Use of verb information in syntactic parsing: evidence from eye movements and word-by-word self-paced reading

We investigated whether readers use verb information to aid in their initial parsing of temporarily ambiguous sentences. In the first experiment, subjects' eye movements were recorded. In the second and third experiments, subjects read sentences by using a noncumulative and cumulative word-by-word self-paced paradigm, respectively. The results of the first two experiments supported Frazier and Rayner's (1982) garden-path model of sentence comprehension: Verb information did not influence the initial operation of the parser. The third experiment indicated that the cumulative version of the self-paced paradigm is not appropriate for studying on-line parsing. We conclude that verb information is not used by the parser to modify its initial parsing strategies, although it may be used to guide subsequent reanalysis.     

Parsing of algebraic expressions by experienced users of mathematics

The present study investigates how experienced users of mathematics parse algebraic expressions. The main issues examined are the order in which the symbols in an expression are scanned and the duration of fixation. Two experiments tracked the order in which the symbols of an expression were scanned. The results were analysed using Markov Chain models of the scanpath data and provided strong support for the hypothesised scanning order: a left-to-right, top-to-bottom syntax-based scanning order. Length of fixation was also analysed in the first experiment. When reading text, readers pause significantly longer at the end of clauses and sentences. A similar pattern was found for mathematical expressions: Symbols at the end of a phrasal constituent were fixated upon for significantly longer than symbols at the start or middle of the phrasal constituent. These results suggest that the parsing of algebraic expressions has marked similarities with the way in which sentences of natural language are processed and reinforces the importance of syntax in their comprehension.     

Phonological codes and functional spelling units: Reality and implications

The grapheme-to-phoneme correspondence procedure for generating nonlexical phonological codes comprises two stages. The first, a parsing procedure, divides a letter string into functional spelling units (FSUs), each unit corresponding to a single phoneme. The second stage assigns the appropriate phoneme to the spelling units according to predictable or regular spelling-tosound patterns. In a lexical decision task, it was found that disrupting the spelling units, by alternating the case of letters within the units (e.g., ChuRCH) caused the pseudohomophone effect to be abolished. In contrast, disrupting the visual appearance of the letter strings but leaving the FSUs intact (e.g., CHurCH) allowed the pseudohomophone effect to emerge. The results are considered in the light of the recent “activation and synthesis” model proposed by Glushko (1979).     

Lexical availability and grammatical encoding scope during spoken sentence production

Three sentence production experiments investigate the relationship between lexical and structural processing scope. Speakers generated sentences with varying phrase structures in response to visual displays (e.g., The dog and the hat move above the fork and the tree/The dog moves above the hat and the fork and the tree). On half of the trials, one of the pictures in the arrays was previewed. Filler sentences varied preview position and sentence structure from trial to trial. When speakers could not anticipate the position of the previewed picture in the upcoming sentence (Experiment 1), preview benefit for pictures corresponding to the second noun to be produced was limited to pictures that fell within the sentence-initial phrase. When the linear position of the previewed picture was predictable, preview benefits were observed for the second noun to be produced, irrespective of phrase position (Experiment 2). However, no preview benefits were observed for the third noun to be produced (Experiment 3). In contrast, significant effects of initial phrase structure were observed in all experiments, with latencies increasing with initial phrase length. The results are consistent with speakers operating a phrasal scope for structural planning within which the scope of lexical access can vary.     

Lexical decision in sentences: Effects of syntactic structure

A lexical decision paradigm was used to examine syntactic influence on word recognition in sentences. Initial fragments of sentences were presented visually (CRT display) one word at a time (at reading speeds), from left to right. The string terminated with the appearance of a lexical decision target. The grammatical structure of the incomplete sentence affected lexical decision reaction time (RT). In Experiment 1, modal verb contexts followed by main verb targets and preposition contexts followed by noun targets produced lower RTs than did the opposite pairings (i.e., modal/noun and preposition/verb). In Experiment 2, transitive verb contexts followed by noun targets and subject noun phrase contexts followed by verb targets yielded lower RTs than did the opposite pairings. Similar contrasts for adjective targets did not yield comparable effects in Experiment 2, but did when the adjective was the head of a predictable phrase (Experiment 4). In Experiment 3, noun targets yielded lower RTs than did verb targets after contexts of a transitive verb followed by a prepositional phrase. An account of these effects is offered in terms of parsing constraints on phrasal categories.     

Planning in sentence production: Evidence for the phrase as a default planning scope

Controversy remains as to the scope of advanced planning in language production. Smith and Wheeldon (1999) found significantly longer onset latencies when subjects described moving-picture displays by producing sentences beginning with a complex noun phrase than for matched sentences beginning with a simple noun phrase. While these findings are consistent with a phrasal scope of planning, they might also be explained on the basis of: (1) greater retrieval fluency for the second content word in the simple initial noun phrase sentences and (2) visual grouping factors. In Experiments 1 and 2, retrieval fluency for the second content word was equated for the complex and simple initial noun phrase conditions. Experiments 3 and 4 addressed the visual grouping hypothesis by using stationary displays and by comparing onset latencies for the same display for sentence and list productions. Longer onset latencies for the sentences beginning with a complex noun phrase were obtained in all experiments, supporting the phrasal scope of planning hypothesis. The results indicate that in speech, as in other motor production domains, planning occurs beyond the minimal production unit.     

A Multiple‐Channel Model of Task‐Dependent Ambiguity Resolution in Sentence Comprehension

Traxler, Pickering, and Clifton (1998) found that ambiguous sentences are read faster than their unambiguous counterparts. This so‐called ambiguity advantage has presented a major challenge to classical theories of human sentence comprehension (parsing) because its most prominent explanation, in the form of the unrestricted race model (URM), assumes that parsing is non‐deterministic. Recently, Swets, Desmet, Clifton, and Ferreira (2008) have challenged the URM. They argue that readers strategically underspecify the representation of ambiguous sentences to save time, unless disambiguation is required by task demands. When disambiguation is required, however, readers assign sentences full structure—and Swets et al. provide experimental evidence to this end. On the basis of their findings, they argue against the URM and in favor of a model of task‐dependent sentence comprehension. We show through simulations that the Swets et al. data do not constitute evidence for task‐dependent parsing because they can be explained by the URM. However, we provide decisive evidence from a German self‐paced reading study consistent with Swets et al.'s general claim about task‐dependent parsing. Specifically, we show that under certain conditions, ambiguous sentences can be read more slowly than their unambiguous counterparts, suggesting that the parser may create several parses, when required. Finally, we present the first quantitative model of task‐driven disambiguation that subsumes the URM, and we show that it can explain both Swets et al.'s results and our findings.     

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.     

When Does Gender Constrain Parsing? Evidence from ERPs

We review the implications of recent ERP evidence for when and how grammatical gender agreement constrains sentence parsing. In some theories of parsing, gender is assumed to immediately and categorically block gender-incongruent phrase structure alternatives from being pursued. In other theories, the parser initially ignores gender altogether. The ERP evidence we discuss suggests an intermediate position, in which grammatical gender does not immediately block gender-incongruent phrase structures from being considered, but is used to dispose of them shortly thereafter.     

Parsing to Learn

Learning a language by parameter setting is almost certainly less onerous than composing a grammar from scratch. But recent computational modeling of how parameters are set has shown that it is not at all the simple mechanical process sometimes imagined. Sentences must be parsed to discover the properties that select between parameter values. But the sentences that drive learning cannot be parsed with the learner's current grammar. And there is not much point in parsing them with just one new grammar. They must apparently be parsed with all possible grammars, in order to find out which one is most successful at licensing the language. The research task is to reconcile this with the fact that the human sentence parsing mechanism, even in adults, has only very limited parallel parsing capacity. I have proposed that all possible grammars can be folded into one, if parameter values are fragments of sentential tree structures that the parser can make use of where necessary to assign a structure to an input sentence. However, the problem of capacity limitations remains. The combined grammar will afford multiple analyses for some sentences, too many to be computed on-line. I propose that the parser computes only one analysis per sentence but can detect ambiguity, and that the learner makes use of unambiguous input only. This provides secure information but relatively little of it, particularly at early stages of learning where few grammars have been excluded and ambiguity is rife. I consider three solutions: improving the parser's ability to extract unambiguous information from partially ambiguous sentences, assuming default parameter values to temporarily eliminate ambiguity, reconfiguring the parameters so that some are subordinate to others and do not present themselves to the learner until the others have been set. A more radical alternative is to give up the quest for error-free learning and permit parameters to be set without regard for whether the parser may have overlooked an alternative analysis of the sentence. If it can be assumed that the human parser keeps a running tally of the parameter values it has accessed, then the learner would do nothing other than parse sentences for comprehension, as adults do. The most useful parameter values would become more and more easily accessed; the noncontributors would drop out of the running. There would be no learning mechanism at all, over and above the parser. But how accurate this system would be remains to be established.     

Processing coordinated structures: incrementality and connectedness

We recorded participants' eye movements while they read sentences containing verb-phrase coordination. Results showed evidence of immediate processing disruption when a reflexive pronoun embedded in the conjoined verb phrase mismatched the sentence subject. We argue that this result is incompatible with models of human parsing that employ only bottom-up parsing procedures, even when flexible constituency is employed. Models need to incorporate a mechanism similar to the adjoining operation in Tree-Adjoining Grammar, in which one structure is inserted into another.     

Seeing a phrase "time and again" matters: the role of phrasal frequency in the processing of multiword sequences

Are speakers sensitive to the frequency with which phrases occur in language? The authors report an eye-tracking study that investigates this by examining the processing of multiword sequences that differ in phrasal frequency by native and proficient nonnative English speakers. Participants read sentences containing 3-word binomial phrases (bride and groom) and their reversed forms (groom and bride), which are identical in syntax and meaning but that differ in phrasal frequency. Mixed-effects modeling revealed that native speakers and nonnative speakers, across a range of proficiencies, are sensitive to the frequency with which phrases occur in English. Results also indicate that native speakers and higher proficiency nonnatives are sensitive to whether a phrase occurs in a particular configuration (binomial vs. reversed) in English, highlighting the contribution of entrenchment of a particular phrase in memory.     

Memory requirements and local ambiguities of parsing strategies

We present a method for calculating lower bounds on the space required and local ambiguities entailed by parsing strategies. A fast, compact natural language parser must implement a strategy with low space requirements and few local ambiguities. It is also widely assumed in the psycholinguistics literature that extremely limited short-term space is available to the human parser, and that sentences containing center-embedded constructions are incomprehensible because processing them requires more space than is available. However, we show that the parsing strategies most psycholinguists assume require less space for processing center-embedded constructions than for processing other perfectly comprehensible constructions. We present alternative strategies for which center-embedded constructions do require more space than other constructions.     

Plausibility and argument structure in sentence comprehension

In two experiments, we investigated how reading time was affected by the plausibility of the prepositional phrase in subject-verb-noun-phrase-prepositional-phrase sentences, and the status of the prepositional phrase as argument versus adjunct of the verb. Highly plausible prepositional phrases were read faster than less plausible ones, and argument prepositional phrases were read faster than adjuncts. These effects appeared both in a self-paced reading experiment and in an experiment that measured eye movements during normal reading. The effects of plausibility were substantially larger and longer lasting than the effects of argument status, but both appeared very early in the reading of the prepositional phrase. The implications of these effects for models of parsing and sentence interpretation are discussed.     

A computational model of human parsing

This report describeslicensing-structure parsing (LS-parsing), a computational model of human parsing. LS-parsing corresponds to human parsing at three points: (1) the order in which the LS parser builds nodes is psychologically more accurate than the orders in which either LL or LR parsers build nodes, (2) the LS parser's preferences in resolving local ambiguities are preferable to Frazier's strategies on both empirical and theoretical grounds, and (3) the backtracking strategies the parser uses when it has made an error at an earlier choice point model the distinction betweenweak andstrong garden paths-strong garden paths being irrecoverable, weak garden paths causing psychological difficulty, but not preventing recovery of the correct structure.     

Case marking in the Parsing of Complement Sentences: Evidence from Eye Movements

An eye-tracking experiment investigated the role of case-marking in parsing. We manipulated the case of pronouns in reduced complement sentences like I recognized you and your family would be unhappy here and I recognized she and her family would be unhappy here , in which the nominative pronoun she immediately disambiguates the sentences, in contrast to the ambiguous you . The nominative pronoun she disambiguates the sentence because I recognised she is ungrammatical, and thus she and her family must be the subject of an embedded sentence and not the NP-object of the preceding verb. Subjects took longer to read she and her family than you and your family during initial processing. The pattern reversed at the disambiguating phrase would be . Unambiguous control sentences containing the complementizer that did not produce case-marking effects. These results demonstrate very rapid effects of case-marking on parsing. Either case information is used immediately, or it is employed after an extremely short delay. We discuss implications for current theories of parsing.     

Working memory and planning during sentence production

Speakers retrieve conceptual, syntactic and lexical information in advance of articulation during sentence production. What type of working memory (WM) store is used to hold the planned information before speaking? To address this question, we measured onset latencies when subjects produced sentences that began with either a complex or a simple initial noun phrase, while holding semantic, phonological or spatial information in WM. Although we found that subjects had longer onset latencies for sentences beginning with a complex noun phrase, showing a phrasal scope of planning, the magnitude of this complexity effect was not affected by any type of WM load. However, subjects made more syntactic errors (but not lexical errors) for sentences beginning with a complex noun phrase, suggesting that advance planning for these phrases occurs at a syntactic rather than lexical–semantic level, which may account for the lack of effect with various types of WM load in the current study.     

Syntactic prediction in language comprehension: evidence from either...or

Readers' eye movements were monitored as they read sentences in which two noun phrases or two independent clauses were connected by the word or (NP-coordination and S-coordination, respectively). The word either could be present or absent earlier in the sentence. When either was present, the material immediately following or was read more quickly, across both sentence types. In addition, there was evidence that readers misanalyzed the S-coordination structure as an NP-coordination structure only when either was absent. The authors interpret the results as indicating that the word either enabled readers to predict the arrival of a coordination structure; this predictive activation facilitated processing of this structure when it ultimately arrived, and in the case of S-coordination sentences, enabled readers to avoid the incorrect NP-coordination analysis. The authors argue that these results support parsing theories according to which the parser can build predictable syntactic structure before encountering the corresponding lexical input.     

A Probabilistic Model of Lexical and Syntactic Access and Disambiguation

The problems of access—retrieving linguistic structure from some mental grammar —and disambiguation—choosing among these structures to correctly parse ambiguous linguistic input—are fundamental to language understanding. The literature abounds with psychological results on lexical access, the access of idioms, syntactic rule access, parsing preferences, syntactic disambiguation, and the processing of garden-path sentences. Unfortunately, it has been difficult to combine models which account for these results to build a general, uniform model of access and disambiguation at the lexical, idiomatic, and syntactic levels. For example, psycholinguistic theories of lexical access and idiom access and parsing theories of syntactic rule access have almost no commonality in methodology or coverage of psycholinguistic data. This article presents a single probabilistic algorithm which models both the access and disambiguation of linguistic knowledge. The algorithm is based on a parallel parser which ranks constructions for access, and interpretations for disambiguation, by their conditional probability. Low-ranked constructions and interpretations are pruned through beam-search; this pruning accounts, among other things, for the garden-path effect. I show that this motivated probabilistic treatment accounts for a wide variety of psycholinguistic results, arguing for a more uniform representation of linguistic knowledge and for the use of probabilistically-enriched grammars and interpreters as models of human knowledge of and processing of language.