Investigating Thematic Roles through Implicit Learning: Evidence from Light Verb Constructions

Abstract

The syntactic structure of a sentence is usually a strong predictor of its meaning: Each argument noun phrase (i.e., Subject and Object) should map onto exactly one thematic role (i.e., Agent and Patient, respectively). Some constructions, however, are exceptions to this pattern. This paper investigates how the syntactic structure of an utterance contributes to its construal, using ditransitive English light verb constructions, such as “Nils gave a hug to his brother,” as an example of such mismatches: Hugging is a two-role event, but the ditransitive syntactic structure suggests a three-role event. Data from an eye-tracking experiment and behavioral categorization data reveal that listeners learn to categorize sentences according to the number of thematic roles they convey, independent of their syntax. Light verb constructions, however, seem to form a category of their own, in which the syntactic structure leads listeners down an initial incorrect assignment of thematic roles, from which they only partly recover. These results suggest an automatic influence of syntactic argument structure on semantic interpretation and event construal, even in highly frequent constructions.

Event Structures Drive Semantic Structural Priming, Not Thematic Roles: Evidence From Idioms and Light Verbs

“…However, unlike idioms, light verbs have also been proposed to participate in argument sharing: In addition to being associated with the two event participants of typical agent‐patient verbs, light verbs may also inherit the dative event structure from the main verb itself (e.g., Jackendoff, ; Wittenberg et al., ). This type of account accords well with existing data showing that participants treat light verbs differently from compositional datives (Wittenberg et al., ), and typically as intermediate between compositional datives and canonical agent‐patient verbs (Wittenberg & Snedeker, ; Wittenberg et al., ). For example, Wittenberg and Snedeker (; see also Wittenberg et al., ) found that participants treated light verbs as having three rather than two event participants at least some of the time (23%), although the overwhelming majority of cases were treated as two‐participant events (75%) .…”

“…This type of account accords well with existing data showing that participants treat light verbs differently from compositional datives (Wittenberg et al., ), and typically as intermediate between compositional datives and canonical agent‐patient verbs (Wittenberg & Snedeker, ; Wittenberg et al., ). For example, Wittenberg and Snedeker (; see also Wittenberg et al., ) found that participants treated light verbs as having three rather than two event participants at least some of the time (23%), although the overwhelming majority of cases were treated as two‐participant events (75%) . However, in our Experiment 1, we found no difference between idioms and light verbs in their effectiveness as primes for compositional dative targets; if anything, idioms were slightly better primes than light verbs.…”

“…Lastly, our results also have implications for the semantic representation of light verbs. Like idioms, light verbs are characterized by non‐canonical mappings between syntax and semantics: In the cases discussed here, three syntactic arguments map to two semantic arguments (Wittenberg & Snedeker, ; Wittenberg et al., , ). However, unlike idioms, light verbs have also been proposed to participate in argument sharing: In addition to being associated with the two event participants of typical agent‐patient verbs, light verbs may also inherit the dative event structure from the main verb itself (e.g., Jackendoff, ; Wittenberg et al., ).…”

The Roles of Implicit Causality and Discourse Context in Pronoun Resolution

“…The eye-movement data were analyzed using a nonparametric permutation test involving the comparison between a linear mixed effects model fitted to the data and reference linear mixed effects models fitted to permuted datasets, which allowed us to test whether our predictors were associated with an increase in looks to the NP 1 character at any time during our auditory region of interest, without the need for specific hypotheses about when looking patterns would diverge (Maris and Oostenveld, 2007;Wittenberg et al, 2017).…”

Event structures drive semantic structural priming, not thematic roles: Evidence from idioms and light verbs

“…Crucially, because they are not fully compositional, idioms and light verbs do not have the same semantic representations that compositional dative structures do Wittenberg, Khan, & Snedeker, 2017). Both variants of compositional datives, for instance, involve three event participants, differentiated from each other either by their event structures or thematic mappings ( Fig.…”