Alan Garnham

Avoiding the garden path : eye movements in context

Pragmatic factors, such as referential context, influence the decisions of the syntactic processor. At issue, however, has been whether such effects take place in the first or second pass analysis of the sentence. It has been suggested that eye movement studies are the only appropriate means for deciding between first and second pass effects. In this paper, we report two experiments using ambiguous relative/complement sentences and unambiguous controls. In Experiment 1 we show that referential context eliminates all the first pass reading time differences that are indicative of a garden path to the relative continuation in the null context. We observe, however, that the context does not eliminate the increased proportion of regressions from that disambiguating continuation. We therefore introduce a regression-contingent analysis of the first pass reading times and show that this new measure provides an important tool for aiding in the interpretation of the apparently conflicting data. Experiment 2 investigated whether the results of Experiment 1 were an artifact of the kinds of questions about the contexts that were asked in order to encourage subjects to attend to the contexts. The results demonstrated that the use of explicity referential questions had little effect. There was some small evidence for a garden path effect in this second experiment, but the regression-contingent measure enabled us to locate all garden path effects in only a small proportion of trials and to conclude that context does influence the initial decisions of the syntactic processor.

How language relates to belief, desire, and emotion understanding

Abstract Two experiments examined syntax and semantics as correlates of theory-of-mind (ToM). In Experiment 1 children’s language was examined at 3 years of age in relation to ToM at 3, 3.5, 4, and 5.5 years. Semantics predicted unique variance in later belief understanding but not desire understanding. Syntax did not explain unique variance in belief or desire. In Experiment 2 two measures of syntax and a measure of semantics were used with 65 3–5-year-olds. The syntax measures tested children’s understanding of word order and embedded clauses. They were related to false belief, but contrary to some predictions, were also related to emotion recognition. Performance on language control tasks with low syntactic demands correlated equally well with false belief. In both experiments performance on the syntax and semantics tasks was highly inter-correlated. We argue that ToM is related to general language ability rather than syntax or semantics per se.

Referential continuity and the coherence of discourse

Abstract Two experiments were carried out to investigate the role of referential continuity in understanding discourse. In experiment 1, a group of university students listened to stories and descriptive passages presented in three different versions: the original passages, versions in which the sentences occured in a random order, and randomised versions in which referential continuity had been restored primarily by replacing pronouns and other terms with fuller and more appropriate noun phrases. The original stories were remembered better, and rated as more comprehensible, than the random versions, but the restoration of referential continuity ameliorated the effects of randomisation. The descriptive passages had little referential continuity from one sentence to the next, and as expected the effects of randomisation on comprehensibility and memory were negligible. In experiment 2, a group of skilled comprehenders and a group of less skilled comprehenders were selected from a population of 7–8-year-old children. The difference between the groups was known to be largely their inferential ability in reading texts. Both groups read a series of short stories presented in the same three versions as used in the previous experiment. As predicted, the ameliorating effects on memory of restoring referential continuity in a randomised story were confined to the skilled group. The results are discussed in relation to the theories of story grammar, text microstructure, and mental models of discourse.

The use of stereotypical gender information in constructing a mental model: Evidence from English and Spanish

Four experiments were carried out to investigate how general knowledge about the stereotypical gender of participants in a text influences comprehension. A self-paced reading task was used to present short texts comprising one, two, or three sentences. The first sentence of each text introduced a stereotypically masculine or feminine participant (e.g. doctor, nurse), or a neutral one. The last sentence introduced a pronoun (he/she) that could match or mismatch the gender of the referent. The first experiment, which was carried out in English, showed that reading times for the last sentence were longer when there was a mismatch than when there was a match between the gender of the pronoun in the last sentence and the stereotypical gender of the referent in the first sentence. In contrast to English, the gender of the participant can be disambiguated by a preceding article (el/la) in Spanish. The results of the second, third, and fourth experiments, which were carried out in Spanish, showed that reading times for the first sentences were longer when there was a mismatch than when there was a match between the gender of the article and the stereotypical gender of the participant. However, reading times for the last sentences did not differ. Overall, the results suggest that information about the stereotypical gender of the participants in a text is incorporated into the representation as soon as it becomes available, and that it affects the ease with which the text is understood.

Mental models as representations of text

A theory of text representation is outlined which suggests that a text is encoded in a mental model that contains representations of only those individuals and events that are relevant to the interpretation of the text in question. Such models are constructed on-line in response to cues in the text and with reference to knowledge about the world. An experiment is reported which demonstrates that mental models represent the events most probably described in a text. The results confirm previous findings that memory for the content of a passage is not based on the construction of one of its linguistically motivated representations. However, unlike most previous findings, they do support a positive proposal about the nature of the representations on which memory for content is based.

Slips of the tongue in the London-Lund corpus of spontaneous conversation

This paper presents a list of slips of the tongue which occur in a corpus of English conversation transcribed from tape recordings. The kinds of error included in the list are briefly discussed, as are the criteria for detecting errors. The work forms a basis for an estimate of the frequency of such errors in ordinary speech.

Immediate activation of stereotypical gender information.

This article reports six experiments in which we explored whether gender stereotype information is typically invoked when certain role and profession terms are read and the extent to which the use of such information is under the reader’s strategic control. All of the experiments used a design in which subjects had to decide whether two terms (one an occupation and one a kinship term) could refer to the same person (e.g.,surgeon-brother orsurgeon—sister). The presentation conditions and the instructions were varied from experiment to experiment, to try to encourage the subjects to respond strategically and to suppress their use of gender stereotypes when responding. The results support not only the hypothesis that information about the stereotypical gender associated with occupations and roles is typically incorporated into the reader’s representation immediately, but also the hypothesis that such information is difficult or impossible to suppress. The implications of these findings for current theories of text processing and text representation are discussed.

Effects of syntax in human sentence parsing: evidence against a structure-based proposal mechanism

In 1992, D. C. Mitchell, M. M. B. Corley, and A. Garnham presented new evidence to suggest that contextual information does not influence the parsers initial decisions. They suggested, however, that if sufficient material separates the choice point from the point of syntactic disambiguation, the processor may have sufficient time to revoke an initial structure-based decision in favor of a more contextually compatible analysis, and so avoid the garden path. They described a reading time experiment that they claim is incompatible with an initial context-based decision. In the present article we argue that Mitchell et al.s contexts were in fact ineffective. We describe an experiment based on a subset of the Mitchell et al. design, but with differently structured contexts, and present eye movement data that are compatible with the claim that contextual information can influence the parsers initial decisions

Are inferences from stereotyped role names to characters' gender made elaboratively?

Two experiments provided evidence that gender stereotype inferences from role names—for example, that a surgeon is (probably) male—are made in a forward, elaborative, direction. We used sentences in which a person’s gender was never made explicit, but was implied in two different ways. The two ways were by the use of a role name, and by mentioning an item of clothing (e.g., a bikini) or a biological characteristic (e.g., giving birth) that is typically associated with females or males. The two pieces of information (role name and clothing/biological characteristic) were presented in different orders in the two experiments. In both cases a mismatch between the associated genders slowed reading, showing that gender information has been activated. It is argued that if an inference about gender is made on the basis of the second piece of information, hence slowing comprehension, it is unlikely that the inference about gender based on the first piece of information was not made immediately.

Generically intended, but specifically interpreted: When beauticians, musicians, and mechanics are all men

The influence of stereotype and grammatical information (masculine intended as generic) on the representation of gender in language was investigated using a sentence evaluation paradigm. The first sentence introduced a role name (e.g., The spies came out …) and the second sentence contained explicit information about the gender of one or more of the characters (e.g., …one of the women …). The experiment was conducted in French, German, and English. In contrast to English, stereotypicality of role names had no influence on readers’ male biased representations in French and German, where interpretations were dominated by the masculinity of the masculine (allegedly) intended as generic.