Department of Linguistics

While research in the acquisition of syntax has largely focused on the necessity of abstract representations and the poverty of the stimulus with respect to these representations, very little research has asked how learners use the input to identify these representations. At the same time, research showing that infants are highly sensitive to the statistical structure of the input is often silent about the nature of the acquired representations. I present several experiments illustrating the role of statistical learning in a selective theory of syntax acquisition. I show (a) that infants can use statistical information to identify hierarchical phrase structure in an artificial grammar, (b) that the acquired representations allow for generalization to unobserved sentence structures, and (c) that statistical generalizations to be found in the input have consequences for morphosyntax that go beyond what can be inferred simply from the distributions. Hence, to the extent that learners use statistical information in learning syntax, they are doing so by comparing that information against the predictions of precise alternative syntactic representations.

Abstract in PDF

I will parallel sets of data on comparison constructions from several languages. On the basis of the cross-linguistic differences observed, we propose three parameters of language variation. The first parameter concerns the question of whether or not a language's grammar has incorporated scales into adjective meanings. The second parameter differentiates between languages that allow quantification over degrees in the syntax and those that do not. Finally, we propose a syntactic parameter that concerns options for syntactically filling the degree argument position of a gradable predicate.

The English sentence below is unambiguous: it has reading (i) but not (ii): (1) In May only Mary began to get good roles. (i) Mary is the only person (say, actress) whose situation changed in such a way that before May she didn't get good roles, and after May she was getting ones. (ii) The overall situation changed in such a way that whoever might have been getting good roles before May, after May only Mary was getting ones. Reading (ii) is not even expressible in English in a straightforward manner. But in Hungarian, Italian, Spanish, Russian, Finnish, Shupamem, German, Dutch, etc. there is a straightforward way to express (ii). For example: (2) Ma'jusban elkezdett csak Mari kapni jo' szerepeket. May-in prefix-began only Mary get-inf good roles-acc [only reading (ii)] (3) In mei begon alleen Marie goede rollen te krijgen. in May began only Mary good roles to get [ambiguous, but reading (ii) may be preferred] It will be argued that, despite the superficial similarity, Hungarian-type and Dutch-type languages pull different tricks. In connection with the Dutch-type trick, one question is whether the quantificational content of "begin" is represented explicitly in syntax, or only in the semantics. The talk will present some initial considerations in this connection.

(In collaboration with Joe Pater, UMass Amherst, and Michael Becker, Reed College) Phonological dependencies in natural language tend to be assimilatory or dissimilatory, i.e., they relate two tokens of the same phonetic feature in an utterance. Six minimally-different experiments in phonotactic pattern learning by English speakers support the hypothesis that single-feature dependencies, even when they are not phonetically grounded, are detected more readily than two-feature dependences, even when they are phonetically grounded. The single-feature learning bias is shown to emerge in a constraint-inducing and -weighting phonotactic learner when ``Greek-letter'' variables are restricted to instances of the same feature. The result of this restriction is that single-feature dependencies are supported by multiple overlapping constraints, leading to faster learning in Maximum Entropy and Gradual Learning Algorithm learners, whereas two-feature dependencies must be learned piecemeal. This treatment unifies the bias towards syntagmatically-simple patterns with that towards paradigmatically-simple ones, and points towards applications of constraint multiplicity and constraint generality in explaining learning and typology.

This talk starts out from the old question of whether a sluicing example such as (1) is derived from a full wh-question as in (2) (Ross 1969, Merchant 2001) or from a copular clause as in (3) (Pollman 1975, Erteshik-Shir 1977). (1) John met someone, but I don't know who. (2) John met someone, but I don't know who . (3) John met someone, but I don't know who . I show that of the ten arguments that Merchant (2001:115-127) gives against the analysis in (3), only one strongly suggests that the structure in (3) never occurs (all the other arguments being compatible with a scenario in which both (2) and (3) are optionally available): in languages with morphological case marking, the sluiced wh-phrase never shows up in the case predicted by the copular source. At the same time, however, there is independent evidence suggesting that in certain cases copular clauses can be used as underlying structure for a sluice. In this talk I explore the consequences of these at first sight contradictory data for the licensing and recoverability requirements of sluicing.

Please see the Recursion Conference website: http://www.umass.edu/linguist/recursionconf/

The website for the event can be found here: http://people.umass.edu/harris/snews/snews.html

HUMDRUM, a graduate workshop on Optimality Theory, will feature talks from phonologists at UMass Amherst and Rutgers. We will offer refreshments beginning at 1:30, with talks running from 2:00 until 5:00. The workshop will be followed by a party hosted by John McCarthy and Ellen Woolford. All are welcome to attend. The schedule is here in PDF.

Ilaria Frana defends her Ph.D. dissertation.

Join Seth Cable and Peggy Speas of the Department of Linguistics in viewing the film, THE LINGUISTS. Discussion and snacks will follow. Scientists estimate that of 7,000 languages in the world, half will be gone by the end of this century. On average, one language disappears every two weeks. THE LINGUISTS joins David Harrison and Gregory Anderson, scientists racing to document languages on the verge of extinction. David and Greg's 'round-the-world journey takes them deep into the heart of the cultures, knowledge, and communities at stake. See the trailer at www.thelinguists.com. THE LINGUISTS world premiered at the 2008 Sundance Film Festival. The only film funded by the National Science Foundation ever at Sundance, THE LINGUISTS has since screened at more than thirty festivals worldwide. Directed by Seth Kramer, Daniel A. Miller & Jeremy Newberger; 2008, 65 minutes.

The Linguistics Department will honor graduating seniors with a department-wide ceremony and informal dinner.