James Sneed German
Laboratoire Parole et Langage, Aix-en-Provence james.german@lpl-aix.fr
In English, the reference of free, personal pronouns may depend on the prosodic context in which they occur. In (1), for example, the reference of him appears to depend on whether the pronoun receives a nuclear pitch accent (Akmajian & Jackendoff 1970).
In one type of account, the presence of an accent on a pronoun simply
“switches” its reference relative to that of some unaccented counterpart (Solan 1983, Kameyama 1999). On this view, accents have an interpretation that is specific to their occurrence on pronouns and independent of information structure (henceforth IS). On another view, contrasts like (1) are predictable from the interpretation of accent patterns within a model of IS (Venditti et al. 2001, deHoop 2004). This talk reports on a perception study that compares these two classes of theories in cases where their predictions conflict. The findings show that the relationship between prosody and pronominal reference crucially depends on IS, and cannot be accounted for by the notion of a reference switch.
The materials included 32 two-sentence contexts like that in (2), which were presented both auditorily and visually to 32 adult speakers of North American English.
In one condition, the second sentence was produced with no accent on the pronoun as in (2iia). This condition established that the preferred reference of the unaccented pronoun is for the subject of the matrix clause of the preceding sentence (i.e., Alex in 2i; 66%). According to a “switching” account, therefore, the preferred referent of the accented pronoun (i.e., 2iib) should be the subject of the embedded clause (i.e., Roger). By contrast, Schwarzschild’s (1999) model of information structure predicts, as I will show, that the accented pronoun in (2iib) cannot refer to the embedded subject of (2i).
In a forced-choice task, subjects chose between two paraphrases of the target sentence (i.e., 2iia/2iib). Forty-four fillers served as distractors and as controls on subjects’ attentiveness. Contra the switching account, accented pronouns actually showed an increased preference for the matrix subject of the preceding sentence (78%, p<0.01). In a related set of conditions, the context (2i) was modified so that, according to Schwarzschild’s model of IS, the accented pronoun was instead consistent only with reference to the embedded subject. In this modified context, the accented pronoun showed a strong dispreference for reference to the matrix subject (21%, p<0.001), while the preference in the unaccented case did not differ significantly differ from that in the original context (72%).
Together, these results show that pronominal reference is sensitive to the interpretation of prosody vis-à-vis information structure, even when this contradicts the predictions of a reference switch effect. I show that the analysis of the findings in terms of specific hearer inferences from IS not only extends to classic examples like (1), but is also the more parsimonious account. In addition, I show how this tendency for free pronouns to reflect the structure of local
83
discourse (Kehler 2002) may be used as a discrete detection measure for exploring further aspects of the mapping between prosody and meaning.
Examples
(1) John hit Bill, and then George hit him/HIM. (him=Bill, HIM=John) (2) i. At the hotel, Alex reminded Roger to ask for the executive suite.
iia. Later that night, he made a request.
L+H* L-H% H* L-L%
iib. Later that night, he made a request.
L+H* L-H% H* L-L%
Figures
Figure 1. Proportion of matrix subject responses by prosodic pattern (x-axis) and by context (solid vs. dashed line). Levels of context: A model of information structure predicts that the accented pronoun may refer (i) only to the subject of the preceding matrix clause (dashed line), or (ii) only to the subject of the preceding embedded clause (solid line).
References
Akmajian, A. & Jackendoff, R. (1970). Coreferentiality and stress. Linguistic Inquiry, 1, 124-126.
de Hoop, H. (2004). On the interpretation of stressed pronouns. In R. Blutner & H. Zeevat (eds.), Optimality theory and pragmatics (pp. 25-41). New York: Palgrave Macmillan.
Kameyama, M. (1999). Stressed and unstressed pronouns: Complementary preferences.
In P. Bosch & R. van der Sandt (eds.), Focus: Linguistic, cognitive, and computational perspectives (pp. 206-321). New York: Cambridge University Press.
Kehler, A. (2002). Coherence, reference and the theory of grammar. Stanford, CA: CSLI.
Schwarzschild, R. (1999). Givenness, AvoidF and other constraints on the placement of accent. Natural Language Semantics, 7, 144-177.
Solan, L. (1983). Pronominal reference: Child language and the theory of grammar.
Dordrecht: D. Reidel.
Venditti, J., Stone, M., Nanda, P., & Tepper, P. (2001). Toward an account of accented pronoun interpretation in discourse context: evidence from eye-tracking. Rutgers Center for Cognitive Science: Technical Report RuCCSTR-68.
85
The prosody of focus in Italian and German productions of Lecce