Discourse structure and relative clause processing

Studies in several languages have shown that subject-relative clauses are easier to process than object-relative clauses. Mak, Vonk, and Schriefers (2006) have proposed the topichood hypothesis to account for the preference for subject-relative clauses. This hypothesis claims that the entity in the relative clause that is most topicworthy will be chosen as the subject. By default, the antecedent of the relative clause will be chosen as the subject of the relative clause, because it is the topic of the relative clause. However, when the noun phrase (NP) in the relative clause is also topicworthy, the preference for the antecedent to be the subject will disappear. This was confirmed in two experiments. In Experiment 1, we tested relative clauses with a personal pronoun in the relative clause. We obtained a preference for object-relative clauses, in line with the assumption that personal pronouns refer to a discourse topic and are thus topicworthy. In Experiment 2, the discourse status of the NP in the relative clause was manipulated; either it was not present in the preceding context, or it was the discourse topic. The experiment showed that when the NP in the relative clause refers to the discourse topic, the difficulty of object-relative clauses is reduced, in comparison with relative clauses with an NP that is new in the discourse, even in the absence of any explicit cue in the relative clause itself. The experiments show that discourse factors guide processing at the sentence level.     

The correspondence between sentence production and corpus frequencies in modifier attachment

We examined the production of relative clauses in sentences with a complex noun phrase containing two possible attachment sites for the relative clause (e.g., "Someone shot the servant of the actress who was on the balcony."). On the basis of two corpus analyses and two sentence continuation tasks, we conclude that much research about this specific syntactic ambiguity has used complex noun phrases that are quite uncommon. These noun phrases involve the relationship between two humans and, at least in Dutch, induce a different attachment preference from noun phrases referring to non-human entities. We provide evidence that the use of this type of complex noun phrase may have distorted the conclusions about the processes underlying relative clause attachment. In addition, it is shown that, notwithstanding some notable differences between sentence production in the continuation task and in coherent text writing, there seems to be a remarkable correspondence between the attachment patterns obtained with both modes of production.     

Discourse influences during parsing are delayed.

Subjects read sentences containing either a syntactically ambiguous prepositional phrase attachment or a syntactically ambiguous reduced relative clause. The sentences were embedded in passages of text that were consistent with either the minimal or non-minimal attachment reading. In addition, a discourse factor (i.e., whether or not the target sentence was in the discourse focus) was varied. Subjects' eye movements were recorded as they read the passages of text. Our primary finding was that subjects were garden-pathed even when there was biasing context. However, when the target sentence was in the discourse focus, subjects were able to recover more readily from their initial erroneous parse of the sentence. The data thus support models of sentence parsing that postulate that the parsing of a sentence is based upon structurally based principles and the influence of semantic or pragmatic information makes itself felt only after the initial parsing decision has been made.     

The role of pragmatic principles in resolving attachment ambiguities: Evidence from eye movements

In an eyetracking experiment, participants read sentences that contained a prepositional phrase (PP) that could be attached to one of two preceding verbs. To understand the sentence, readers had to select which verb would serve as the host for the PP. In some of the sentences, the critical verbs and the PPs were part of a subordinate clause in which lexical and syntactic factors were expected to matter, but discourse factors favoring attachment of the material to positions where it would be relevant to the main assertion of the sentence were not expected to matter. In other sentences, the critical material was tested in main clause contexts in which the main assertion principle was predicted to apply. Hierarchical linear modeling showed that online attachment preferences were affected by clause type (main vs. subordinate). Specifically, the preference for a local verb over a distant verb was greater when the critical material appeared within a subordinate clause than when it appeared within a main clause. This pattern of results can be explained by the operation of the relativized relevance or main assertion principle, which in our sentences meant that participants favored attachment of the PP to the first verb over the second.     

Emotional nouns affect attachment decisions in sentence completion tasks

We report three sentence completion experiments in which we manipulate the emotional dimension of the nouns in a complex noun phrase (NP) that precedes a relative clause (RC), as in the classic ambiguity in Someone shot the servant of the actress who was on the balcony. The aim was to see whether nouns such as orgy or genocide affect the well-established preference of Spanish to adjoin the relative clause high in the tree (to servant instead of actress in the example above). We manipulated the valence and arousal of the lexical entities residing in the NP. Our results indicate that (a) the inclusion of either pleasant or unpleasant words induces changes in the usual NP1 preference found in Spanish; (b) the effects of high-arousal words are especially clear, in that they pull RC adjunction towards the NP where they are located, be it the NP1 or the NP2; and (c) in the context of sentence production, these kinds of words seem intense enough to promote changes in (and even reverse) a solid syntactic bias. We discuss these findings in the light of existing theories of syntactic ambiguity resolution.     

Learning individual talkers' structural preferences

Listeners are often capable of adjusting to the variability contained in individual talkers' (speakers') speech. The vast majority of findings on talker adaptation are concerned with learning the contingency between phonological characteristics and talker identity. In contrast, the present study investigates representations at a more abstract level - the contingency between syntactic attachment style and talker identity. In a 'visual-world' experiment, participants were exposed to semi-realistic scenes depicting several objects (e.g., an adult man, a young girl, a motorbike, a carousel, and other objects) accompanied by a spoken sentence with a structurally ambiguous relative clause (e.g., 'The uncle of the girl who will ride the motorbike/carousel is from France.' In the context of the scene, 'motorbike' suggested the uncle as the agent of the riding, whereas 'carousel' suggested the girl as the agent). For half the experimental items, one version of the sentence was read by one talker, who always uttered sentences that resolved, pragmatically, to the high attachment (the uncle as the agent), and the other by another talker, who always uttered sentences resolving to the low attachment (the girl as the agent). For the other half of the experimental items, both versions were read by a third talker who produced both high and low attachments. It was found that, after exposure to these stimuli, and for new sentences not heard previously, participants learnt to anticipate the 'appropriate' attachment depending on talker identity (with no attachment preference for the talker who produced both attachment types). The data suggest that listeners can learn the relationship between talker identity and abstract, structural, properties of their speech, and that syntactic attachment decisions in comprehension can reflect sensitivity to talker-specific syntactic style.     

Working memory contributions to relative clause attachment processing: A hierarchical linear modeling analysis

An eye-movement-monitoring experiment tested readers' responses to sentences containing relative clauses that could be attached to one or both of two preceding nouns. Previous experiments with such sentences have indicated that globally ambiguous relative clauses are processed more quickly than are determinately attached relative clauses. Central to the present research, a recent study (Swets, Desmet, Hambrick, & Ferreira, 2007) showed that offline preferences for such sentences differ as a function of working memory capacity. Specifically, both English and Dutch participants' preference for the second of two nouns as the host for the relative clause increased as their working memory capacity increased. In the present study, readers' working memory capacity was measured, and eye movements were monitored. Hierarchical linear modeling was used to determine whether working memory capacity moderated readers' online processing performance. The modeling indicated that determinately attached sentences were harder to process than globally ambiguous sentences, that working memory did not affect processing of the relative clause itself, but that working memory did moderate how easy it was to integrate the relative clause with the preceding sentence context. Specifically, in contrast with the offline results from Swets and colleagues' study, readers with higher working memory capacity were more likely to prefer the first noun over the second noun as the host for the relative clause.     

The on-line resolution of the sentence complement/relative clause ambiguity: evidence from Spanish

Two self-paced reading experiments investigated syntactic ambiguity resolution in Spanish. The experiments examined the way in which Spanish subjects initially interpret sentences that are temporarily ambiguous between a sentence complement and a relative clause interpretation. Experiment 1 examined whether the sentence complement preference found in English is observed in Spanish speaking subjects. In Experiment 2, verbal mood was manipulated in order to study the influence of verb-specific information on sentence processing. Since subcategorization for a subjunctive complement clause is generally assumed to be a lexical property of some verbs, the manipulation of the mood of the embedded verb affords us an interesting and novel way to examine the influence of lexical information on syntactic ambiguity resolution. Experiment 1 showed that Spanish speakers initially interpret the ambiguous that-clause as a sentence complement. Experiment 2 showed that verb-specific information, in particular, the information that specificies that a verb subcategorizes for a subjunctive complement, is accessed and used rapidly and affects the ambiguity resolution process. The results are discussed in relation to current models of sentence processing.     

A Hierarchical Linear Modeling Analysis of Working Memory and Implicit Prosody in the Resolution of Adjunct Attachment Ambiguity

An eye-movement monitoring experiment investigated readers’ response to temporarily ambiguous sentences. The sentences were ambiguous because a relative clause could attach to one of two preceding nouns. Semantic information disambiguated the sentences. Working memory considerations predict an overall preference for the second of the two nouns, as does the late closure principle (Frazier, On comprehending sentences: Syntactic parsing strategies. Ph.D. dissertation, University of Connecticut. West Bend, IN: Indiana University Linguistics Club, 1979). Previous studies assessing preferences for such items have obtained mixed results. On-line assessments show that working memory affects the degree of preference for the first noun, with lower capacity readers having a greater preference for the second noun (Felser et al., Language Acquisition: A Journal of Developmental Linguistics, 11, 127–163, 2003; Traxler, Memory & Cognition, 35, 1107–1121, 2007). Off-line assessments indicate the opposite pattern of preferences when the test sentences are displayed on a single line (Swets et al., Journal of Experimental Psychology: General, 136, 64–81, 2007). However, when implicit prosody is manipulated by displaying the sentences with a break between the second noun and the relative clause, the off-line assessments indicate that readers prefer to attach the relative clause to the first noun. In this experiment, readers’ undertook a working memory assessment and then read test sentences that were displayed across two lines, with a break appearing after the second noun and before the relative clause. The eye-tracking data indicated an overall preference to attach the relative clause to the first noun, and there was little indication that working memory moderated the degree of preference for this configuration. Hence, it appears that readers’ implicit prosodic contours rapidly affect resolution of adjunct attachment ambiguities.     

Case Matching and Relative Clause Attachment

Two accounts of relative clause attachment will be discussed, the case-matching hypothesis proposed by Sauerland and Gibson (1998) and the attachment-binding dualism (Hemforth et al., in press a, b). While the case-matching hypothesis predicts that relative clauses are preferentially attached to NPs whose case matches that of the relative pronoun, attachment binding predicts that NPs are preferentially attached to the most salient host, that is NP1 in constructions with two NPs. We conducted two off-line studies, one sentence completion task and one magnitude estimation experiment using subject (nominative pronoun) and object (accusative pronoun) relative clauses that can be attached to either of the two nouns in a complex subject (NP1 = nominative, NP2 = genitive) or object NP (NP1 = accusative, NP2 = genitive). While attachment binding predicts an across-the-board NP1 preference, the case-matching hypothesis predicts an NP1 prefence only in the case of subject (object) NPs followed by subject (object) relative clauses. The results of both experiments provide evidence for attachment binding and against case matching.     

Cross-structural priming: prepositional phrase attachment primes relative clause attachment

In this paper we show that attachment height (high vs. low attachment) of a modifier to a complex noun phrase (CNP; e.g., "the servant of the actress"), can be primed between dissimilar syntactic structures. In a sentence completion experiment, we found that the attachment height of a prepositional phrase (PP) in the prime sentence primed the attachment height of a relative clause (RC) in the target sentence. This cross-structural priming effect cannot be explained in terms of the priming of specific phrase-structure rules or even sequences of specific phrase-structure rules (Scheepers, 2003), because the attachment of a PP to a CNP is generated by a different phrase-structure rule than the attachment of an RC. However, the present data suggest that the location at which the RC is attached to the CNP is mentally represented, independent of the specific phrase-structure rule that is attached, or by extension, that the abstract hierarchical configuration of the full CNP and the attached RC is represented (Desmet & Declercq, 2006). This is the first demonstration of a cross-structural priming effect that cannot be captured by phrase-structure rules.     

Recency and lexical preferences in Spanish

One experiment provided evidence in support of Gibson, Pearlmutter, Canseco-Gonzalez, and Hickok's (1996) claim that a recency preference applies to Spanish relative clause attachments, contrary to the claim made by Cuetos and Mitchell (1988). Spanish speakers read stimuli involving either two or three potential attachment sites in which the same lexical content of the two-site conditions appeared in a different structural configuration in the three-site conditions. High attachment was easier than low attachment when only two sites were present, but low attachment was preferred over high attachment, which was in turn preferred over middle attachment, when three sites were present. The experiment replicated earlier results and showed that (1) attachment preferences are determined in part by a preference to attach recently/low, and (2) lexical biases are insufficient to explain attachment preferences.     

Overt reanalysis strategies and eye movements during the reading of mild garden path sentences

In an eye movement experiment, we examined the use of reanalysis strategies during the reading of locally ambiguous but globally unambiguous Spanish sentences. Among other measures, we examined regressive eye movements made while readers were recovering in reading mild garden path sentences. The sentences had an adverbial clause that, depending on the mood (indicative vs. subjunctive) of the subordinate clause verb, could attach high (to the main verb of the sentence) or low (to the verb in the subordinate clause). Although Spanish speakers favor low attachment, the high attachment version was quite easy to understand. Readers predominately used two alternative strategies to recover from the mild garden path in our sentences. In the more common reanalysis strategy, their eyes regressed from the last region (disambiguation+ 1) directly to the main verb in the sentence. Following this, they reread the rest of the sentence, fixating the next region and the adverb (the beginning of the ambiguous part of the sentence). Less frequently, readers regressed from the last region (disambiguation+1) directly to the adverb. We argue that both types of strategies are consistent with a selective reanalysis process as described by Frazier and Rayner (1982).     

The influence of only on syntactic processing of "long" relative clause sentences

We report an eye movement experiment investigating the influence of the focus operator only on syntactic processing of "long" relative clause sentences. Paterson, Liversedge, and Underwood (1999) found that readers were garden pathed by "short" reduced relative clause sentences containing the focus operator only . They argued that due to thematic differences between "short" and "long" relative clause sentences, garden path effect might not occur when "long" reduced relative clause sentences are read. Eye-tracking data show that garden path effects found during initial processing of the disambiguating verb of "long" reduced sentences without only were absent or delayed in the case of counterparts with only . We discuss our results in terms of current theories of sentence processing.     

Children's comprehension of relative clauses

A review of the literature on children's use of relative clause constructions reveals many contradictory findings. The suggestion is that some studies fail to take into account the two factors of embeddedness (role of complex noun phrase within the sentence) and focus (role of head noun in the relative clause). The experiment reported here attempted to reconcile the disparate findings and extend the range of constructions examined. 114 children between the ages of 3 and 7 served as subjects in a test of comprehension using an act-out procedure of 9 different relative clause sentences that exhaust the possible combination of 3 roles of the complex noun phrase in the sentence and 3 roles that the head noun plays within the relative clause (in each case, subject, driect object, and indirect object). All constructions were understood better with increasing age of the children sex and sentence set were nonsignificant variables. The results reveal a difficulty in ordering of the 9 types of construction that is in keeping with a prediction based on surface structure processing strategies.This research was supported by grant No. BNS73-09150 from the National Science Foundation to Professor Roger Brown.     

Effects of Local and Global Context on Processing Sentences with Subject and Object Relative Clauses

An eye tracking study investigated the effects of local and global discourse context on the processing of subject and object relative clauses, whereby the contexts favored either a subject relative clause interpretation or an object relative clause interpretation. The fixation data replicated previous studies showing that object relative clause sentences were more difficult to process than subject relative sentences. Crucially, however, the reading difficulty asymmetry between subject and object relative clause sentences disappeared when the sentences were presented with a local or a global discourse context that favored the objects in the object relative clauses. These findings demonstrate that the evidence for a syntax-based account of sentence processing is found when sentences are presented in isolation. However, if sentences are placed more naturally, in context, discourse factors outweigh the initial structural assignment.     

A Study on Late Closure in Spanish: Principle-grounded vs. Frequency-based Accounts of Attachment Preferences

A much-debated issue in current research on sentence parsing concerns the resolution of attachment ambiguities. Parsing theories differ on the procedures used to guide on-line attachment decisions: principle-grounded theories (e.g. garden-path/ construal) propose universal principles that minimize processing load; frequency-based accounts (such as linguistic tuning) claim that attachment decisions are shaped by readers' previous experience with their particular language; lexically based models, in turn, assert that attachment choices are determined by the properties of individual lexical items in the sentence. This paper reports six studies on late closure in the resolution of attachment ambiguities in Spanish: a questionnaire study on attachment preferences and three self-paced reading experiments where ambiguous PPs could be attached as arguments of two VP-hosts in NP-VP1-NPVP2-PP structures. The results show a clear preference towards low attachment (late closure), thus supporting principle-grounded theories. In addition, two corpus studies were carried out to obtain records of relative frequencies of the attachment choices involved in the experiments. A coarse-grained measure revealed that, in accordance with linguistic tuning, low-attachment structures are more common in Spanish NP-VP1-NP-VP2-PP sentences. However, a fine-grained count showed that low-attachment preferences cannot be explained by arguing that the specific verbs positioned lower on the tree (VP2) are more likely to take an extra argument than those located at a higher position (VP1), as lexical models would assume.     

Relative Clause Attachment: Nondeterminism in Japanese Parsing

Like many other languages, Japanese is ambiguous about the potential attachment sites of a relative clause which appears alongside a complex noun phrase. One important class of parsing theories (nondeterministic accounts) predicts that the relative clause will initially be attached to the first available host and later assigned to an alternative site, following the intervention of certain discourse processes. An alternative (deterministic) account maintains that the preliminary attachment site remains unchanged throughout the course of processing the sentence. The results of a questionnaire and a self-paced reading study with Japanese materials of this kind provided evidence of phased attachment first to one noun host and then to another. It is argued that these results are compatible only with nondeterministic accounts of parsing.     

The influence of referential processing on sentence complexity

This paper reports the results of five experiments designed to investigate the effects of referential processing on sentence complexity. Gibson (Cognition, 68 (1998) 1) suggested that sentence complexity is related to the locality of integrations between dependent syntactic heads, and that an appropriate measure of locality is the number of new discourse referents intervening between the endpoints of those integrations. The experiments in this paper test, modify and extend Gibson's (1998) claims. Each experiment manipulated noun phrases (NPs) in the subject positions of object-extracted relative clauses in order to determine how different types of NPs affected sentence complexity. Experiments 1, 2 and 3 used questionnaires to gauge sentence complexity, whereas Experiments 4 and 5 used self-paced reading. The results from Experiments 1, 2, 4 and 5 suggest that the complexity of the experimental items was more closely related to the Givenness status of the embedded subject in the Givenness Hierarchy than to whether the embedded subject was old or new to the discourse. Experiment 3 compared materials in which a quantifier was rotated through subject positions of a nested relative clause structure. The results of this experiment support a discourse-processing-based distance metric for computing locality and provide evidence against a pure similarity-based account of structural complexity such as proposed by Bever (Bever, T. G. (1970). The cognitive basis of linguistic structures. In J. R. Hayes (Ed.), Cognition and the development of language (pp. 279-362). New York: Wiley).     

The processing of subject and object relative clauses in Spanish: An eye-tracking study

A normative study and an eye-tracking experiment investigated the influence of animacy on the processing of subject and object relative clauses in Spanish. The results showed that object relative clauses caused more difficulty than subject relative clauses, but that animacy modulated this preference. The overall pattern was similar to findings in other languages. However, because of the syntactic characteristics of Spanish relative clauses, the results give novel insights into the processing mechanisms that underlie relative clause processing.