David Embick

Morphology. From University of Pennsylvania.

Words are not better than phrases

As recognized by Poser (1992) and others, the alternation between one-word "synthetic" forms and two-word "analytic" forms-- e.g., smarter/*more smart, *intelligenter/more intelligent -- promises to reveal much about the relationship between the "word system" (Lexicon) and "phrase system" (syntax). Most of the literature on this topic is devoted to extending Aronoff's (1976) blocking "out of the Lexicon", so that words can compete with phrases. Thus for this line of thought, smarter blocks *more smart in the same way that glory blocks *gloriosity. Numerous accounts of this effect ground the blocking in a general principle that I call LEXICAL PREFERENCE: the principle that words are better than phrases, and therefore block them. This talk examines LEXICAL PREFERENCE along two lines, and demonstrates that no such principle is operative in grammar. The first point is that the interaction between analytic and synthetic forms like the comparative case above is not blocking. Analyses that work in these terms, by implementing "Poser Blocking" (Poser 1992, Hankamer and Mikkelsen 2005, among others) make incorrect predictions about the conditions under which analytic and synthetic forms compete. This point is illustrated with reference to English comparative/superlative formation, secondarily with reference to definite article affixation in Danish and Swedish. The second point is that, even if we abandon "Poser Blocking", it does not appear that LEXICAL PREFERENCE is doing much (if any) of the work in adjudicating between one-word and two-word expressions; other rules are at play. This point is illustrated with French preposition/determiner fusion (du vs. *de le), which raises some interesting questions of its own. The talk concludes with a discussion how these points relate to (and in most cases follow from) a syntactic (i.e. Non-Lexicalist) approach to morphology.

READINGS (available at http://www.ling.upenn.edu/~embick/ab.html)

Most relevant:

• Embick and Marantz (2006) "Architecture and blocking"

Also relevant:

• (2006) "Blocking effects and analytic/synthetic alternations"
• (2006) "Linearization and Local Dislocation..."