This book "asserts that the origin and spread of languages must be examined primarily through the time-tested techniques of linguistic analysis, rather than those of evolutionary biology" and "defends traditional practices in historical linguistics while remaining open to new techniques, including computational methods" and "will appeal to readers interested in world history and world geography."
AUTHOR: Jaszczolt, Kasia M. TITLE: Default Semantics SUBTITLE: Foundations of a Compositional Theory of Acts of Communication PUBLISHER: Oxford University Press YEAR: 2005 ISBN: 0199261989 ANNOUNCED IN: http://linguistlist.org/issues/17/17-638.html
Alexander Onysko, University of Innsbruck, Austria
This book proposes an integrative semantic theory of acts of communication by merging two distinct approaches to meaning construction and analysis: dynamic truth-conditional semantics and truth-conditional pragmatics. Kasia Jaszczolt specifically draws from the formal language of Discourse Representation Theory (DRT), which she adapts to be able to encode pragmatic information in her formal representations of Default Semantics Theory. The theory offers a cognitively plausible account of how utterance meaning is composed and interpreted in discourse.
The book is structured in two parts. Part 1 determines the theoretical foundations of Default Semantics Theory and Part 2 discusses applications of the theory for explaining the semantics of definite descriptions, propositional attitude reports, futurity, anaphora, sentential connectives, and number terms. The individual applications are exemplified from the perspective of the English language.
In the introduction Kasia Jaszczolt comes straight to the point and describes the essential claims of Default Semantics (repeated below in abbreviated form of the original, cf. xvi-xvii):
- The theory of discourse meaning is truth conditional and dynamic.
- Pragmatic information contributes to truth-conditional content.
- The representation of truth-conditional content is a merger of information from (i) word meaning and sentence structure (ii) conscious pragmatic processes (iii) default meanings
- Default meanings are conceived of as (i) cognitive defaults (properties of the human thinking process) (ii) social-cultural defaults (stemming from the way society and culture are organized)
- Default Semantics applies a dynamic approach to mergers in order to represent the meaning of multi-utterance discourse.
-Default Semantics uses an adapted and extended formalism of DRT, which is applied to the product of the merger, i.e. the merger representation.
- Merger representations are 'abstracts over thoughts'.
Thus, Default Semantics (hence DS) falls in line with the major views of neo-Gricean pragmatics, relevance theory, and optimality theory pragmatics that linguistic meaning underdetermines the content or proposition expressed and that pragmatic information is necessary for meaning completion. DS aims to achieve this holistic approach to meaning through merger representations which unite information from word meaning and sentence structure (WS), cognitive defaults (CD), social-cultural defaults (SCD), and from conscious pragmatic processes (CPI).
Following the introduction Chapter 1 discusses some fundamental questions that lie at the heart of DS. First of all, a merger of the semantic/pragmatic, the pragmatic/syntactic and the syntactic/semantic interfaces is postulated because in DS the interaction of the sources of meaning formation (e.g. WS as contributing syntactic information) are part of forming a compositional representation which appears as unified meaning on the level of merger representation. So for Jaszczolt, the relevant question in semantics is not what levels of meaning can be distinguished but where does meaning come from? However, DS also operates on a set of contributors to utterance meaning that is WS, CD, SCD, and CPI confined in merger representations (MRs) so that the question arises as to what extent these components contribute to the meaning of MRs. At the present stage of the theory all sources of information are conceived to contribute to meaning '''on an equal footing' that is by a 'conspiracy', an interaction whose mechanism is for the moment beyond our interest'' (p. 8). To investigate this interaction is actually a crucial question to solve within the framework of DS.
Instead, the author moves on to contend that by putting the construction of meaning on the level of MRs the dilemmas of ambiguity and underspecification vanish in the theory of DS. This claim is partly supported by the application of the modified Occam's Razor which denies the formation of unnecessary ambiguities. In turn, this leads to the assertion of parsimony of levels (PoL) meaning that ''levels of senses are not to be multiplied beyond necessity'' (p. 14). To complement the basic outline of DS, Jaszczolt touches upon the issue of why to apply a formal logical language for semantic representations. Furthermore, she critically discusses the notion of what is said and denounces the middle level of meaning which for lack of evidence can be sacrificed by Occam's Razor. The author also falls in line with proponents of direct communication (cf. Recanati 2002) dispensing with the idea of constant inferencing in communication. Alternatively, direct, i.e. literal, communication is the default mode of utterance decoding and inferences are only made if there are signals that diverge from a default utterance interpretation.
Chapter 2 expands on the notion of default meaning and establishes a basic distinction in cognitive and social-cultural defaults. Cognitive defaults are tied to the intentionality of mental states. Based on the principles that ''intentions allow for degrees'' and that ''the primary role of intention in communication is to secure the referent of the speaker's utterance'' (p. 51-52) three main scenarios of intentionality are distinguished. These correspond with three readings as demonstrated in the following sentence:
The author of Oscar and Lucinda is a very good writer (p. 50)
The strongest intentionality reading is captured by reference to the extension in the real world, i.e. the specific individual that has written Oscar and Lucinda (Peter Carey). The dispersed intentionality reading is essentially a misattribution of the referential extension in the real world, e.g. the speaker refers to Roddy Doyle while Peter Carey is the real author of the novel. Finally, the weak intentionality reading relates to the fact that the extension in the real world remains empty but categorial reference is made in the sense of 'whoever wrote the novel is a very good writer'. From these possible scenarios of intentionality, Jaszczolt concludes that strongest intentionality is the communicative norm and thus the cognitive default in utterance interpretation.
As far as social-cultural defaults are concerned, they can have an effect on the formation of merger representations dependent on the degree of cultural ingrainedness. Thus, the sentence ''Pablo's painting is of a crying woman'' will trigger, according to common cultural knowledge in westernized cultural areas, an interpretation that reference is made to a painting by Pablo Picasso. This example shows that such an interpretation is dependent on an individual's socially and culturally determined belief system and encyclopaedic knowledge which can be easily overruled by contextually imbued inferences. In fact, the notion of social-cultural defaults appears to work on a cline of contextual specification, where inferences based on general knowledge, i.e. social-cultural defaults, are inversely proportional to contextual specification of the utterance. Similarly, Jaszczolt observes that ''the boundary between […] social-cultural defaults and social-cultural inferences can only be assumed as methodologically desirable and psychologically plausible'' (p. 56). This leads her to conclude that DS has little to say about social-cultural defaults because their investigation lies outside the scope of semantic analysis and is an issue of anthropological linguistics and sociolinguistics.
After discussing defaults, Chapter 3 illustrates the central claim of Default Semantics Theory: the formation of meaning in merger representations. The basis of merger representations is the compositional nature of utterance meaning. In detail ''the meaning of the act of communication is a function of the meaning of the words, the sentence structure, defaults, and conscious pragmatic inference'' (p.72). These sources of meaning form the level of merger representation which is the unit of meaning analysis in the processing of truth conditional content and which is also the formalizable level of meaning analysis. Since compositionality of meaning only holds at the level of merger representation, compositionality is not expected on the level of sentence structure alone. This postulate differentiates Default Semantics Theory from Discourse Representation Theory (DRT, cf. Zeevat 1989) and Optimality-theory pragmatics (cf. Blutner and Zeevat 2004). According to the model of utterance interpretation in DS, a model hearer will in the first stage process truth-conditional content from a merger of information of (a) combination of word meaning and sentence structure, (b) conscious pragmatic inference, (c) cognitive defaults, and (d) social-cultural defaults. At a second stage in the process of utterance interpretation the hearer will resort to the processing of implicatures based on social-cultural defaults and conscious pragmatic inference.
The formal language of merger representations is based on an adaptation of the code of DRT with the caveat that formalization of the sources of meaning information applies only at the level of merger representations, i.e. to the output of the individual sources. Thus, ''formalization is always epiphenomenal on something that is in itself not formalizable'' (p. 87).
The chapters on applications of the theory in Part 2 follow the same basic outline. Different natural language phenomena are analyzed as merger representations following a differentiation of DS from the approach of discourse representation structure (DRS). Chapter 4 focuses on an analysis of definite descriptions. According to the degree of intentionality (strong, dispersed, and weak) expressed in the utterance ''the best architect designed this church'', separate MRs are given that exemplify default reading (e.g. Antoni Gaudi designed Sagrada Familia), non-default referential reading (e.g. Simon Guggenheim designed Sagrada Familia) and attributive reading (whoever designed Sagrada Familia is the best architect). Chapter 5 extends the formalization of MRs to propositional attitude reports as in ''Tom believes that the best architect designed this church''. DS tries to account for the discrepancies in knowledge between the holder of the belief and the reporter, as well as between the reporter and the hearer of the belief. Accordingly, a tripartite classification of MRs, in analogy to definite descriptions, account for different readings of the belief report: 'de re' which is the default reading relating to the proper referent in the real world, 'de dicto' which combines default reading with dispersed extension, and 'de dicto proper' in which both the believe and the attributive reading are bound to CPI.
The analysis of futurity in English is one of the core chapters of Part 2. The diverse functions of 'will' are based on the interface of modality and futurity encapsulated by 'will' as epistemic modality, as dispositional modality, and as future reference. Abstracting from Aristotle's classification of future as modal (cf. 1928), DS operates on an understanding of future as a bundle of time lines, representative of various undecided options that lie ahead of the present moment of an utterance. In order to cater for different degrees of modality in future time expressions, Jaszczolt introduces the Acc operator which is modelled on Grice's Acc operator (Grice, 2001). In analogy to the tripartite distinctions in previous chapters, the Acc operator is indexed in DS according to three degrees of modality/intentionality ranging from 'tf' as futurity and little modality, to 'fp' as reduced commitment on part of the speaker, i.e. increased modality, and to 'rf' showing the lowest degree of intentionality and the highest degree of modality. Since the essence of future is modality, the default is the most modal of the three forms, depicted in MRs as the regular future. These formalized states of modality/intentionality are also modelled in MRs on four other ways of expressing futurity/modality in English: futurative progressive ('fp'), tenseless future ('tf'), epistemic necessity 'will' ('rf'), and dispositional necessity ('rf'). Despite the plausible treatment of futurity, DS is still lacking integration of other modal expressions such as 'would', 'could', 'can', 'may', and 'might'. It also needs to solve the question of how formalized states of modality and intentionality can be put into a gradable relation to each other.
The remaining chapters deal with DS for presupposition as anaphora, sentential connectives, and the semantics of number terms. Binding and accommodation are central to the discussion of DS for presupposition as anaphora. Following the basic logic of equating default with the highest degree of intentionality, Jaszczolt argues that binding is the default choice in regular unstressed reference to topics while stressed anaphora, i.e. focus, leads to ambiguity resolution via accommodation in the context of the speech situation. The chapter on sentential connectives determines that conjunctions in English acquire enriched semantic meaning post-propositionally through CPI and by way of shortcuts through inference by SCD. Thus, 'and' in the sentence ''I dropped the glass and it broke'' evokes the interpretation of a causal and sequential event according to real world experience, i.e. by social-cultural default. In the same chapter, the author contradicts a dogmatic view of the restricted conditional reading 'iff' (if and only if) as being the only condition upon which the proposition in the main clause becomes true. For example, the condition expressed in ''if you mow the lawn, you'll get 5 dollars'' only holds true in the immediate space set up by the condition and is not the only way to get 5 dollars from the speaker in the real world. This observation ties in with recent approaches to conditionals in terms of mental spaces (cf. Dancygier and Sweetser 2005).
The applications of Default Semantics Theory come to a close with an investigation of the meaning of number terms. In general, number terms are scalar concepts that allow for different readings of at most, at least, and exactly. Arguing in terms of intentionality, the exact reading appears to be the default interpretation of number terms. However, the actual meaning of number terms is very much dependent on conventionalized utterance contexts as when number terms occur together with units of measurement, e.g. 2 km, 2 weeks. In such cases number terms acquire an approximative quality. To expand on these observations, the degree of approximation seems dependent on the precision of the units of measurement which also function as scalar concepts. Thus, 2 mm generally evokes a more exact reading than 2 km by sheer subordination of the unit of measurement.
As the summary indicates, Jaszczolt's book is a very complex and dense piece of work. With her theory of Default Semantics she successfully ventures into new ground of semantic territory. Her advance is cognitively plausible and the particular strength of the theory is its holistic approach merging pragmatic, social-cultural, syntactic, and lexical components of meaning. However, her exploration also stirs up new questions which are left unanswered at the present stage of the theory, as the author herself remarks. Some of these problematic issues are broached below.
While an inferentialist approach to utterance interpretation is disclaimed on grounds of a lack of cognitive evidence, the postulate of direct communication as the basic mode of utterance production and processing seems similarly unfounded and merely based on a belief in cognitive economy. This belief is also at work in the understanding of the modified Occam's Razor, leading to the postulation of parsimony of levels (PoL). Even though the claim of direct communication seems currently preferable due to empirical reasons, the lack of adequate investigations thus far should not suffice to deny prima facie the existence of intricate inferential patterns in utterance processing. This is certainly an area where our understanding is in dire need of appropriate psycholinguistic research.
The depiction of merger representations as the central element of Default Semantics also calls for future refinements. Despite aptly delineating the individual contributors to meaning (WS, SCD, CD, and CPI) and proposing that they principally contribute to meaning on an equal basis, the latter claim brings up the issue of how possible interactions of these components can be operationalized in different discourse situations. This is particularly evident in the category of social-cultural defaults, which, even though integrated into the model of merger representations, lack an adequate definition within the framework of DS. In the applications, SCDs only surface in the interpretation of 'and' as temporal and causal.
Furthermore, in the model of utterance interpretation, SCD and conscious pragmatic inference apply both at the stage of merger representations and at the stage of processing implicatures. This reflects a scale of SCD and CPI from deeply ingrained, i.e. largely invariant, to contextually dependent. The dividing line between SCD1/CPI1 and SCD2/CPI2 remains fuzzy in the present conception of the model which leads to an erosion of Stage II. This is emphasized in the applications of DS which focus on the building and interpretation of merger representations (Stage I) only. Thus, for future research it would be interesting to see the interrelation between the processing of truth-conditional content (Stage I) and the processing of implicatures (Stage II) at work in diverse communicative contexts.
According to DS, default is related to the degree of intentionality of mental states whereby the highest degree of intentionality translates into the default state of interpretation. Again the basic understanding seems cognitively plausible and is in sync with earlier claims of direct communication, i.e. parsimony of levels and the principle of primary intention. However, the current depiction of three stages of intentionality which lead to three different types of reading, proper extensional reference, dispersed extension, and categorial reference or attributive reading (empty extension), stand vaguely connected with strong intentionality, weaker intentionality, and weakest intentionality. An operationalization of degrees of intentionality and its mapping onto default and other types of readings seems essential in order to put the understanding of default as strong degree of intentionality on more solid grounds and render this principle applicable for further research.
Finally, a note on the presentation of the arguments is necessary. Due to the fact that the author summarizes the essence of the theory in the first chapter, the remaining chapters serve to develop these initial claims. This presupposes a certain amount of repetition which is generally welcome considering the complexity of the approach. However, tedious repetition of the main arguments (e.g. ''merger representations are abstracts over thoughts'', or ''DS offers a truth-conditional compositional theory of meaning of acts of communication'') in the theoretical part and in almost every single chapter of Part 2 appears as counterproductive to making a convincing case for the theory. In fact, this might give the impression that the author herself is not really sure of the foundations she convincingly argued for in the initial chapters of the book.
This argumentative dissonance should not cloud the interest in Default Semantics and its obvious advance in the analysis and understanding of utterance meaning. Constructed on credible cognitive foundations and successfully drawing a holistic picture of meaning interpretation, Default Semantics does not shy away from laying bare the limits of our current understanding of utterance semantics. This is why Default Semantics is a particularly stimulating and thought provoking theory that will continue to shape our vision of meaning.
Aristotle. 1928. The Works of Aristotle, Oxford University Press, London
Blutner, R. and H. Zeevat. 2004. Optimality Theory and Pragmatics. Basingstoke: Palgrave Macmillan.
Dancygier, Barbara, Eve Sweetser. 2005. Mental Spaces in Grammar: Conditional Constructions. Cambridge: Cambridge University Press.
Grice, P. 2001. Aspects of Reason. Richard Warner (ed.) Oxford: Oxford University Press
Recanati, F. 2002. ''Does linguistic communication rest on inference?'' Mind and Language 17: 105-26.
Zeevat, H. 1989. ''A compositional approach to Discourse Representation Theory''. Linguistics and Philosophy 12: 95-131.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Alexander Onysko is working as post-doctoral researcher and lecturer of linguistics at the English Department of Innsbruck University, Austria. His research interests lie in the fields of language contact, multilingualism, and cognitive semantics.