Features, universals, and the internal structure of words
Bobaljik (2012 - Universals in Comparative Morphology) presents and defends a set of generalizations about suppletion in comparative adjectives robust enough to contend as linguistic universals. We find analogues of good - better - best, but not patterns like good - better - goodest. The explanation offered in that work led to the postulation of sometimes hidden morpho-syntactic structure even in simple words, ending with the suggestion that such structure is motivated by universal limits on functional morphemes. The explanation of why patterns like *good-better-goodest do not occur provides us with a template for looking for structure in words in domains not investigated earlier from this perspective. Armed with this 'structure-detector', we will look at other domains, including a large survey of pronominal paradigms (Smith, Moskal, Xu, Kang, and Bobaljik, submitted). In pronouns too we find that some patterns of suppletion for case and number are widely attested, while others are virtually unattested. The conclusions developed on the basis of adjectives lead us to posit hidden structure in pronouns as well, as an explanation for apparent universals. After presenting this argument, I contrast the structural approach with a new perspective (Bobaljik and Sauerland, submitted) on how morphological feature inventories may be defined without a priori notions of what the features are. This approach yields a ban on ABA patterns in certain contexts, and I discuss some ways in which these approaches differ, and possible considerations to distinguish them.