, Ottoman) ~ vat-
(Bashkir, Chuvash, Tatar), lep ‘flame’ (Kirgiz) ~ lav (Kazakh, Tatar, Uyghur), sen- ‘to
trust’ (Kazakh) ~ şan- (Chuvash), etc. In addition to simplex and derivatives, phrases
or compounds are also considered in the paper: birisi gün ‘the day after’ (< ol birisi
gün), eşu ‘that’s’ (< ana şu), yıbar- ‘to send’ (< Old Turkic ıdu ber-) or oblique forms
such as yessi ‘owner’ (Anatolian dialect), yiliğe ‘for good’ (Anatolian dialect). Though
Türk devoted his article to word-initial vowel deletion, he also exemplified following
CV and VC elisions: çık- (< Old Turkic taşık-), nāk ‘cheek’ (< yanak), lak ‘yeanling’
(< ulak < ūlak < oglak).
Phonosyntax in the WIVD mechanism
As mentioned above, scholars agree on some points that induce initial vowel drop.
The whole material tells us that phonosyntax might as well be responsible for WIVD.
The phonsyntax VCV... dramatically preponderates on WIVD occurrences: alaçık >
laçık, ılıca > lıca, emekle- > mekle-, ınan- > nan-, ısırga > sırga, umutlu > mutlu etc. This
phonosyntax, however, include some exceptions, such as ıltam > ltam, ısmarla-
> smarla-, işte > şte.
Do Turkic linguistic sub-groups have peculiar patterns for WIVD?
Given the syllable number of each word in whole material, it is obvious that Oghuz
and Kipchak sub-groups dramatically differ from each other at syllable number in
any word with WIVD. In the Oghuz sub-group where Turkish and Gagauz instances
are significant in number, the predominant part consists of trisyllabic or polysyllabic
words, whereas disyllabic ones are in minority. As for Kipchak sub-group, even
though there is no preponderance on disyllabic or trisyllabic words, disyllabic words
are not restricted as many as in Oghuz sub-group.
Authenticity of WIVD instances
Some words, after several phonetic developments, might look like that they are
WIVD cases. What causes this as follows:
1) Word-initial metathesis as VCC... > CVC (e.g., ıslıh > sılıh, uksa- > kusa-)
2) Elision of prothetic vowel
The second condition lessens the genuineness of WIVD cases, since prothesis is
characteristic attribute of Turkic languages when they copy a lexeme starting with
/l/ or /r/. For instance, laçın ‘falcon’ is in every scholarly work regarded as a WIVD
instance (as, alaçın/ılaçın > laçın), yet the oldest version of the word is attested as laçın/
lāçın in 8-9th cc. Turkic texts (see Clauson, 1972, p. 763) or flamur ‘linden’ is thought
by Gülensoy (1987, p. 115) to be a WIVD as ıhlamur > flamur, but the word appears
as flamouri in donor language, namely Greek language. Thus, before determining a
word as WVID, one must make sure about each word’s past.

Situația generală. Vechile texte românești prezintă destul de frecvent un tip de situație reperabil în vorbire, caracterizat de un anumit grad de variabilitate. Dincolo de faptul în sine, prima tentație cognitivă este ca, după stabilirea etimonului, să se afle cauzele acelei oscilații, impactul și semnificațiile ei lingvistice. Un astfel de caz este dat de avatarurile cuvintelor poruncă şi porunci (< vsl. *poronka, poronciti).