Adele E. Goldberg

The nature of generalization in language

Abstract This paper provides a concise overview of Constructions at Work (Goldberg 2006). The book aims to investigate the relevant levels of generalization in adult language, how and why generalizations are learned by children, and how to account for cross-linguistic generalizations.

Can thematic roles leave traces of their places

An important question in the study of language production is the nature of the semantic information that speakers use to create syntactic structures. A common answer to this question assumes that thematic roles help to mediate the mapping from messages to syntax. However, research using structural priming has suggested that the construction of syntactic frames may be insensitive to variations in thematic roles within messages (Cognition 35 (1990) 1; Psychological Review 99 (1992) 150). Because these studies involved structural alternations whose syntax covaries with the order of thematic roles, it is difficult to assess any independent contribution that role information may make to the positioning of phrases. In this study, we primed the order of the roles without changing the syntactic structure of the sentences produced, and found that the order of the roles was influenced by the priming manipulation. This implies that thematic roles or the features that differentiate them are active within the mapping between messages and sentence structures.

The island status of clausal complements: Evidence in favor of an information structure explanation

Abstract The present paper provides evidence that suggests that speakers determine which constructions can be combined, at least in part, on the basis of the compatibility of the information structure properties of the constructions involved. The relative “island” status of the following sentence complement constructions are investigated: “bridge” verb complements, manner-of-speaking verb complements and factive verb complements. Questionnaire data is reported that demonstrates a strong correlation between acceptability judgments and a negation test used to operationalize the notion of “backgroundedness”. Semantic similarity of the main verbs involved to think or say (the two verbs that are found most frequently in long-distance extraction from complement clauses) did not account for any variance; this finding undermines an account which might predict acceptability by analogy to a fixed formula involving think or say. While the standard subjacency account also does not predict the results, the findings strongly support the idea that constructions act as islands to wh-extraction to the degree that they are backgrounded in discourse.

The role of prediction in construction-learning

It is well-established that (non-linguistic) categorization is driven by a functional demand of prediction. We suggest that prediction likewise may well play a role in motivating the learning of semantic generalizations about argument structure constructions. We report corpora statistics that indicate that the argument frame or construction has roughly equivalent cue validity as a predictor of overall sentence meaning as the morphological form of the verb, and has greater category validity. That is, the construction is at least as reliable and more available than the verb. Moreover, given the fact that many verbs have quite low cue validity in isolation, attention to the contribution of the construction is essential.

Corpus evidence of the viability of statistical preemption

Abstract The present paper argues that there is ample corpus evidence of statistical preemption for learners to make use of. In the case of argument structure constructions, a verbi is preempted from appearing in a construction A, CxA, if and only if the following probability is high: P(CxB|context that would be suitable for CxA and verbi). For example, the probability of hearing a preemptive construction, given a context that would otherwise be well-suited for the ditransitive is high for verbs like explain that overwhelmingly appear in the dative, and low for verbs like tell that readily appear in the ditransitive. Strength of statistical preemption is determined both by this probability, and by the frequency (ln (F)) of a verb in a preemptive construction when the context is at least as well suited to the preempted construction. The critiques of preemption by Stefanowitsch (Cognitive Linguistics 19: 513–531, 2008, this volume) are countered by arguing that the relevant probabilities were not considered. Moreover, we find evidence that constructions are somewhat less constrained when yoked to non-alternating verbs, as Stefanowitsch (cf. this volume) suggests should be the case.

Incidental verbatim memory for language

Abstract It is widely believed that explicit verbatim memory for language is virtually nonexistent except in certain circumstances, for example if participants are warned they are to receive a memory test, if the language is ‘interactive’ (emotion-laden), or if the texts are exceedingly short and memory is tested immediately. The present experiments revisit the question of verbatim memory for language and demonstrate that participants do reliably recognize and recall full sentences that they are exposed to only once at above chance rates (Experiments 1 and 3). The texts are 300 words long, non-interactive, and no advanced warning of a memory test is given. Verbatim memory is demonstrated even when lexical content and memory for gist are controlled for (Experiments 2 and 4). The most striking finding is one of incidental recall: even after a six-day delay, participants reliably reproduce sentences they have heard before when asked to describe scenes, even though they are not asked to recall what they had heard (Experiment 5).

The partial productivity of constructions as induction

Abstract Whether words can be coerced by constructions into new uses is determined in part by semantic sensicality and statistical preemption. But other factors are also at play. Experimental results reported here suggest that speakers are more confident that a target coinage is acceptable to the degree that attested instances cover the semantic space that includes the target coinage. The relevance of coverage is supported by combined effects of type frequency and variability of attested instances [Experiments 1a–1b], and an expected interaction between similarity and variability [Experiment 3]. Similarity to an attested instance is also found to play a role: speakers are more confident of a target coinage when the coinage is more similar to an attested instance [Experiment 3]. Experiment 2 provides a manipulation check that indicates that participants are in fact basing their confidence ratings on the perceived productivity of constructions. The results reported here lend support to the idea that the productivity of constructions depends on general properties of induction.

Metaphorical sentences are more emotionally engaging than their literal counterparts

Why do people so often use metaphorical expressions when literal paraphrases are readily available? This study focuses on a comparison of metaphorical statements involving the source domain of taste (e.g., “She looked at him sweetly”) and their literal paraphrases (e.g., “She looked at him kindly”). Metaphorical and literal sentences differed only in one word and were normed for length, familiarity, imageability, emotional valence, and arousal. Our findings indicate that conventional metaphorical expressions are more emotionally evocative than literal expressions, as the amygdala and the anterior portion of the hippocampus were more active in the metaphorical sentences. They also support the idea that even conventional metaphors can be grounded in sensorimotor and perceptual representations in that primary and secondary gustatory areas (lateral OFC, frontal operculum, anterior insula) were more active as well. A comparison of the individual words that distinguished the metaphorical and literal sentences revealed greater activation in the lateral OFC and the frontal operculum for the taste-related words, supporting the claim that these areas are relevant to taste.

Distinguishing grammatical constructions with fMRI pattern analysis.

All linguistic and psycholinguistic theories aim to provide psychologically valid analyses of particular grammatical patterns and the relationships that hold among them. Until recently, no tools were available to distinguish neural correlates of particular grammatical constructions that shared the same content words, propositional meaning, and degree of surface complexity, such as the dative (e.g., Sally gave the book to Joe) and the ditransitive (e.g., Sally gave Joe a book). We report the first fMRI data that distinguish such closely related, abstract grammatical patterns. Multi-voxel pattern analysis (MVPA) proved capable of discriminating at above-chance levels between activity patterns arising during reading of dative and ditransitive sentences. Region-of-interest analyses reveal that the union of certain language-relevant areas, anterior and posterior BA22, BA44/45 and BA47, yield classification accuracy above chance and above that of control conditions in the left hemisphere but not in the right. Looking more closely at the LH ROIs, we find that the combination of areas aBA22 and BA47 is sufficient to distinguish the two constructions better than the controls and better than chance. The fact that both of these areas-particularly BA47-have been implicated in semantics, lends support to claims that the two constructions are distinguishable semantically. More generally, the ability to distinguish closely related grammatical constructions using MVPA offers the promise of addressing traditional theoretical questions on a neuroscientifically grounded basis.

Subject-auxiliary inversion: A natural category

Abstract This article argues that subject auxiliary inversion in English (SAI) provides an example of a syntactic generalization that is strongly motivated by a family of closely related functions. Recognition of the functional properties of each subconstruction associated with SAI allows us to predict many seemingly arbitrary properties of SAI: e.g., its (partial) restriction to appear in main clauses, the fact that the inversion only involves the first auxiliary, and the fact that its use in comparatives is more limited. The dominant feature of SAI, being nonpositive, is also argued to motivate the syntactic form of SAI. It is suggested that attention to the rich data inherent in language and to findings in categorization research simultaneously serves to reinforce and benefit our understanding of both language and categorization more generally.