Multiple Reduplikation: Typologie and Theorie
Zusammenfassung der Projektergebnisse
This project conducted a typological investigation of multiple reduplication (=MR) and hence instances of two or more reduplicative morphemes within a word. This empirical research resulted in a representative database of 35 MR patterns that are classified according to different morphological and phonological criteria. This typology was then taken as empirical foundation for several theoretical arguments for a purely phonological account to reduplication that is based on prosodic affixation. For one, the project confirmed the hypothesis put forth in Urbanczyk (2000) and Shaw (2005) that there is language-specific variation about which string serves as the base for reduplication. The two base-concepts I could find in my representative database were the adjacent string and the stem. I argue that these base-concepts fall out from general Linearity and Continguity constraints and no reference to an abstract concept of base is needed (as in Urbanczyk (2000) and Shaw (2005)). The typology also revealed an interesting new pattern of general reduction in MR contexts. The complete avoidance of multiple reduplicants had already received a little theoretical attention in the past but no purely phonological reduplication account assuming prosodic affixation was ever proposed for such an avoidance. Most interestingly, the typology showed that complete avoidance of two reduplicants is only the most drastic case of a general reduction pattern: Some MR patterns show a partial reduction of one reduplicant in contexts where another reduplicant is present. This new empirical pattern led to two theoretical proposals that explain reduction in MR contexts. Reduction in MR contexts (complete or partial) falls out as an epiphenomenon under a prosodic affixation account from the general phonological repair operations of coalescence and prosodic integration. Crucially, the trigger for these repair processes was the general phonological tendency to avoid unfaithful operations – in this case, of fission processes that result in phonological copying. No constraint or mechanism specific to reduplication contexts is hence necessary in such an account. A more general theoretical proposal was made by me that did not only explain the reduction in MR contexts but also captured the independent generalization that both reduplicants or copied bases can show phonological reduction effects that are impossible outside of reduplication contexts. The redefinition of fission as redistribution of underlying activity in a model assuming Gradient Symbolic Representations predicts both empirical generalizations since it implies that every copy operation weakens all the elements which are involved in copying.
