Aktiviteter

Language and Comprehension: Input, Process, Product

Symposium arranged by Danish Functional Linguistics

Time: October 13, 10.00-17.15 

Place: Room 16.1.8, University of Copenhagen (Amager)

Abstracts:

Vyvyan Evans 
Beyond Conceptual Metaphor Theory: Towards a Usage-based Account of Figurative Language
 

Conceptual metaphor theory, hereafter CMT (Lakoff 1993; Lakoff and Johnson 1980; Lakoff and Turner 1989), treats figurative language such as metaphor and metonymy as the outcome of stable knowledge structures (cross- and within-domain mappings) stored in long-term memory. More recently, CMT has increasingly argued for more abstract mappings, ‘primary metaphors’ , which are ever more distant from the actual language data that they are held to license (Lakoff and Johnson 1999).
However, this view of figurative language as deriving from ‘sub-symbolic’ knowledge ‘structures’ fails to address the dynamic and situated nature of figurative language in actual language use. Moreover, it fails to explain how figurative language as evidenced in discourse, particularly publicly-mediated discourse, evolves over time (see Musolff 2004; Zinken et al., In press).
Indeed, some researchers have argued that figurative language crucially involves situated inferencing (Sperber and Wilson 1995), relies upon dynamic processes of on-line meaning-construction giving rise to ‘emergent’ meaning (Fauconnier and Turner 2002) and draws upon the richly detailed ’semantic potential’ associated with linguistic units such as words (i.e., encyclopaedic knowledge, e.g., Croft 1993).
In this presentation, I present and take issue with some of the central assumptions of CMT, most notably its view of language use being determined by sub-symbolic cross-domain ‘structures’. I suggest that figurative language is better treated as an outcome of regular processes of prompted meaning-construction. Accordingly, I then argue for a dynamic, situated (i.e., usage-based) perspective (Evans To appear; Evans and Zinken 2005). I do so by briefly presenting the architecture for a new usage-based theory, and illustrate with typical examples found in the literature.
I argue that the new model can better explain similarities and differences between literal and figurative language as well as between metaphor and metonymy. In particular it allows us to address the difference between literal and figurative language use as a matter of degree, rather than as a principled distinction in meaning construction processes. I also consider how the usage-based model can successfully capture i) the relation between semantic representation and conceptual representation; ii) the difference in meaning construction involving metaphorically used forms on the one hand, and conventionally used forms, including conventionalised 'metaphors', on the other; and iii) the processes involved in meaning construction in general, and figurative meaning construction in particular.
 

TIL PROGRAMMET

________________________________________
 

Peter Harder
The Instructional Perspective on Meaning

The standard Platonic heritage in semantics does not only consist in a view of true meaning as objective and immutable. It also involves a view of meaning as ‘canonical output’: meaning is the ‘take-home’, definitive essence of a message. In the last decades, this picture has been challenged from a number of different angles: procedural semantics (in computation, and spreading to various linguistic frameworks such as mental space theory and relevance theory) social constructionism (in various forms of discourse analysis), Bakhtininan text analysis (which has also inspired linguists, including polyphony theorists); in the cognitive linguistics community it has manifested itself in the view of meaning as an instruction to perform a ‘mental simulation’ of a specific type. A shared element in this change is the increasing focus on the dynamics of meaning construction, which highlights process over output and in some versions ultimately implies that no definitive meaning can ever be ascribed to an utterance (“every decoding is another encoding”, in the words of David Lodge’s literature professor Morris Zapp).
This change has not been generally adopted in semantics, probably because it has been occurring from a number of different perspectives: ‘semantics’ is still generally understood to be a discipline dealing with the ‘content’ of linguistic utterances, as opposed to dynamic factors involving their use and interpretation. In this introductory talk I argue that a wholly ‘instructional’ account of meaning, i.e. an account viewing all meanings as input (rather than output) to the process of understanding, may be an important step towards a precise understanding of what meaning is, including also the classical ‘output’ properties of meaning.

TIL PROGRAMMET

____________________________________

Peter Juul Nielsen
Finnish Object Case: A Unified Description of Multifunctionality in a Sign Contrast

Finnish has 15 productive cases, and two of these, the accusative (ACC) and the partitive (PART), can be used for marking the object of a transitive clause:

(1) Minä löysin sinut
I+NOM found you+ACC
”I found you”

(2) Minä rakastan sinua
I+NOM love you+PART
”I love you”

The rules for choosing between the two object cases form a complex system which can be boiled down to three semantic oppositions: (I) entity delimitation: bounded vs. unbounded entity; (II) aspectual delimitation: bounded vs. unbounded State of Affairs (SoA); and (III) actuality: actual vs. non-actual propositional content (which is reminiscent of a polarity opposition).
The basic principle for choosing object case is that the values of all the three parameters have to favour ACC for it to be chosen: the object NP has to refer to a bounded entity, and the clause has to describe an aspectually bounded SoA the actuality of which is not negated or presented as doubtful or unlikely. Thus PART is chosen whenever one or more of the contrasting features are present.
I propose an analysis of this phenomenon based on an instructional interpretation of the semantics of contrasting signs. With such an approach it is possible to avoid the reductionism of calling it all just aspect (as has been proposed) while still retaining a sort of unified content, thus avoiding abandoning the concept of the linguistic sign as seen in Dik’s FG. I will describe the sign contrast ACC vs. PART as a contrast between two essentially under-specified and context-dependent invariant instructions that both refer to a specific Finnish conception of the prototypical transitive SoA (in the vein of Hopper & Thompson and Slobin). The content of ACC is identity with the transitivity prototype. The content of PART is to signal under-specified, abstract deviation from the prototype and acts as an instruction to the addressee to specify the relevant parameters from among those that govern the choice of PART. This presupposes the contextual presence of semantic specifications that will delimit this choice.
 

TIL PROGRAMMET

____________________________________

Lotte Dam & Helle Dam-Jensen
Reflections on the Interplay between Language and Context

It is a basic assumption that linguistic meaning is constructed on the basis of language combined with inputs from other sources. On this view, it is the aim of this paper to initiate a discussion of how information is structured at the lexical and at the functional level of linguistic analysis and how it interacts with context. This account makes a distinction between, on the one hand, the contribution of linguistic context and, on the other, the interpretive impact of the situational context.
 

TIL PROGRAMMET

____________________________________

Jan Nuyts
Conceptual versus Communicative Meanings: The Case of Deontic Modality and Directivity
 

Models of the layered representation of tense-aspect-modality categories as developed in Functional Grammar and Role and Reference Grammar include illocution in this hierarchical system (it is even ‘heading’ the system). As I have argued before (Nuyts 2001), in purely ‘logical’ terms this would not seem to be a cognitively plausible assumption. Illocution on one hand and semantic (‘qualificational’) categories such as aspect, tense or types of modality on the other are notions of a completely different nature, the latter having to do with the speaker’s understanding of reality (broadly defined), the former having to do with communicative goals and objectives. In other words, in a cognitive model qualificational categories (and their hierarchy) belong in conceptual structure, whereas communicative notions such as illocution do not, and are (in a language production perspective) only part of the devices which ‘translate’ conceptual information (including qualificational categories) into linguistic acts.
Unexpected empirical support for this view has turned up in a recent corpus-based investigation into the ‘deontic’ uses of a few modal auxiliaries in Dutch (mogen ‘may’ and moeten ‘must’; Nuyts, Byloo and Diepeveen 2005). Deontic modality is traditionally defined in terms of the notions of permission and obligation and other related concepts – notions which are obviously strongly reminiscent of illocution-related mood concepts such as the imperative or the hortative. The data show that – next to a few other meanings of no relevance here, including an evidential one – one must distinguish between dynamic, deontic and directive uses, whereby the deontic category should be defined purely in (conceptual) terms of moral acceptability or necessity, and the directive category is a matter of illocutionary notions, which may but need not be ‘inspired’ by deontic considerations. (The invenstigation also offers nice evidence in terms of the ‘division of labor’ between directive uses of the modals and the imperative.)
The difference between the deontic and the directive use of these forms obviously also taps into the issue of ‘instructional meanings’, even if not in a simple way – but at least one can say that the directive use (quite like the non-declarative mood categories) is explicitly instructional (it ‘means’ an instruction to the hearer to act in a certain way) whereas the deontic use (quite like other qualificational categories) is not explicitly instructional, at least not in the same sense of the notion ‘instruction’.
Nuyts, J. (2001). Epistemic modality, language and conceptualization. Amsterdam:
Benjamins.
Nuyts, J., P. Byloo and J. Diepeveen (2005). On deontic modality, directivity, and mood: A
case study of Dutch ‘mogen’ and ‘moeten’. Wilrijk: Antwerp Papers in Linguistics 110.
 

TIL PROGRAMMET

____________________________________
 

Henning Nølke
French Énonciation Linguistics: On Connectors
 

French énonciation linguistics deals in particular with the instructions yielded by the linguistic form with respect to the utterance act (l’énonciation) and, hence, the meaning in minimal (“constructed”) context including the interlocutors.
After a short introduction to (my conception of) instructional semantics, I shall discuss aspects of “discursive dynamics” (Ducrot’s TAL: The theory of argumentation within language) and of polyphony concentrating upon coding aspects of these phenomena. Taking the function of some French connectors as my key example, I shall try to show how these expressions organize discourse interpretation by means of their particular instructions. The analysis will also focus upon the role assigned to pragmatic and cognitive interpretation principles in this system.
This will be the opportunity to discuss the relations between conceptual and procedural meaning, and maybe to challenge the classical point of view according to which reference and truth conditions play the central (or primary) role in linguistic semantic analysis.
 

TIL PROGRAMMET

____________________________________

Michael Fortescue
The role of linguistic input and the ‘propositional prehension’ in constructing a mental model of a short anecdote
 

It has long been clear that narratives are rarely retained and reconstructed verbatim - what is typically stored in memory (as a ‘mental model’, to use the term introduced by Johnson-Laird 1983) is rather far removed from the original wording and may be actively retold from a much reduced or transformed trace of the original linguistic input. Standard recall methods can be combined with a procedural approach to linguistics to test hypotheses concerning the processes involved. The particular framework adopted in this paper is the Pattern and Process approach inspired by Whitehead (Fortescue 2001, 2003). The two essential concepts from that source that I shall employ are ‘prehension’ and ‘nexus’. In considering what minimal set of ‘instructional’ processes might be required to create a mental model from the words of a story that are read it is suggested that the ‘propositional prehension’ plays a central role in the building up of memorized structure. This approach exploits the notion of the decay of the ‘weak’ nexus links that characterize the relationship between form and function in symbolic systems - these contrast with the strong nexus links binding memory traces of real world situations and events. It assumes that the memory trace left by a simple story (read or heard) is the result of the gradual disintegration of weak links between strings of words and their content, ultimately leaving just the strong ones between the relevant parts of the associated content.
The Whiteheadian framework is broadly compatible with the motivating thrust behind mental model theory, although it does not sharply distinguish between propositional and sensory representations as the latter usually does (but cf. Garnham 1996 for dissent within the ranks). The distinction between propositional structure and propositional content is, on the other hand, crucial here. The framework not only intimately draws on context and general world knowledge but also combines the referential and the propositional approach to meaning in terms of two different species of acts that are involved in building up representations of events, namely the ‘indicative’ and the ‘predicative’, which combine to produce propositional prehensions. These processes, it is claimed, are inextricably interwoven in the construction of mental models, regardless of modality. In order to substantiate their psychological ‘reality’, the retelling of an anecdote by 24 native English-speaking subjects was elicited to abstract the constant structure behind the idiosyncratic versions produced by individual subjects. This empirical data was used to test specific predictions based on the theoretical framework.

Fortescue, Michael. 2001. Pattern and Process: A Whiteheadian perspective on linguistics.
Amsterdam/Philadelphia: John Benjamins.
Fortescue, Michael. 2003. The Pattern and Process of language in use: a test case. In Johanna Seibt
(ed.) Process Theories, Cross-Disciplinary Studies in Dynamic Categories, 177-218. Dordrecht: Kluwer Academic.
Garnham, Alan. 1996. Theories of language comprehension. In A. Garnham & J. Oakhill (eds.)
Mental Models in Cognitive Science, 35-52. Hove: Psychology Press.
Johnson-Laird, Philip. 1983. Mental Models. Cambridge: Cambridge University Press.

TIL PROGRAMMET

>> Forside