CONSTRAINTS ON POETIC LICENCE IN THE QUTADGU BILIG : THE CONVERB AND AORIST VOWELS

In Azeri Turkish, the opposition atır : atar hears the weight of a full-fledged grammatical (aspect) category. Not so in Tür key-Turkish: A foreigner uttering the sentence Beni kim gör er? instead of Beni kim görür? 'Who would see me?' would no doubt be understood correctly, although having clearly violated a strict norm of the language. The aorist vowels may occasionally he necessary for disambiguation: kuş would be subject in Kuş dama konar mı? but object in the impersonal Kuş dama konur mu? Somebody who says yap-'ü-ar instead of yapılır (in analogy to yapar, say) might be understood as having meant yapi + lar 'the buildings', or as having constructed the aorist of a new denominal verb yap% + la-, which the hearer may think he would need to learn. But then, in this case, an ambiguous context is hard to think of. In Old Turkic, both the vowel of the aorist and that of the converb (as a rule equal to the former) bear very little functional load, obviously varying like in Turkey-Turkish. Whatever grammatical content different converb/aorist forms may have had in Proto-Turkic, this was already lost by the time of the earliest existing texts. The semantic emptiness of the converb/aorist vowel must have been encouraged, if not caused, by the metanalysis of verbs which retained an original stem vowel in these forms, as ban-r > bar-ir and kali-r > kal-ir2. But perhaps, things were the other way around: Perhaps the reason that the original vowel was retained here was the fact that the content of the other (originally morphological) converb/aorist vowels had already become empty.