Abstract
Andrew Spencer (University of Essex)
Does English have productive compounding?
How can compounds be distinguished from phrases? None of the standard criteria (including stress) give unequivocal results. Phrases supposedly modify nouns in syntax, while single words modify inside compounds, but under Bare Phrase Structure (Chomsky 1995) this has to be reappraised. The phrase markers for black bird and blackbird would be identical:
Adopting the BPS approach (but not necessarily Minimalism!), I argue that only NN 'compounding' is fully productive in English: AN compounding (blackbird) is invariably lexicalized (and often semantically opaque), while NA, AA (ice/icy cold) compounding is a restricted type of syntactic modification. Under BPS this means that compounds will be categorially indistinguishable from phrases and endocentric NN compounding amounts to a syntactic license for a bare N to modify another N (in contrast to, say, French or Russian syntax). This includes synthetic compounds, where 'modification' is interpreted as (inherited) argument-structure satisfaction. Appositional compounds (woman doctor) have two co-heads (Bresnan, 2000). Language-particular principles determine how they are inflected (women doctors, woman doctors).
Syntactic arguments:
Stress: The distinction between forestress (Compound Stress, bláckbird) and afterstress (Phrasal Stress, blackbírd) is notoriously difficult to apply. However, afterstress tends to be associated with appositional readings rather than modificational ones: apprentice instrúctor (appositional) vs. appréntice instructor (synthetic compound, 'one who instructs apprentices').
This supports an model on which there can be morphosyntactic constructions (cf. Booij, to appear, on Dutch compounds), in this case defined by stress patterns (cf. Zwicky 1986) defined over syntactic phrases. There remains the question of why different languages permit different types of lexicalised compound. I suggest that this is not 'grammatical' in the strict sense but is best understood under the version of connectionist architecture argued for by Krott, Baayen & Schreuder (2001).
© INSTITUT DE LINGÜÍSTICA APLICADA - UNIVERSITAT POMPEU FABRA