Gaberell Drachman, Univ. Salzburg
Concord in Morphology
The term Concord canonically refers to epiphenomena of agreement in syntax. Thus it may refer to agreement between the verb and its external &/or internal arguments or between determiner, adjective and head-noun within DetP. But we must also consider Concord chains such as are employed in the treatment of negative Concord in French and English (Roberts & Roussou 1999). Further, concord has a potential checking function: there may be obligatory agreement between inflectional tense or aspect markers and various types of adverb, as in Alexiadou (1995). But this paper proposes to extend the notion of concord to morphology.
Drachman and Malikouti-Drachman 2000 (DMD) first made use of the notion concord to cover agreement which is word-internal, subsuming the discontinuous set of morphemes that overlap in realising a common semantic content within the word, as in Matthews' (1974) 'multiple exponency', well exemplified in Latin, and even found in German ge-putz-t. Extreme examples would be Case-doubling in Pilbara (W. Australian) languages, or causative doubling in (Bantu) Jita (Downing 2001).
In addition, DMD innovated the paired notions 'dominant exponent' (cf. the term 'principle exponent' in Carstairs-McCarthy 1992:212, in a somewhat different context) and 'concordant set'. The 'dominant exponent' is the one showing the most robust concrete paradigm; eg. the dominant exponent of 'past' in ancient Greek (AG) was the augment prefix. The 'concordant set' comprises the remaining exponents, here, the person/number endings-set also realising 'past', which function to enhance that meaning. On such assumptions, we must e.g. claim that since the personal pronouns of Modern Greek are liable to proDrop, they are only concordant and enhancing to the more robust system of verbal endings, which is thus dominant for Person/Number. Compare the converse dominant (and obligatory) status of the clitic pronouns in Modern French.
The paper will go on to explore certain areas of Greek morphology potentially testing such an analysis. Thus, (1) as mentioned above, the AG augment makes a highly plausible candidate for a dominant exponent. We will seek to locate dominant exponents and the corresponding concord sets for the other elements of inflection in Greek, namely mood, voice and aspect. (2) we follow conventional wisdom in considering inflectional systems as central to our problem, but will extend the discussion of Concord to derivation and compounding. In Greek compounding, e.g. the dominant exponent is the so-called joining-vowel (whose absence is very largely predictable), the concordant elements including unmarked stress placement, as in xióni+neró > xion-ó-nero 'snow-water' (3) we have referred to Concord-chains in syntax, and will thus try to justify an extension of the chain-metaphor to morphology; one relevant example concerns the historical change, via grammaticalisation and degrammaticalisation, in the relative status of dominant and concordant/enhancing exponents of Tense in Greek (DMD).
Finally we consider the conflict between the distinctiveness of the dominant exponent and the redundancy of the concordant exponents. Do these notions corrrespond to constraint types in the sense of Optimality Theory? We might interpret distinctiveness as constraining individual forms as an Input-Output constraint (Faithfulness), and as constraining paradigms as an Output-Output constraint (Consistency of content or prosody). But redundancy does not correspond either to the alternative IO constraint type, viz. Markedness, or to an OO constraint such as Consistency. Rather, it is a special case of non-consistency, but one too complex to correspond to a simple down-grading of consistency. ? radical solution might be to employ distinct (part) grammars for production and perception -- the former licensing biunique M-distinctiveness, the latter licensing M-enhancement or redundancy.
© INSTITUT DE LINGÜÍSTICA APLICADA - UNIVERSITAT POMPEU FABRA