Meaning and production pressures in speakers’ choices: partitive some Judith Degen T. Florian Jaeger Department of Brain and Cognitive Sciences, University of Rochester ————————————————————————– ————————————————————————–
Abstract
The test case: simple vs. partitive some
In a corpus study on the choice of simple some vs. partitive some of DT, we find that the pressure to precisely encode an intended meaning operates in parallel with production pressures. The more similar the meaning of two forms, the larger the effect of production pressures. ————————————————————————–
————————————————————————– In psycholinguistics: a window onto production pressures
————————————————————————– Dataset 1237 cases of some-NPs (269 partitives, 23%) from Switchboard corpus after excluding cases that can only occur in one of the two forms (pronouns, singular count nouns, idioms) ————————————————————————– Predictors entered in mixed-effects logit model predicting partitive use:
————————————————————————– In theoretical linguistics: a window onto meaning differences
Meaning (discourse accessibility, Reed, 1991)
Choice between syntactic forms assumed to be driven by meaning differences (Dor, 2005; Gropen et al., 1989; Pinker, 1989)
————————————————————————–
Study 2: Gradient Alternation Hypothesis
I(SOME | context) = − log( p(some | context) + p(some of DT | context))
————————————————————————–
Number of cases
Results of MTurk ratings Form simple partitive
40 20 0
Methods. For each case, collected 10 Residual mean similarity similarity ratings of original to sentence with some (of) omitted to obtain a measure of some-NP strength (exploiting presuppositionality) −2
−1
0
1
————————————————————————–
————————————————————————– ResultsOld/mediated NP Meaning factors are strongest, * Topical/subject NP *** but both UID factors and one Modified head ***availability factor affect partiCount head *** tive choice in predicted direction: more partitives with inHead frequency * creasing information of SOME Animate NP Info(SOME | NP head) and decreasing availability of * Info(SOME | previous word) head. * 0.0
Coefficient
1.0 0.8 0.6 0.4 0.2 0.0 2.5 2.0 1.5 1.0 0.5 0.0 2.5 2.0 1.5 1.0 0.5 0.0 1.5 1.0 0.5 0.0 0.1 0.0 −0.1 −0.2 −0.3 0.5 0.0 −0.5 0.3 0.2 0.1 0.0 0.2 0.1 0.0 −0.1
strong(similarity <= breakpoint) simple and partitive more similar
———————————— ————————————– simple some strength determined by similarity breakpoints —————————————— ——————————– Old/mediated NP
●
●
Topical/subject NP Modified head
●
●
Count head
●
Head frequency
●
Animate NP Info(SOME | NP head)
●
●
Info(SOME | prev. word) −0.2
0.0
0.2
0.4
0.6
−0.2
0.0
0.2
0.4
0.5
1.0
1.5
Change in log odds
Results of fitting model to partitive and weak / strong simple some weak (similarity > breakpoint) simple and partitive less similar
Production pressures Availability UID
- previous mention of NP - frequency of head - I(SOME| NP head) - topicality of some-NP - animacy of head - I(SOME| previous word) - modification of head Alex ���� ate some (of the) cashews � �� � previous word NP head - head type (mass/count)
“a difference in syntactic form always spells a difference in meaning” (Bolinger, 1968)
−3
Hypotheses
Study 1: parallel pressures
Robust communication: Uniform Information Density Within the bounds defined by grammar, produce utterances that distribute information uniformly across the Info(u) = -logp(u) linguistic signal (Jaeger, 2010; Levy & Jaeger, 2007)
(Milsark, 1974; Ladusaw, 1994; Horn, 1997)
————————————————————————–
————————————————————————–
Availability-based production: Avoid suspension of speech: utter available material first
2 meanings: weak sm vs. strong some
cashews. [combined meaning contribution]
2. Gradient Alternation Hypothesis: Effects of production pressures are more pronounced, the more similar the meanings that the speaker intends to convey by using one of the two forms are.
Extensive study of the choice between syntactic forms assumed to be meaning-equivalent, e.g., that-omission (Ferreira & Dell, 2000), active/passive (Bock & Irwin, 1980), ditransitive (Bresnan et al., 2007)
60
Alex ate SOME
1. Meaning and production pressures operate in parallel in quasialternations (when meanings of two forms are similar enough).
Introduction - syntactic alternations. . .
80
————————————————————————– ——Alex ate some cashews. [simple some; shorter form] Alex ate some of the cashews. [partitive some; longer form]
Observations 500 600 700 800 900 1000 p value ● <.05 ● <= .1 ● >.1
———————————————————————— –
Conclusion
• Production pressures apply to the choice between forms that are not meaning equivalent. We conclude that, rather than production pressures applying after restriction of permissible forms by semantics, the pressure to robustly communicate a core meaning applies in parallel with the pressure to find the most precise form to encode an intended meaning. • This is compatible with probabilistic approaches to form choice: the more similar the meaning (and associated inferences) of two forms, the more likely the choice between the two is to be affected by production pressures.
0.6
Mean similarity breakpoint
meaning factors contribute to both weak/strong and simple/partitive difference – production pressures primarily to simple/partitive (strong dataset)
Bock, K., & Irwin, D. E. (1980). Syntactic Effects of Information Availability in Sentence Production. Journal of Verbal Learning and Verbal Behavior, 19, 467 – 484. Bolinger, D. (1968). Entailment and the meaning of structures. Glossa, 2(2), 119 – 127. References Bresnan, J., Cueni, A., Nikitina, T., & Baayen, H. (2007). Predicting the Dative Alternation. In G. Boume, I. Kraemer, & J. Zwarts (Eds.), Cognitive foundations of interpretation (pp. 1–33). Amsterdam: Royal Netherlands Academy of Science. Dor, D. (2005, January). Toward a Semantic Account of that-Deletion in English. Linguistics, 43(2), 345–382. Available from http://www.degruyter.com/view/j/ling.2005.43.issue-2/ling.2005.43.2.345/ling.2005.43.2.345.xml Ferreira, V. S., & Dell, G. S. (2000, June). Effect of ambiguity and lexical availability on syntactic and lexical production. Cognitive psychology, 40(4), 296–340. Available from http://www.ncbi.nlm.nih.gov/pubmed/10888342 Gropen, J., Pinker, S., Hollander, M., Goldberg, R., & Wilson, R. (1989). The Learnability and Acquisition of the Dative Alternation in English. Language, 65(2), 103 – 257. Horn, L. (1997). All John’s children are as bald as the King of France: Existential import and the geometry of opposition. In Cls 33 (pp. 155 – 179). Jaeger, T. F. (2010, August). Redundancy and reduction: speakers manage syntactic information density. Cognitive psychology, 61(1), 23–62. Available from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2896231&tool=pmcentrez&rendertype=abstract Ladusaw, W. A. (1994). Thetic and categorical, stage and individual, weak and strong. In M. Harvey & L. Santelmann (Eds.), Proceedings of salt iv (pp. 220 – 229). Cornell U. DMLL, Ithaca, NY. Levy, R., & Jaeger, T. F. (2007). Speakers optimize information density through syntactic reduction. In B. Schl\"okopf, J. Platt, & T. Hoffman (Eds.), Advances in neural information processing systems (Vol. 19, pp. 849 – 856). Cambridge, MA: MIT Press. Available from http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Speakers+optimize+information+density+through+syntactic+reduction#0 Milsark, G. (1974). Existential sentences in English. Unpublished doctoral dissertation, MIT. Pinker, S. (1989). Learnability and Cognition. Cambridge, MA: MIT Press. Reed, A. M. (1991). On interpreting partitives. In D. Napoli & J. Kegl (Eds.), Bridges between psychology and linguistics: A swarthmore festschrift for lila gleitman (pp. 207 – 223). Hillsdale, NJ: Erlbaum.
