The difficult mountain: enriched composition in adjective–noun phrases

When readers need to go beyond the straightforward compositional meaning of a sentence (i.e., when enriched composition is required), costly additional processing is the norm. However, this conclusion is based entirely on research that has looked at enriched composition between two phrases or within the verb phrase (e.g., the verb and its complement in . . . started the book . . .) where there is a discrepancy between the semantic expectations of the verb and the semantics of the noun. We carried out an eye-tracking experiment investigating enriched composition within a single noun phrase, as in the difficult mountain. As compared with adjective–noun phrases that allow a straightforward compositional interpretation (the difficult exercise), the coerced phrases were more difficult to process. These results indicate that coercion effects can be found in the absence of a typing violation and within a single noun phrase.     

Processing temporary syntactic ambiguity: The effect of contextual bias

This paper reports two experiments using sentences with a temporary ambiguity between a direct object and a sentence complement analysis that is resolved toward the normally preferred direct object analysis. Postverbal noun phrases in these sentences could be ambiguously attached as either a direct object or the subject of a sentence complement, whereas in unambiguous versions of the sentences the subcategorization of the verb forced the direct object interpretation. Participants read these sentences in relatively long paragraph contexts, where the context supported the direct object analysis (“preferred”), supported the sentence complement analysis (“unpreferred”), or provided conflicting evidence about both analyses (“conflicting”). Self-paced reading times for ambiguous postverbal noun phrases were almost equivalent to the reading times of their unambiguous counterparts, even in unpreferred and conflicted context conditions. However, time to read a following region, which forced the direct object interpretation, was affected by the interaction of verb subcategorization ambiguity and contextual support. The full pattern of results do not fit well with either an unelaborated single-analysis (“garden path”) model or a competitive constraint-satisfaction model, but are consistent with a race model in which multiple factors affect the speed of constructing a single initial analysis.     

Processing temporary syntactic ambiguity: the effect of contextual bias

This paper reports two experiments using sentences with a temporary ambiguity between a direct object and a sentence complement analysis that is resolved toward the normally preferred direct object analysis. Postverbal noun phrases in these sentences could be ambiguously attached as either a direct object or the subject of a sentence complement, whereas in unambiguous versions of the sentences the subcategorization of the verb forced the direct object interpretation. Participants read these sentences in relatively long paragraph contexts, where the context supported the direct object analysis ("preferred"), supported the sentence complement analysis ("unpreferred"), or provided conflicting evidence about both analyses ("conflicting"). Self-paced reading times for ambiguous postverbal noun phrases were almost equivalent to the reading times of their unambiguous counterparts, even in unpreferred and conflicted context conditions. However, time to read a following region, which forced the direct object interpretation, was affected by the interaction of verb subcategorization ambiguity and contextual support. The full pattern of results do not fit well with either an unelaborated single-analysis ("garden path") model or a competitive constraint-satisfaction model, but are consistent with a race model in which multiple factors affect the speed of constructing a single initial analysis.     

The influence of semantic fit on on-line sentence processing

The interactive influence of verb complement preferences and noun phrase semantic fit on resolution of temporary syntactic ambiguity was investigated in an eye movement experiment. The present semantic fit manipulation included noun phrases that fit well as direct objects of the verbs that they followed and noun phrases that were possible but less likely direct objects of the verbs in question. This contrasted with existing research on the use of verb complement preferences and semantic fit during sentence processing, in which processing of noun phrases that are possible direct objects has been compared with processing of noun phrases that are not possible direct objects of the verbs that they follow. Verb complement preference information and noun phrase semantic fit interacted at early stages of on-line sentence processing. Implications of these results for interactive and structural models of sentence processing are discussed.     

Syntactic structure and the garden path

The experiment investigated locally ambiguous English sentences containing “complement” verbs such as believe, which can be followed either by a direct object or by a complement clause. These two sentence types were compared with unambiguous sentences in which the complement clause was introduced by the word that. Subjects processed numerous examples of these sentences in a word-by-word self-paced reading task. At the disambiguation point after the ambiguous noun phrase, longer reading times were obtained for reduced complement constructions compared with direct object sentences. Such an effect has been attributed to the operation of the parsing principle Minimal Attachment (Frazier and Rayner, 1982). This principle predicts that subjects assume falsely that the noun phrase after the complement verb in the reduced complement constructions is the direct object, resulting in the need for subsequent structural reanalysis. However, longer times in the disambiguating zone were also found for the unambiguous that complements. Thus, the complexity difference seems not to represent “garden-pathing” as a result of the operation of Minimal Attachment, but may instead reflect the extra complexity caused by having to handle two sets of clausal relations instead of just one.     

The return of “visiting relatives”: Pragmatic effects in sentence processing

Two experiments were performed to investigate the role of syntactic and pragmatic cues on the disambiguation of noun phrases of the form VERB+ing NOUN+s, like visiting relatives, that can be interpreted as either singular or plural noun phrases. Both experiments used a self-paced reading task in which reading times were measured for two words, a verb and an adverb, immediately following the potentially ambiguous noun phrase. The interpretation of the noun phrase as singular or plural was biased by pragmatic cues in the first experiment and by syntactic cues in the second experiment. In both experiments, subjects were faster to read the adverb following the verb when the interpretation biased by the cues agreed in number with the verb that immediately followed the target noun phrase than when it did not agree with the verb. These results suggest that pragmatic cues, like syntactic cues, can be utilized rapidly in sentence processing.     

American English Usage Frequencies for Noun Phrase and Tensed Sentence Complement-Taking Verbs

The present research documents the American English usage frequencies for 136 verbs that occur with noun phrase and tensed sentence complements (e.g., accepted, noun phrase complement: The student knew the answer yesterday, tensed sentence complement: The student knew the answer was correct). This class of verbs has been the focus of numerous empirical studies investigating how syntactic ambiguity is resolved during language comprehension (Ferreira & Henderson, 1990; Garnsey, Pearlmutter, Myers, & Lotocky, 1997; Holmes, Stowe, & Cupples, 1989; Kennedy, Murray, Jennings, & Reid, 1989; Kennison, 1995; Rayner & Frazier, 1987; Trueswell, Tanenhaus, & Kello, 1993). The frequencies for noun phrase complements, tensed sentence complements, prepositional phrase complements, infinitival complements, and other usages are provided, as well as the frequencies with which specific verbs occur with the overt complementizer that in tensed sentence complements.     

Limitations on the use of verb information during sentence comprehension

An eye tracking experiment was conducted in order to investigate the role of verb information in resolving structural ambiguity during sentence comprehension. Reading time was measured on sentences containing temporarily ambiguous noun phrases (e.g., “The athlete revealed the problem⊙) that were continued as tensed sentence (S) complements or noun phrase (NP) complements. Ambiguous noun phrases were preceded either by verbs occurring most frequently with NP complements (NPbiased) or verbs occurring most frequently with S complements (S-biased). Reading time was also measured on sentences containing unambiguous S complements preceded by either NP-biased or S-biased verbs. The results showed that contrary to predictions made by verb guidance theories (e.g., constraint satisfaction; MacDonald, Pearlmutter, & Seidenberg, 1994a, 1994b; Trueswell & Tanenhaus, 1994), for both NP- and S-biased verb conditions, sentences containing temporarily ambiguous noun phrase complements were read most quickly, and sentences containing temporarily ambiguous S complements were read more slowly than those containing unambiguous S complements.     

Spanish-English L2 speakers' use of subcategorization bias information in the resolution of temporary ambiguity during second language reading

Using a self-paced moving window reading paradigm, we examine the degree to which structural commitments made while 60 Spanish-English L2 speakers read syntactically ambiguous sentences in their second language (L2) are constrained by the verb's lexical entry about its preferred structural environment (i.e., subcategorization bias). The ambiguity under investigation arises because a noun phrase immediately following a verb can be parsed as either the direct object of the verb 'The CIA director confirmed the rumor when he testified before Congress', or as the subject of an embedded complement 'The CIA director confirmed the rumor could mean a security leak'. In an experiment with 59 monolingual English participants, we replicate the findings reported in the previous literature demonstrating that native speakers are guided by subcategorization bias information during sentence interpretation. In a bilingual experiment, we then show that L2 subcategorization biases influence L2 sentence interpretation. The results indicate that L2 speakers keep track of the relative frequencies of verb-subcategorization alternatives and use this information when building structure in the L2.     

Complement coercion is not modulated by competition: evidence from eye movements

An eye-movement study examined the processing of expressions requiring complement coercion (J. Pustejovsky, 1995), in which a noun phrase that does not denote an event (e.g., the book) appears as the complement of an event-selecting verb (e.g., began the book). Previous studies demonstrated that these expressions are more costly to process than are control expressions that can be processed with basic compositional operations (L. Pylkkanen & B. McElree, 2006). Complement coercion is thought to be costly because comprehenders need to construct an event sense of the complement to satisfy the semantic restrictions of the verb (e.g., began writing the book). The reported experiment tests the alternative hypotheses that the cost arises from the need to select 1 interpretation from several or from competition between alternative interpretations. Expressions with weakly constrained interpretations (no dominant interpretation and several alternative interpretations) were not more costly to process than expressions with a strongly constrained interpretation (1 dominant interpretation and few alternative interpretations). These results are consistent with the hypothesis that the cost reflects the on-line construction of an event sense for the complement.     

Remembering complex sentences

Recall of complex sentences at two retention intervals was examined, using sentences which varied simultaneously in three ways, being either active or passive, of low or high Yngve depth, and predictable or unpredictable. Recall of any particular sentence was cued with either the logical subject, logical object, verb or adverbial phrase noun. In general, unpredictable sentences were recollected better than predictable ones, low Yngve depth sentences were recalled better than high Yngve depth ones, and passive sentences were retained better than active ones. The most effective cue was the object, followed in turn by the subject and the adverbial phrase noun, with the verb being least effective. Implications of these findings are discussed.     

Lexical structure in parsing long-distance dependencies

We review a series of experiments investigating lexical influences in parsing sentences with long-distance dependencies. We report three primary results. First, gaps are posited and filled immediately following verbs that are typically used transitively, even when the filler is an implausible object of the verb. However, gaps are not posited after verbs that are typically used intransitively. Second, plausibility determines whether or not a filler is treated as the object of a verb when the verb is typically used with both a direct object and an infinitive complement. Finally, verb control information is used immediately in determining which noun phrase will be interpreted as the understood subject of an infinitive complement.     

Lexical guidance in sentence processing?

An eyetracking experiment was conducted to explore a self-paced reading effect reported by Mitchell (1987). Mitchell found that a noun phrase (NP) was read slowly when it immediately followed an intransitive verb, as long as the verb and NP appeared in the same presentation region. This effect has been used to support the claim that verb subcategorization information is not used initially in sentence parsing. However, the effect did not appear in the eyetracking experiment reported in the present paper, supporting criticisms that Mitchell’s segmentation procedure distorted the parsing process.     

The parser doesn't ignore intransitivity, after all

Several previous studies (B. C. Adams, C. Clifton, & D. C. Mitchell, 1998; D. C. Mitchell, 1987; R. P. G. van Gompel & M. J. Pickering, 2001) have explored the question of whether the parser initially analyzes a noun phrase that follows an intransitive verb as the verb's direct object. Three eye-tracking experiments examined this issue in more detail. Experiment 1 replicated the finding that readers experience difficulty on this noun phrase in normal reading and found that this difficulty occurs even with intransitive verbs for which a direct object is categorically prohibited. Experiment 2, however, demonstrated that this effect is not due to syntactic misanalysis but to disruption that occurs when a comma is absent at a subordinate clause/main clause boundary. Experiment 3 replicated the finding (M. J. Pickering & M. J. Traxler, 2003; M. J. Traxler & M. J. Pickering, 1996) that when a noun phrase "filler" is an implausible direct object for an optionally transitive relative clause verb, processing difficulty results; however, there was no evidence for such difficulty when the relative clause verb was strictly intransitive. Taken together, the 3 experiments undermine the support for the claim that the parser initially ignores a verb's subcategorization restrictions.     

Semantic facilitation and lexical access during sentence processing

An experiment was conducted testing predictions derived from context-dependent and context-independent models of lexical access. Four types of unambiguous test sentences were constructed. The direct object of each test sentence was preceded by a verb that was either semantically related or unrelated to it, and by an adjective that was semantically related or unrelated. Context-dependent models predict that the speed with which the object noun is retrieved from the mental lexicon will be faster when the verb and/or the adjective is semantically related; context-independent models predict no such facilitation. Forty-four subjects each heard 32 test sentences and were asked to monitor within the sentence for a word-initial target phoneme. The target phoneme occurred on the word following the object noun. Reaction times to detect the targets were obtained. According to context-dependent models, these times should be shorter when related words precede the object noun, and that is what was found. It was also observed that the facilitation effects due to the related verbs and adjectives were additive. Implications of these results were discussed.     

The Resolution of the Dative NP Ambiguity in Korean

An eye-movement study examined the ambiguity of the dative noun phrase (NP) in Korean, which may be attached either to the main verb or to a relative verb. Experimental sentences of the form NP-Nom NP-Dat NP-Nom V (dative) NP-Acc V (dative or simple transitive; Nom = nominative; Dat = dative; V = verb; Acc = accusative) were tested. Garden-path theory predicts that comprehension difficulty occurs when the main verb is transitive, forcing the dative NP to be attached to the relative verb, because the dative NP will have already been attached in the clause posited when the first nominative NP is read. In contrast, lexically based theory predicts that comprehension difficulty occurs when the main verb is a dative verb, because the dative NP will be attached to the first verb encountered (the relative verb). The present study did not fully support either theory, because the differences seen in the experimental conditions appeared, at least in part, in the control conditions. However, the results were much different from the prediction of lexically based theory. Reading was faster, and regressive eye movements fewer, in the dative-ambiguous condition than in the transitive-ambiguous condition. Some reasons for the unexpected difficulty of the transitive-unambiguous condition are advanced.     

Argument Status and PP-Attachment

Prepositional phrase attachment was investigated in temporarily ambiguous sentences. Both attachment site (noun phrase or verb phrase) and argument status (argument or adjunct) were manipulated to test the hypothesis that arguments are processed differently than adjuncts. Contrary to this hypothesis, some previous research suggested that arguments and adjuncts are initially processed in the same manner, following a general bias to attach prepositional phrases to the verb phrase whenever possible [Clifton, Speer, & Abney (1991) Journal of Memory and Language, 30, 251–271]. The current study supports the hypothesis for differential processing, even during the initial stages of syntactic analysis. In an eye movement experiment, readers spent less first-pass time on argument prepositional phrases (PPs) than adjunct PPs. The results support a view in which a noun’s or verb’s argument structure can facilitate the analysis of its arguments.This research was supported by NSF grants SBR-9729056 and SBR-9720473 to the first author. We gratefully thank Gerry Altmann, Rick Lewis, Brian McElree, Don Mitchell, Neal Pearlmutter, Shari Speer, and Suzanne Stevenson for useful discussions and helpful comments. We would also like to acknowledge the Rutgers University Center for Cognitive Science, where the eye movement data were collected and an initial draft of the paper was written.     

Structural limits on verb mapping: the role of abstract structure in 2.5‐year‐olds’ interpretations of novel verbs

Two experiments showed that 2.5‐year‐olds, as well as older children, interpret new verbs in accord with their number of arguments. When interpreting new verbs describing the same motion events, children who heard transitive sentences were more likely than were children who heard intransitive sentences to assume that the verb referred to the actions of the causal agent. The sentences were designed so that only the number of noun‐phrase arguments differed across conditions (e.g. She’s pilking her over there versus She’s pilking over there). These experiments isolate number of noun‐phrase arguments (or number of nouns) as an early constraint on sentence interpretation and verb learning, and provide strong evidence that children as young as 2.5 years of age attend to a sentence’s overall structure in interpreting it.     

Investigating the effects of distance and number interference in processing subject-verb dependencies: an ERP study

Event-related potentials (ERPs) were used to investigate whether the processing of subject-verb dependencies is influenced by (1) the linear distance between the subject and the verb and (2) the presence of an intervening noun phrase with interfering number features. Linear distance did not affect integration and diagnosis or revision processes at the verb, as indexed by early negative and P600 components. This is in accordance with hierarchy-based models of reanalysis, but is problematic for distance-based integration models. However, tracking of the subject features is affected by linear factors: more judgment errors were made in the long compared to the short condition. Furthermore, the presence of a plural object between the singular subject and the verb led to more judgment errors, and an enhanced positivity around 250 ms for the grammatical verbs. This suggests that linear factors affect feature tracking, but not integration processes following feature retrieval or repair processes following the detection of a mismatch.     

Predicting complex syntactic structure in real time: Processing of negative sentences in Russian

In Russian negative sentences the verb’s direct object may appear either in the accusative case, which is licensed by the verb (as is common cross-linguistically), or in the genitive case, which is licensed by the negation (Russian-specific “genitive-of-negation” phenomenon). Such sentences were used to investigate whether case marking is employed for anticipating syntactic structure, and whether lexical heads other than the verb can be predicted on the basis of a case-marked noun phrase. Experiment 1, a completion task, confirmed that genitive-of-negation is part of Russian speakers’ active grammatical repertoire. In Experiments 2 and 3, the genitive/accusative case manipulation on the preverbal object led to shorter reading times at the negation and verb in the genitive versus accusative condition. Furthermore, Experiment 3 manipulated linear order of the direct object and the negated verb in order to distinguish whether the abovementioned facilitatory effect was predictive or integrative in nature, and concluded that the parser actively predicts a verb and (otherwise optional) negation on the basis of a preceding genitive-marked object. Similarly to a head-final language, case-marking information on preverbal noun phrases (NPs) is used by the parser to enable incremental structure building in a free-word-order language such as Russian.