Text

theses

ETD doctoral

Pronouns appear frequently in natural language, thus ensuring that speakers are faced with the task of resolving pronominal reference. Traditionally, syntactic structure is thought to place constraints on pronominal interpretation. But what other factors affect pronominal reference, and do these other factors interact with – and possibly override – structural constraints? In this dissertation, I appeal to backwards anaphora with Principle C effects as a case study, and experimentally investigate a range of factors affecting pronominal reference, even when the structural constraint in the form of Principle C predicts that coconstrual is barred. The findings demonstrate that variation in speakers’ acceptability judgments is a result of an interaction between structural and extra-syntactic factors.

Typically, when a pronoun c-commands a name, as in (1), coconstrual between the two is ruled out. This failure of coconstrual is traditionally referred to as the “Principle C effect” (Bruening 2014, Chomsky 1981, Johnson 2012, Rizzi 2004, Safir 1999, Sportiche 1998). Yet, some speakers occasionally allow for structurally marked coconstruals, suggesting that pragmatics and discourse pressures may be weighed on par with syntactic constraints on interpretation, cf. (1) and (2)-(3).

(1) *Hei said that Johni would win. (Chomsky 1981)
(2) Shei was out of spirits when I last talked to Maryi. (Bolinger 1977)
(3) The teacher warned himi that in order to succeed[,] Walteri was going to have to work a lot harder from now on. (McCray 1980)

While a few theoretical proposals have been presented to account for the variability in the data (Chien and Wexler 1990, Grodzinsky and Reinhart 1993, Heim 1982, Higginbotham 1985, Reinhart 1983, Safir 2004), they draw primarily on intuitive judgments and fail to explicitly identify the specific conditions that give rise to varying acceptability. To fill this gap, I employ offline measures (forced choice task and judgment tasks) and systematically manipulate a range of independent variables to demonstrate that unavailability of coconstrual in structurally marked backwards anaphora traditionally referred to as the “Principle C effect” is, in fact, a more nuanced phenomenon.

I propose that a more accurate way of conceiving of the depressed acceptability of coconstrual in structurally marked backwards anaphora is in terms of the “overall obviation effect.” Based on previously unavailable experimental data, I argue that this effect is both composite and gradable. It incorporates the Principle C effect, which is a strong, but not a categorical restriction on pronominal interpretation. The magnitude of the overall obviation effect further varies depending on a wide range of factors that are common to all cases of backwards anaphora: structurally marked and structurally neutral. These other factors include, but are not limited to, plausibility of coconstrual, salience of the pronominal antecedent (subjecthood, topicality, prosodic prominence), (Not)-At-Issue status of the proposition containing the two nominals, and order of operations during incremental processing. These findings allow us to begin to make clear predictions about linguistic environments that will (or will not) give rise to coconstrual and enrich our understanding of the complex mechanisms behind pronominal reference resolution.

ETD_11253

Ph.D.

Includes bibliographical references

doi:10.7282/t3-hp5h-zk10

The author owns the copyright to this work.