Foundations of Language: Brain, Meaning, Grammar, Evolution

(ff) #1
d. week by week; book after book; bumper to bumper (General form: N P N)^86 (Williams 1994; Oehrle 1998)

These are far from“core”constructions of Englishin Chomsky's (1981) sense, but theyare used allthetime and must
be learned by the child. Universal Grammar must therefore make provision for them. Once one bluntly admits the
possibility in the grammar of such“syntactic junk”—constructions that cannot be derived in a well-regimented way
fro many standard structure—it is easier to condone better-behaved constructional idioms such as those in the
previous section. (Syntactic“blends”of the sort discussed by Bolinger (1961) might fall in this category as well.)


This approach leads to a number of theoretical issues for future research. Thefirst is: How structurally complex can
constructions be? This issue has been addressed to some extent in the literature on idioms (e.g. Bresnan 1982b;
Marantz 1984); I do not know if it has been addressed in terms of the broader class of syntactic constructions. Going
to extreme possibilities, it might be intriguing to regard poetic meters and verse forms as phonological constructions
that must unify with a text. And since we spoke earlier of song lyrics being stored as large lexical items, we might
consider treating as constructions those song lyrics that contain a variable to befilled creatively in successive verses,
such as the children's game song (30).


(30) Let everyone VP like me!


Let everyone VP like me!


Come on and join into the game,


You'llfind that it's always the same.


(where VP is a one- or two-syllable VP naming an action, e.g.clap hands, laugh, fall down)


Another issue is how far the constructional approach extends into“core”grammar. Construction Grammar claims
thatallsyntax is builtfro mconstructions clippedtogether. Goldberg's (1995) and Langacker's (1987; 1992) versionsof
Construction Grammar claim that all syntactic structures are form-meaning pairings, and therefore inherently
meaningful; Fillmore and Kay (1993) are more agnostic on the meaningfulness of constructions. My own preference


LEXICAL STORAGE VS. ONLINE CONSTRUCTION 179


(^86) This construction actually falls into“families,”some of which are productive and some of whichare not. A few cases with non-identical nouns must be listed as individual
idioms:cheek by jowl, hand overfist, head over heels, etc. Otherwise the nouns must be identical.N after N is productive and can appear both in NP and adverbialpositions; its
pronominalform isone after another. N by N,N toN, andN for Narealsoproductiveandcanbeusedadverbiallyoras prenominaladjectives;their pronominalforms areone
by/to/for one.

Free download pdf