The time course of familiar metonymy.

Abstract

Metonymic words have multiple related meanings, such as college, as in the building (“John walked into the college”) or the educational institution (“John was promoted by the college”). Most researchers have found support for direct access models of metonymy but one recent study, Lowder and Gordon (2013), found delayed reading times for metonymic sentences relative to literal controls, in support of an indirect access account. We conducted a speed-accuracy-tradeoff experiment to test whether their result was caused by lower retrieval probabilities, consistent with direct or indirect access models of metonymy, or slower retrieval dynamics, consistent only with indirect access accounts. We found lower retrieval probabilities for the metonymic sentences but no difference in the dynamics parameters. These results therefore suggest that literal senses do not have priority during processing and that established metonymic senses can be accessed directly.

It is not always a matter of time: Addressing the costs of metaphor and metonymy through a speed-accuracy trade-off study.

“…For metonymy, we reported a lower asymptote compared to literal expressions, as in the work of Bott et al (2016), but, differing from it, we also observed dynamics differences, with metonymical sentences being processed with a later intercept than literal sentences. Bott et al (2016) explained the lower asymptote by arguing that the retrieved metonymic SPEED-ACCURACY TRADE-OFF IN METAPHOR AND METONYMY 13 interpretations were less plausible than the literal ones. Participants were less able to find an interpretation in which context and target word meaning were consistent for sentences in the metonymic condition than the literal condition.…”

Linguistic Representation and Processing of Copredication

“…Indeed, Lowder and Gordon (2013) have proposed that "sentence structure is a key factor to consider in developing psycholinguistic models" of figurative language; this may also be the case for models of copredication. They found that institutional senses of college-type metonymies were more difficult to process than physical senses, but that this cost was lessened if the institutional reading was embedded in an adjunct structure (see also Bott et al 2016 for evidence that these institutional senses can still be directly accessed from the nominal, as opposed to being indirectly accessed via the 'literal'/physical sense; a finding which is in line with assumptions in this thesis about the senses of complex polysemous lexical entries being independently represented within the nominal representation).…”

Speed-Accuracy Tradeoff Modeling and its Interface with Experimental Syntax

“…Finally, additional SAT investigations have tested models and theories involving metonymy, metaphor, enriched composition, and scalar implicatures. Metonymic expressions were less likely to be computed than literal controls, with no differences in time course (Bott et al 2016), supporting direct access to metonymic senses, and arguing against an indirect, literalfirst type of model. Similarly, figurative interpretations were less likely to be recovered or computed than literal ones, with no differences in time course (McElree and Nordlie 1999), also arguing against a serial, literal-first approach.…”

“…The predictions that are crucial for evaluating many fundamental hypotheses about language processing, particularly those that address basic architectural issues, often concern the respective speed of processing for different types of expressions. Time course measures are critical for answering questions such as whether the interpretation of one type of expression increases the complexity of a particular operation, whether it recruits an altogether different type of operation, or whether it requires more operations than another (e.g., Bott, Bailey, and Grodner 2012;Bott, Rees, and Frisson 2016;Griffith 1995, 1998;McElree and Nordlie 1999;McElree, Pylkkänen, Pickering, and Traxler 2006). Time course measures also provide the primary means of investigating general architectural issues, such as whether there are contingencies in the organization of component operations, with some operations having temporal priority over others, versus organized in an interactive fashion (Bornkessel, McElree, Schlesewsky, and Friederici 2004;Martin and McElree 2018;McElree 1993).…”