ALT 8 The Dynamics of Probabilistic Eighth Biennial Conference of the Association for Grammar: Implications for Linguistic Typology Typology University of California, Berkeley July 23–26, 2009 Joan Bresnan Dynamics of Grammar: Implications for Typology – p.1/48 Dynamics of Grammar: Implications for Typology – p.2/48 Dynamics of Grammar Dynamics of Grammar How is (probabilistic) grammar dynamic? Incrementality in production. “the syntactic structure of an utterance re ects in part the order in which information becomes available during the syntactic stages of processing” (Prat-Sala & Branigan 2000) Bock 1982; De Smedt & Kempen 1987; Kempen & Hoenkamp 1987; Levelt 1989 Dynamics of Grammar: Implications for Typology – p.3/48 Dynamics of Grammar: Implications for Typology – p.4/48
Incremental production Incremental production Example from Ferreira 1996: You formulate your message incrementally: You are formulating a message: Sheila gave something happened – the giving of toys to some children by a person named Sheila Which do you choose? the children or toys ? Dynamics of Grammar: Implications for Typology – p.5/48 Dynamics of Grammar: Implications for Typology – p.6/48 Incremental production Incremental production Sheila gave Sheila gave Sheila gave toys Sheila gave the children Sheila gave toys to Sheila gave the children toys Sheila gave toys to the children Dynamics of Grammar: Implications for Typology – p.7/48 Dynamics of Grammar: Implications for Typology – p.8/48
Incremental production Incremental production Which do you choose? the children or toys ? Activation is increased by • discourse accessibility (Bock & Irwin 1980, Select the item with highest activation at the time the Prat-Sala & Branigan 2000) postverbal position is to be lled. • animacy (Bock, Loebell, & Morey 1992) • effects of prior processing (Bock 1986; Pickering, Branigan, and McLean 2002) • ... Dynamics of Grammar: Implications for Typology – p.9/48 Dynamics of Grammar: Implications for Typology – p.10/48 Incremental production Incremental production Victor Ferreira (1996) “Is it better to give than to Bresnan, Cueni, Nikitina, & Baayen 2007: donate? Syntactic exibility in language production” discourse given � not given JML 35: 724–55 animate � inanimate Experimental participants formulated sentences faster de nite � inde nite with give than with donate . pronoun � non-pronoun less complex � more complex Incremental theories (in contrast to competition structural persistence theories) predict this, because the availability of alternative ways to ll the postverbal slot facilitates V NP rec NP thm formulation with whichever argument is more V NP thm PP rec activated. Dynamics of Grammar: Implications for Typology – p.11/48 Dynamics of Grammar: Implications for Typology – p.12/48
Incremental production Incremental production ...the children ... Similar ndings (though quantitatively different) for Sheila gave them some toys. • dative alternation • active/passive voice alternation ...the toys ... • genitive alternation ( the woman’s shadow, the Sheila gave them to some children. shadow of the woman ) Sheila gave the children toys. Bock & Irwin 1980, Weiner & Labov 1981, Thompson 1990, Hawkins 1994, Collins 1995, Bresnan et al. 2007, Bock et al. 1992, Prat-Sala & Branigan 2000, Estival & Myhill 1988, Rosenbach 2003, 2005, O’Connor, Anttila, Fong & Maling 2004, Hinrichs & Szmrecsányi 2007, Tagliamonte & Jarmasz 2008, Shih et al. 2009, . . . Dynamics of Grammar: Implications for Typology – p.13/48 Dynamics of Grammar: Implications for Typology – p.14/48 Dynamics of Grammar Dynamics of Grammar How is (probabilistic) grammar dynamic? Incrementality of comprehension. Many studies: Boston, Hale, Kliegl, Patil, & Vasishth 2008; Demberg & Keller 2008; Levy 2008, Gibson 1998; Grodner & Gibson 2005; Altmann & Kamide 1999; Kamide, Altmann, & Haywood 2003; Rayner, Warren, Juhasz, & Liversedge 2004; Rayner 1998 Dynamics of Grammar: Implications for Typology – p.15/48 Dynamics of Grammar: Implications for Typology – p.16/48
Incremental comprehension Incremental comprehension Behavioral indices of comprehension dif culty (e.g. Coupling comprehension with production: eye-tracking) sensitive to: Ford & Bresnan (2009) • probabilistic expectation of a syntactic category Word recognition during reading as a function of given the preceding sentence fragment corpus model probabilities for the dative construction • the number and position of syntactic constituents in the sentence • expectation of semantic referents • semantic plausibility of words in their context • ... Dynamics of Grammar: Implications for Typology – p.17/48 Dynamics of Grammar: Implications for Typology – p.18/48 Incremental comprehension Incremental comprehension Corpus model classi cation accuracy, spoken English: Speaker: I’m in college, and I’m only twenty-one but I had a speech class last semester, constructions baseline accuracy study and there was a girl in my class who did a speech on home care of the elderly. dative 79% 94% * And I was so surprised to hear how many people, you know, the older people, are like, fastened to their beds so they can’t get out just because, you know, genitive 59% 89% ** they wander the halls. And they get the wrong medicine, just because, you know, the aides or whoever *Bresnan, Cueni, Niktina, Baayen 2007 **Shih, Grafmiller, Futrell, Bresnan 2009 Dynamics of Grammar: Implications for Typology – p.19/48 Dynamics of Grammar: Implications for Typology – p.20/48
Incremental comprehension Incremental comprehension Dependent variable = the decision latencies on the Speaker: preposition to : I’m in college, and I’m only twenty-one but I had a speech class last semester, and there was a girl in my class who did a speech on home care of the elderly. give the wrong medicine to them And I was so surprised to hear how many people, you know, the older people, are like, fastened to their beds so they can’t get out just because, you know, Predictor = the conditional probabilities of a they wander the halls. And they get the wrong medicine, just because, you prepositional to dative given only the verb, theme, and know, the aides or whoever previous contextual information (from the Bresnan et just ____ ___ _____ ________ __ ____ ____ ______ al 2007 dataset) note: recipient is irrelevant Dynamics of Grammar: Implications for Typology – p.20/48 Dynamics of Grammar: Implications for Typology – p.21/48 ! 4 ! 2 0 2 Incremental comprehension oz us Prediction: faster decision times for the word to in 6.4 more probable structures mean log RT 6.2 6.0 ! 4 ! 2 0 2 corpus conditional log odds Dynamics of Grammar: Implications for Typology – p.22/48 Dynamics of Grammar: Implications for Typology – p.23/48
Incremental comprehension Incremental comprehension • We extracted all instances of verb dative NP from Kuperman & Bresnan conducted a pilot study on the Dundee Corpus (Pynte & Kennedy 2006): Dundee. • We measured the probability of verb dative NP thm the eye-movements of 10 experimental participants and verb dative NP rec in two other corpora of British recorded during their reading of editorials of the journalistic texts (Grimm and Bresnan 2009). British newspaper The Independent . Prediction: faster reading times for the verb if that verb is followed by a noun phrase with a more probable thematic role (parafoveal on foveal effects) Dynamics of Grammar: Implications for Typology – p.24/48 Dynamics of Grammar: Implications for Typology – p.25/48 Dynamics of Grammar How is (probabilistic) grammar dynamic? Dynamics of Grammar: Implications for Typology – p.26/48 Dynamics of Grammar: Implications for Typology – p.27/48
Continuous change Continuous change Continuous historical change: Saxon ( ’s ) genitive: • increasing from 1960’s and 1990’s in U.S. and Prepositional ( to ) dative: U.K. (Hinrichs and Szmrecsányi 2007) • increasing with verb give in spoken New Zealand • increasing in spoken Toronto English in younger English since 1900’s (Bresnan and Hay 2008) speakers (Tagliamonte and Jarmasz 2008) • decreasing from 1960’s and 1990’s in U.S. and U.K. (Grimm and Bresnan 2007) Dynamics of Grammar: Implications for Typology – p.28/48 Dynamics of Grammar: Implications for Typology – p.29/48 Continuous change Continuous change Hawkins (2007) “Processing typology and why psychologists ‘Soft constraints mirror hard constraints’ need to know about it,” New Ideas in Psychology 25: 87–107 Animacy alignment in spoken English dative ...patterns and preferences found in constructions (Bresnan, Cueni, Nikitina, and Baayen performance in languages with several 2007): structures of a given type (e.g. preferences after controlling for possible confounds, inanimate among alternative word orders) are the same recipients are over fi ve times more likely to occur in patterns and preferences one nds across dative PPs than animates languages in the xed conventions of grammars that permit less variation (i.e. in xed word orders) Dynamics of Grammar: Implications for Typology – p.30/48 Dynamics of Grammar: Implications for Typology – p.31/48
Recommend
More recommend