Galina Samoukova, University of Washington
This paper provides an account for the ability of native speakers to analyze words that typically are not included in traditional morphological accounts, i.e., data supplied by child language, folk etymology, and nonce words, that is, words created on the spur of the moment. The main problem with these data is that, being for the most part the result of blending, they do not submit to morphological analysis, as they do not contain morphemes in the traditional sense of the word. For example, in "chubaucher" 'Chubarov's voucher' there are no full morphemes, although we can recognize its meaning and all the allusions contained in it.
I am approaching the problem from the point of view of Renata Raffelsiefen's (1992) theory of recognition, extended to Russian data by Irina Kominz (1994). The main goal of this approach is to account for native speakers' ability to recognize the relatedness among words in a given language. Phonological transparency and semantic relatedness are considered two primary factors responsible for that ability. For example, the fact that "stul" 'chair' and "stul'chik" 'small chair' are both phonologically transparent (i.e. contain an identical phonological sub-string) and semantically related causes native speakers to recognize their relatedness.
According to Raffelsiefen's theory, all the words in the language are interrelated, and recognition of this relatedness allows us to analyze them. Every complex word bears a Base Recognition Relation (BRR) to some semantically and, as a rule, structurally less complex primitive word.
I show that this theory provides a solid account for children's ability to come up with their own etymology of words. The urge to reinterpret primitive words as complex and/ or the attempt to establish a 'false' BRR is the source of folk and child etymology. It is interesting that through this, children often 'reestablish' the formerly existing BRR, because some words in the process of historical development lose their BRR and become primitive.
I also show that the existence of multiple BRR in nonce words is responsible both for our ability to create and understand word play as well as for the problems that some words pose for morphological analysis.