Whamit!

The Weekly Newsletter of MIT Linguistics

LFRG 4/28 - Roman Feiman

Speaker: Roman Feiman (Harvard)
Time: Monday, April 28, 12-1:30
Place: 66-148
Title: How abstract is LF? Differences between quantifiers, similarities between operations

Recent work in psycholinguistics (Raffray and Pickering, 2010) has shown that Logical Form representations can be primed — that how people resolve one scope ambiguity will affect their resolution of another ambiguity with different noun content. This suggests that once constructed, mental representations of the relationships between quantifiers are abstracted from the specific sentence and can be reused. We extend Raffray and Pickering’s paradigm to investigate priming across ambiguous sentences with varying subject quantifiers, using “Every”, “Each”, “All of the” and bare numerals. Priming aside, we find very large differences in the overall biases of these quantifiers to take wide or narrow scope relative to an indefinite object quantifier — large enough to swamp many others factors that have been argued to drive scope ambiguity resolution (e.g. linear order, c-command, thematic hierarchy). We also find that LF representations can be primed for all quantifiers, and that the priming is of the same magnitude for all of them, but only as long as the quantifier words in prime and target trials are the same. This finding suggests that the priming paradigm targets a common operation (like QR), which can act on all quantifiers equally. At the same time, we find no priming across sentences with different quantifiers (except from one bare numeral to another), suggesting that all of the quantifier words we tested have separate representations at LF, and that the common operation responsible for within-quantifier priming is unparsimoniously stored, redundant within the lexical entry of each quantifier. Taken together, these findings call for a different kind of theory of LF — one where there are generalized quantifiers and common operations applying to them (with these operations stored lexically), but also one where differences between individual quantifiers have a strong effect on their scoping behavior.