An analysis of theme vowels in modern greek within distributed morphology

• Autores: Alexandra Galani
• Localización: Interlingüística, ISSN 1134-8941, Nº. 14, 2003, págs. 399-412
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • Theme vowels (TV) are traditionally seen as markers of the conjugational classes in which verbs belong mainly in Slavic and Romance languages but also in Greek. Moreover previous analyses assume that TV are empty morphemes appearing next to a root to form an uninflected stem to which inflectional affixes are further attached. Nonetheless in recent approaches, TV are treated as independent –empty- morphemes (Spencer 1991). These simple enough treatments, though, leave many questions unanswered about the role, the exact positions in which TV appear as well as the features represented in them in Modern Greek (MG). Based on the above claims and comparing gr`af-ome (write-IMP.NON-ACTIVE.PRESENT.1SG) versus kal-`ume (invite-IMP.NA.PR.1SG), we are forced to suggest that there are no TV in MG. On the contrary and in line with empirical observations, we notice that the vowels which follow the roots, differ (-o-, -u-). Consequently we could assume that these vowels should be treated as TV which would also show that these verbs belong in different conjugational classes and enable us to predict that they verbs behave in a different way morphologically, as it is indeed the case. On the other hand, these vowels cannot be seen as independent morphemes but only as part of the morphological cluster representing aspect, voice, agreement and tense (cumulative exponence). By these means we can also interpret the allomorphy they present: gr`af-ese, kal-`ise (root-IMP.NA.PR.2SG). However, this contradicts the claim (empty morphemes) in the literature. Further evidence for the dependence of TV on semantic features derives from the perfective forms gr`af-tika (root-PERF.NA.PST.1SG)-: in such cases TV are not spelled-out overtly which suggests that TV in MG are sensitive with respect to the imperfective aspect. Finally the last piece of empirical data -showing that the position of TV as well as their features in MG cannot be predicted in terms of the existing treatments- derives from verbs such as pl`en-o (wash-IMP.A.PR.1SG), `eplin-a (root-PERF.A.PST.1SG). In such cases we notice that the TV is realised within the stem. So the aim of this paper is to provide an alternative account of TV within Distributed Morphology- which would capture the features discussed above. We propose that the MG verbal morphology is organised in terms of markedness based on the degree of markedness of the TV. This hierarchy is set by the degree of frequency and regularity of the forms in which TV appear. In gr`afome-like verbs (the most frequent and regular verbs (Koutsoudas 1964)), the TV is not overtly realised and we take such verbs to be the default category. Based on the way TV are realised in kal`ume and pl`eno-like verbs, we propose that such verbs should be treated as separate entries in addition to the conclusion that TV represent the features of aspect but they are treated as part of the inflectional endings due to the cumulative exponence of inflection. Each TV is marked for a feature which also specifies and arranges the roots and the inflectional suffixes in the lexicon. Consequently when Vocabulary Insertion applies at the morphological component, only these vocabulary items which match the root specification are inserted in the terminal node INFL. Any other combinations result to ungrammaticality. By these means the correct matching between the roots and the suffixes is accomplished (gr`afome, *gr`afume). Finally we treat TV as the realisations of a morphological well-formedness requirement on the lexical head V in order to explain the position at which TV are realised in pl`eno verbs. Morphological rules may further apply before the structure moves to PF- in order any necessary modifications to be made.