Luc Baronian

Université du Québec à Chicoutimi

Below the morpheme, beyond the morpheme

Abstract: Affixation is the basic morphological operation, yet the reasons why it possesses the range of incarnations it does is not well understood. While prefixation and suffixation are fairly simple operations, infixation is intertwined with prosodic structure and Semitic-type transfixation requires adequate representational theories. Germanic-type ablaut is not usually considered part of affixation, though, formally, it looks like a simplified transfixation.

It is satisfying when a complex pattern can be explained through the interaction of simple items. In this talk, I take a formal look at the parameters that distinguish rewrite rules used for morphological purposes and deduce a typology of affixation from these parameters. In order to posit morphological rules, one must first compare forms related in meaning. In comparing these forms, we notice similarities and differences, thus the input and output of morphological rules obviously have similarities (S) and differences (D). I observe that S and D can vary according to three parameters:

1) S and D can be continuous or discontinuous.

2) S can be unspecified or partially specified.

3) A given D, on one side of the rule, can be replaced or deleted (=replacement with a null D) on the other side of the rule.

Aside for the fact that, for interpretability reasons, a completely unspecified variable cannot be discontinuous, I will show that the logical combinations of these parameters generate a basic typology that contains prefixation, suffixation, infixation, ablaut, transfixation, circumfixation, as well as phonemically-sensitive affixation. We will see further that laxing the second parameter allows us to broadly define clipping, compounding and reduplication.

Much has been made of the distinction between morpheme- and word-based theories of morphology. Both, however, inescapably involve comparing forms. Therefore, the fundamental parameters described in this talk apply to both equally.

Colloquia Abstracts