Book review

Methods in Cognitive Linguistics

Mónica González-Márquez, Irene Mittelberg, Seana Coulson and Michael J. Spivey (eds.) 2007. Methods in Cognitive Linguistics. Amsterdam: John Benjamins.

Reviewed by Martin Hilpert, ICSI Berkeley.

More and more researchers in Cognitive Linguistics employ empirical methodologies in their work. Partly responsible for this trend is the EMCL workshop series, which started as a student-run conference at Cornell University in 2003. Subsequent events took place in England (Portsmouth, 2004) and Spain (Murcia, 2006; in 2008 the workshop will come to Denmark. In the spirit of these meetings, the organizers of the first workshop now present the volume Methods in Cognitive Linguistics, which is intended as a pedagogical handbook.

Aiming to familiarize the reader with the entire spectrum of empirical methods available in current Cognitive Linguistics, the book consists of five thematically bound sections. These are preceded by a foreword from Leonard Talmy, who discusses the capacities and limitations of each methodology and emphasizes the need for a pluralism of approaches.

Section 1 includes a position paper that advocates the use of empirical methods, general outlines of corpus-based and experimental approaches, and a section on the use of statistics. Section 2 presents corpus-based approaches to discourse and grammar. The topic of section 3 is the empirical study of sign language and gesture. Section 4, the most comprehensive part of the book, is devoted to different experimental approaches. Section 5 closes the overview of methods with discussions of brain imaging and computational modeling of language.

Section 1 - Methods and motivations

In the opening paper, Ray Gibbs aims to draw cognitive linguists and psychologists closer together. A central point of the paper is that cognitive linguists need to state the results of their research in such a way that they are empirically testable. More often than not, claims in cognitive linguistics do not take the shape of hypotheses that could, in principle, be falsified.

Mittelberg et al. introduce the reader to corpus-based discourse analysis. They start with a brief overview of the theory behind this approach (Chafe 1998, Hopper 1998, DuBois 2003), thus motivating the methodological choice to study language in use, such as talk in interaction or co-speech gesture. Their practical discussion focuses especially on the combination of qualitative and quantitative approaches, and the chapter ends with an annotated list of commonly used corpora.

González-Márquez et al. provide a corresponding introduction to experimental techniques in cognitive linguistics. The paper first offers a tutorial on how to read an empirical psychology paper and then walks the reader through the essential steps of designing and conducting an experiment. Concepts such as null and experimental hypotheses, independent and dependent variables, and reliability and validity are spelled out clearly and applied to concrete examples.

To conclude the section, Rafael Núñez surveys methods of inferential statistics for cognitive linguistics, in particular the t-test, ANOVA, and the chi-square test.

Section 2 - Corpus and discourse analysis

Waugh et al. apply the theoretical discussion of Mittelberg et al. (see above) to two case studies of discourse phenomena. First, an investigation of personal pronoun usage in spoken French uncovers several recent changes in meaning and function of these items. The second case study is devoted to classroom interaction of native and non-native speakers of English. Both studies involve close readings of contextualized discourse and quantitative evidence in the form of frequencies.

Grondelaers et al. illustrate quantitative corpus research through a study of the Dutch pronoun er 'there' in presentational sentences of the type Tomorrow there is a meeting or Down the hall there is a bathroom. Under certain circumstances, the Dutch equivalents of these sentences allow the omission of the pronoun. It is the goal of the analysis to determine the factors that govern this alternation. Using logistic regression analysis, Grondelaers et al. find that initial locative adjuncts (down the hall) and use of the verb zijn 'to be' strongly correlate with omission of er. They further find that factors of dialectal region and register influence the alternation, if less strongly so.

Section 3 - Sign language and gesture

Wilcox and Morford survey empirical methods in signed language research. The study of signed languages is particularly fruitful from a cognitive linguistic perspective, since the visual modality strongly lends itself to metaphor, metonymy, and iconicity. However, there are some challenges to be met: There are no standardized transcription conventions, and corpora of signed languages are both laborious to compile and difficult to search. Wilcox and Morford thus advocate the inclusion of elicitation and experimentation techniques into a combination of empirical approaches to signed languages.

Sweetser introduces the reader to the analysis of co-speech gesture, which can shed light on the cognitive processes taking place during speech production. For instance, a gesturing speaker may iconically represent an object, a motor routine, or a path in space, thus revealing what conceptualization lies behind the on-going speech. Metaphorical aspects of gesture include the handling of surrogate objects - speakers literally try to hold back hostile arguments or accompany the presentation of a point with an open palm. Finally, deixis and point of view are continuously represented in gesture.

Mittelberg complements Sweetser's theoretical discussion of gesture with concrete guidelines on the recording and editing of video data, transcription issues of gesture-speech synchrony, annotation of gestural features, and the semantic and pragmatic interpretation of gestures.

Section 4 - Behavioral research

Carlson and Hill discuss experimental methods for the study of spatial language. They start with acceptability ratings, in which a subject provides an assessment of how well a sentence such as The toothpaste is above the toothbrush accurately describes a visual stimulus. Similarly, it can be measured how fast subjects can verify that a sentence does or does not correspond to a visual scene. A more interactive method is a placement task, in which subjects are asked to manipulate actual objects. Patterns in the placings allow inferences about the meanings of concepts such as above or in front of. Conversely, subjects can be confronted with a configuration of objects and be asked to describe the situation.

Bergen explores the idea that language understanding involves mental simulation. To understand a verb such as throw is to mentally simulate the physical movement and to imagine the consequences of that action. This theory makes a number of testable predictions. For instance, it predicts compatibility effects, in which a sentence such as The man hammered the nail into the wall primes the idea of downward movement and thus gives subjects an advantage for a subsequent task that involves such movement. Conversely, interference effects show that a mental simulation may make it more difficult to carry out a task that draws upon the same neural circuitry.

Hasson and Giora give a general overview of common methods in psycholinguistic research, in particular lexical decision and naming tasks, memory measures, item recognition measures, reading time experiments, self report measures and priming experiments. Each of these is illustrated with concrete examples.

Richardson et al. present eye-tracking as a revealing experimental paradigm in the study of language and cognition. The speed and involuntary nature of eye movements make it an ideal source for the study of cognition as it happens. The chapter surveys recent findings on word recognition, sentence processing, and the understanding of non-literal language, in particular fictive motion. It also includes a tutorial that explains the basics of eye-tracking methodology.

Brandone et al. make the case that infant studies provide unique insights into the cognitive organization of language in its formative stages. Behavioral methods for infant studies include the habituation paradigm, which can determine what aspects of a stimulus leads infants to perceive it as different from a previous stimulus. A lesser-known methodology is the intermodal preferential looking paradigm. This paradigm can be used to show that given two competing visual stimuli, infants will prefer to look at a visual stimulus that matches a simultaneous auditory stimulus. Given a stimulus such as Find the shoe, a picture of a shoe will receive more attention than a picture of a boat. Brandone et al. show how the two paradigms can be combined to study the acquisition of verb meanings.

Gor contrasts the processing of inflectional morphology in native speakers and L2 speakers in order to evaluate theoretical positions such as dual vs. single-system approaches to morphology. Evidence from nonce-verb generation and lexical decision tasks actually points towards a third alternative, i.e. a hybrid model of morphological processing.

Coulson explains the use of event-related brain potentials (ERPs) for cognitive linguistic studies. To illustrate, the so-called N400 potential refers to electrical activity that occurs in reaction to hearing a word and which is larger for words that are harder to process. It may thus be used for testing hypotheses about the relative processing difficulty of two linguistic stimuli. Coulson shows how ERPs can be operationalized for the empirical study of semantic frames, iconicity, and fictive motion.

To conclude the book, Edelman offers some suggestions on how the study of language can be bridged with its neighboring disciplines in the cognitive sciences, in particular computational and algorithmic approaches to cognition. An algorithm is presented that uses principles from Construction Grammar for the unsupervised identification of multi-word sequences, thus building a constructicon from raw corpus data. The chapter ends with a list of further challenges for computational cognitive linguistics.

Evaluation

A personal statement is in order here - I wish someone would have handed me a book like this one on my first day in graduate school. It would have made a world of a difference. Of course, graduate students are not the only ones who stand to benefit from reading it. The book draws together a wide range of approaches within cognitive linguistics, so that it will prove fruitful for anyone wanting to explore new or unfamiliar areas in this field.

With the praise out of the way, let me address a few criticisms. Introducing a novice readership to the complete current scope of empirical methods in cognitive linguistics is a task that is tremendous, if not altogether impossible. Compromises and trade-offs are inevitable. However, I am not certain that these have always been made at the right points. For instance, the current breadth of corpus-based approaches to cognitive linguistics is, to my mind at least, not adequately reflected in the contents of the volume. Further, each individual contribution is necessarily very brief, so that pedagogical aspects are sometimes lost in the attempt to cover more material. Another point concerns the fact that the articles could cross-reference each other more than they currently do. Given that each section is thematically very coherent and many theoretical concepts reappear, it would be economical to explain a term such as lexical decision task in detail where it appears first and reference that page in later papers. Finally, a short statistics refresher like the chapter by Núñez is very useful, but could be made more effective through corresponding exercises, which do not take much space and effort to produce.

To conclude, this book is sure to get the recognition it deserves. It is well-written, well-edited, and timely. Hopefully, there will be follow-up publications from subsequent EMCL workshops.

Links

© 2002-present ICLA; all rights reserved.
Site admin information and contact