This is the first study that empirically investigates preposition placement across all clause types. The study compares first-language (British English) and second-language (Kenyan English) data and will therefore appeal to readers interested in world Englishes. Over 100 authentic corpus examples are discussed in the text, which will appeal to those who want to see 'real data'
Date: Wed, 11 May 2005 13:12:42 +0200 From: Rick Nouwen <R.Nouwen@lingua.uni-frankfurt.de> Subject: Experimental Pragmatics
EDITORS: Noveck, Ira A; Sperber, Dan TITLE: Experimental Pragmatics PUBLISHER: Palgrave Macmillan YEAR: 2004
Rick Nouwen, Graduiertenkolleg "Satzarten", Johann Wolfgang Goethe- Universität Frankfurt am Main
This book is a collection of papers on studies in the new field of "experimental pragmatics," which combines pragmatics with experimental psychology. Pragmatics, here, is to be understood in a narrow sense, namely as the study of how the interaction of linguistic properties and context influences the interpretation of an utterance. This type of pragmatic theory has traditionally been developed by linguists and philosophers of language who have used their own intuitions as the basis for their argumentation. Only recently has it been acknowledged that experimental methods could play an important role in creating stronger theories of "linguistic pragmatics." At the same time, a growing recognition has emerged that experimental psychology will benefit from studying pragmatics as well, since it has become clear that pragmatic processes play an important role in quite a lot of cognitive tasks. The editors of this collection stress, therefore, that the interaction between the different kinds of methodology used in experimental psychology and pragmatic theory will no doubt form empirically and theoretically stronger theories of both pragmatic phenomena and the underlying cognitive processes.
The book is divided into three parts. The first contains five chapters representing pioneering approaches to experimental pragmatics. These chapters describe longer periods of research on a specific topic and discuss how experimental psychology and linguistic pragmatics have interacted. In the second part, five papers introduce current topics in experimental approaches to pragmatic phenomena. The third and final part is completely devoted to the topic of scalar implicatures.
Part I: Pioneering approaches.
In the first chapter of this part, Herbert Clark and Adrian Bangerter review how the theory of (definite) reference evolved over the years. In particular, they show that this theoretical development was fuelled by three distinct methods: (i) armchair theorising (intuitions), (ii) experimental research and (iii) field observations. The authors discuss how reference was first thought of as an uncooperative process (the picking of a referent out of a set of alternatives) and how a series of theoretical and experimental work caused several adjustments to how we think about reference. After the philosophical work of Grice and theoretical studies into the phenomenon of bridging reference, referring is analysed a cooperative act. Subsequently, experimental research makes clear that reference involves an additional coordination process. In particular, it is shown that the interaction between the speaker and hearer is crucial for the act of referring to proceed smoothly.
In the next chapter, Raymond Gibbs uses four case studies from his own empirical work to illustrate how experimental research methods may be put to work to help settle theoretical debates in linguistic pragmatics. The studies discuss a variety of phenomena, including direct and indirect speech acts and the attributive/referential distinction for definite descriptions. Gibbs argues that studies like his illustrate the merits of a field like experimental pragmatics. The impact of experimental studies of linguistic pragmatic phenomena are in his view not to be underestimated. They offer a falsification framework for pragmatics, while at the same time they may resolve some of the scepticism that exists among psycholinguists and scholars in theoretical pragmatics with respect to the merits and the very possibility of experimental pragmatics.
In the third pioneering study, Sam Glucksberg argues against processing models that assume that non-literal meaning only emerges in a secondary processing stage. In such models, literal meaning is processed effortlessly, while non-literal (or 'speaker's') meaning only emerges after the defective literal meaning leads to additional inferences. Glucksberg discusses several experiments on metaphor comprehension and concept combination which show that pragmatic processes are automatic.
The next chapter, written by Guy Politzer, is concerned with the pragmatic analysis of reasoning tasks. In evaluating tasks, Politzer distinguishes a macro-analysis from a micro-analysis, which both are related to pragmatic theory. In the macro-analysis, the experiment itself (that is, the interaction between experimenter and subject) is essentially seen as a conversational setting and therefore subject to all sorts of assumptions and inferences which need to be reviewed. The micro-analysis of experimental tasks applies to the premisses, consequences, questions etc. that make up the task: it investigates what the actual propositions are the participant uses. Such an analysis is therefore essential to guaranteeing the validity of a task. Politzer illustrates his pragmatic analysis by describing several experimental studies, showing that experimental studies of reasoning cannot do without linguistic- pragmatics.
The final chapter of the first part of the book deals with quantifiers. It sums up 15 years of research into the processing properties of natural language quantifiers by the authors, Anthony Sanford and Linda Moxey. The goal of their research programme is to construct a psychological model of quantifier comprehension and to compare it with the predominantly formal linguistic and logical standard analyses of quantification. They concentrate on two functions of quantifiers their experimental studies have revealed: the so-called focus of a quantifier (the set a quantifier brings to mind) and the possibility for a quantifier to deny a contextual supposition. An important part of the studies described in this chapter deal with discrepancies between the experimentally attested data and predictions made by analytical properties of quantifiers. For instance, initial data suggested that the focus of a quantifiers is dependent on whether or not it has the logical property of being upward entailing. On closer investigation, it turned out, however, that it is rather the psychologically distinct notion of 'denial' that is related to quantifier focus. Sanford and Moxey show that the factors that play a role in quantifier processing are non-analytical and are represented in continua. Such results, the authors argue, illustrate the need for studying language comprehension from the point of view of communication.
Part II: Current Issues in Experimental Pragmatics
In first chapter of this part of the book, Jean-Baptiste Van der Genst and Dan Sperber discuss several experiments designed to test and confirm some explicit central consequences of Relevance Theory (Sperber and Wilson 1995). In particular, they focus on the two principles of relevance theory: (i) that human cognition typically involves the maximisation of relevance and (ii) that utterances involve a presumption of their own optimal relevance. The first, cognitive, principle is tested by a series of experiments on reasoning with comparison relations. It turns out that subjects are willing to draw only those conclusions that are potentially relevant. The second, communicative, principle is studied on the basis of a series of variations on Wason's Selection Task. It is shown that the performance on the task can be experimentally controlled by varying the context and as such the relevance of certain actions. Similar results are found for a production task.
The next chapter is written by Orna Peleg, Rechel Giora and Ofer Fein. These authors study context effects in linguistic processing and argue for the so-called "graded salience hypothesis." According to this hypothesis, during language processing, linguistic (i.e. lexical) meanings are ordered with respect to salience (on the basis of qualitative or quantitative familiarity.) The more salient a meaning is, the faster it will be accessed. A series of four experiments is presented that offers support for this hypothesis and yields interesting data on the interaction between lexical and contextual information.
Seanna Coulson's contribution to the book is the only one dealing with cognitive neuroscience and electrophysiological methods. Coulson gives a comprehensive and accessible overview of techniques in cognitive neuroscience, focusing in particular on so-called event-related brain potentials (ERPs). ERPs can be used to reveal systematic differences in brain responses to pairs of stimuli. They yield waveforms containing positive and negative peaks. The deflections of such a waveform are called components. Several specific ERP components have turned out to be functionally significant and have linguistic relevance. For instance, the amplitude of the so-called N400 component has been shown to vary inversely with the predictability of a certain word in a (preceding) context. Moreover, it has been shown that words in an ambiguous text show relatively greater N400 amplitudes than words in disambiguated text. This component is thus directly relevant to pragmatics. Since, moreover, ERPs offer a continuous direct measure of brain response, they are sensitive to different kinds of processing and consequently offer a more detailed measure than, for instance, reading times do. Coulson exploits this feature in a study of metaphor comprehension and shows that there is a continuum from literal to metaphorical interpretation rather than a dichotomy. Coulson concludes by suggesting some interesting future experiments which involve the study (and discovery) of pragmatic processes using ERPs.
The next study in the book discusses children's understanding of speech acts, with special focus on the case of promises. The authors, Josie Bernicot and Virginie Laval, present two experiments. In the first, the so- called preparatory condition is focused on. This condition says that something is a commissive act only if the listener prefers the speaker to perform the action in question. The experiment shows the relative late acquisition of this condition. A second experiment shows that textual and contextual aspects of processing start to interact around the age of 6. This experiment, which was performed in French, shows that when children are asked to process promises phrased using a future tense (instead of an explicit verb "promettre" (to promise)) in absence of a promise-specific context, they will only reconstruct the promise when the immediate (rather than the simple) future tense is used. In other words, the experiments show the close relation between textual and contextual factors in both language acquisition and language use.
The final chapter in the second part of the book concerns "even-if" conditionals and is written by Simon Handley and Aidan Feeney. Central to the experiments presented in this chapter are conditionals like "Even if I read everything on the reading list, I will fail the exam." Since it is very surprising that someone fails an exam while having read everything on the reading list, subjects are not eager to conclude that the speaker will in fact fail. This provides evidence against an analysis of even-if conditionals as universally quantified statements, where "even if A, B" is interpreted as "in every possible case (including A), B is true". In such an analysis one would expect that the truth of the consequence follows readily from any even-if conditional. Effects like the ones found in Handley and Feeney's experiment are therefore unaccounted for. In general, the experiments indicate that the inferences drawn from conditional assertions display a rich and varied pattern that involves not only logical reasoning but moreover the subject's background knowledge and expectations.
Part III: The Case of Scalar Implicatures
The final four chapters of the book all deal with scalar implicatures, a traditionally hot topic in pragmatics, but now hot in experimental pragmatics as well, as illustrated by this separate book-part addressing the implicature debate.
Three of the chapters on implicatures deal with the debate between two competing theories of pragmatic inferences. According to the "default" theory (for instance, Levinson 2000, referred to as the prototypical version of such a theory in several chapters), implicatures are automatically associated with a term. That is, the use of an expression that is relatively weak on a scale automatically implicates the denial of the stronger expressions on that scale. For instance, "some" is per default interpreted as "some but not all." Only when the inference is found to be incompatible with other information, can the implicature be retracted. The opposing theory (for instance Relevance Theory) views the weak interpretation as most salient to the term and assumes that implicatures involve processing effort.
Anne Bezuidenhout and Robin Morris study what happens when subjects have to process a (small) text containing a scalar term, where the utterances eventually do not support the implicature, but do support the weak interpretation of the term. In the "default" model, initial smooth processing is predicted with an eventually costly retracting of the default inference. In what they call the "underspecification model" (inspired by relevance theory), it is predicted that there will be no need for reanalysis, but that during processing the weak reading will simply be strengthened more and more. Bezuidenhout and Morris performed two eye- tracking experiments to test these predictions. The results are clearly compatible with the underspecification model. However, as the authors argue, the results do not so clearly falsify the default model. Nevertheless, Bezuidenhout and Morris stress that eye-tracker data like these are extremely valuable for the theoretical debate. Eye-tracker results do not just tell us whether or not people are aware of implicatures, but moreover indicate when in the comprehension process such inferences are made.
Also focusing on the debate between the "default" framework and models inspired by Relevance Theory is the chapter by Ira Noveck. He focuses on a specific developmental-pragmatic effect: children tend to give a weak logical interpretation to weak logical terms like "or", "some" or "might." For instance, children are much more likely to interpret "or" as inclusive than adults (who prefer the non-logical stronger exclusive interpretation). Both competing theories may offer an explanation for this effect. According to Relevance Theory, the logical meaning is minimal. The effect is then explained by assuming that the capacity for making pragmatic inferences is increased with age. The default model could explain the effect by assuming that pragmatic inferences only turn into defaults with age. Noveck argues that these explanations have a testable difference. Only according to the explanation inspired by Relevance Theory may the complexity of a task influence the outcome. That is, a sufficiently effortless task should trigger non-logical interpretations even with children that normally show to only adopt the weak meaning. Unfortunately, developmental data is difficult to interpret. Some results supporting the Relevance Theoretical model, however, are borne out. Very easy tasks indeed show adult-like behaviour with young children. Moreover, harder experiments with adult subjects tend to yield more logical meanings.
Anne Reboul's chapter also suggests experimental support for the non- default model of pragmatic inferences. Reboul is interested in comparatives of the form "Better tea than no coffee." In her experiments, subjects are asked to interpret such utterances and then to decide whether the speaker prefers tea or coffee and what he or she was actually given. She argues that the default model predicts that subjects confronted with such a comparison end up with an interpretative dead-end, since they will interpret "tea" as "only tea", rendering the sentence into the nonsensical "Better only tea than only tea." In two experiments, she shows that this prediction is not borne out.
The remaining chapter in the part on implicatures concerns a somewhat different debate on the defaultness of pragmatic inferences. A team of researchers consisting of Gennaro Chierchia, Maria Teresa Guasti, Andrea Gualmini, Luisa Meroni, Stephen Crain and Francesca Foppolo offers experimental results that apply to a theoretical complication introduced in Chierchia's non-experimental work. Chierchia has argued against a Gricean theory of pragmatic inferences, where implicatures are post- propositional "global" inferences.An iImplicature in fact turns out to be a rather local effect. For example, for a sentence "Mary is either reading a paper or seeing some students" the resulting implicated reading should be "Either Mary is reading a paper or she is seeing some students, but she is not seeing all students." The standard approach, however, seems to generate an unwanted inference, namely "it is not the case that Mary is either reading a paper or seeing every student." An additional problem for implicature computation follows from the observation that implicatures do not arise in downward entailing (DE) contexts. But since implicatures emerge locally, how are they going to blocked once embedded in DE- contexts? In Chierchia's proposal, the so-called Semantic Core Model, the "plain" and the "scalar" meaning of a sentence are both computed on the basis of a recursive bottom-up process. These two meanings are eventually compared and the most informative one is added to the context. The model explains readily why implicatures surface locally. Since only the most informative meaning "wins" and negation reverses informativity scales, the model also accounts for the fact that in DE-contexts scalar implicatures are not expected. The experiments presented in this chapter are designed to find support for Chierchia's theoretical model and focus on how adults and children interpret disjunction in DE and non-DE contexts. In a first experiment, it is shown that adults derive the (logical) inclusive reading for "or" rather than the exclusive reading significantly more often in DE contexts than in non-DE contexts. In a second experiment, children turn out to accept the inclusive reading of "or" much more often than adults do. In a third experiment, the predictions of the Semantic Core Model are tested directly, by asking subjects to perform a truth-value judgement task on (acted-out) utterances containing a disjunction embedded in either a DE or a non-DE context. Adults perform the task as the Semantic Core Model predicts: they compute the scalar implicature only in the non-DE context. Half the children are adult-like, while the other half computes the implicatures even in DE contexts. Using a final experiment, the authors intend to show that the children's behaviour cannot be accounted for in a Gricean model. By letting children compare two true utterances one of which is more informative (and therefore more appropriate) than the other, they test whether children know Grice's Maxim of Quantity. It turns out that children perform this task very well. Almost all children are capable of distinguishing the more appropriate true utterance from the less- informative true one.
"Experimental pragmatics" is an interesting and valuable contribution to both linguistic pragmatics and experimental psychology. It offers a clear outline of a new field and it convincingly shows why an experimental approach to pragmatics is attractive and indeed necessary.
I can recommend this book to researchers and students of pragmatics, psycholinguistics and the psychology of reasoning. The individual chapters are accessible even to readers who are less familiar with pragmatic theory or experimental methods. The studies in this book are generally of a high standard.
Clearly, however, the field of experimental pragmatics has not reached maturity yet. An "experimental pragmatics"-sceptic might claim that experimental research into pragmatic phenomena has little theoretical value since the theories in linguistic pragmatics are not explicit enough to be tested. Such criticism is still very much applicable (even if some contributing authors in this volume suggest otherwise). I fear that the theoretical impact of some of the studies is somewhat limited. In quite a few chapters of this book, it is not entirely clear why a certain theory makes the predictions the authors claim it does. In my view, this is because there is still a large gap between the models that are being tested and the linguistic pragmatic theories these models are based on. In particular, "standard" theories of pragmatics often fail to make explicit processing predictions. This is illustrated by the fact that in several chapters in this book, researchers do not test an existing theory, but rather their own processing model that is loosely based on one. Consequently, the falsification of such a model does not necessarily lead to the falsification of the theory. These critical notes do not mean, however, that the studies in this book are unconvincing. Overall, the findings are extremely valuable, even if their direct theoretical consequences are somewhat limited. The ultimate merit of the field of experimental pragmatics (and indirectly of this book) is exactly that by experimentation the theories of pragmatics will gradually lose their psychological vagueness. The large range of impressive experimental results that is presented in this book will force theorists to incorporate explicit experimentally attested processing parameters in their theories.
"Experimental Pragmatics" makes it clear that enormous progress is being made in this young collaboration between two fields and that the experimental results will eventually lead to new models of language use. By doing so, this collection is a big success.
Levinson, S.C. (2000). Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press.
Sperber, D., and Wilson, D. (1995). Relevance: Communication and Cognition (2nd edn). Oxford: Blackwell.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Rick Nouwen is a postdoctoral researcher at the Graduiertenkolleg "Satzarten," J.W. Goethe-Universität Frankfurt/Main in Germany. His research has mainly focused on the formal semantics and pragmatics of plural and quantified anaphora. His current work involves the meaning and use of quantifiers, with special attention to cognitive and logical aspects of reference to quantity.