An Optimality-based Inferential Model For Relevance-theoretic Verbal Comprehension

Relevance theory(RT) was first presented by Sperber & Wilson in 1986. Originating from Gricean studies on cooperative principles and conversional implicature,it exhibits great advancement through expounding verbal communication and the interpretation of natural language from the unprecedented perspective of cognitive pragmatics.It proposes to regard the context as cognition-oriented and mutual manifest,and divides cognitive effects into explicatures and implicatures.Besides,it also verifies contextual variability and the constancy of relevance,and identifies pragmatic inference as nontrivial,nondemonstrative and deductive.In a word,RT makes revolutionary contributions to inferential studies.In the past two decades, relevance theory has drawn much attention and been well discussed in the fields concerned.Though no doubt has been cast on its achievements,RT is not as perfect as its cofounders have expected.One of the principal defects is that a communicative principle of relevance as generalized as the one offered in RT cannot provide specific and detailed description of the process of verbal comprehension at a microcosmic level. All kinds of relevance-oriented inference count on and only on the guidance of cognitive and communicative principles of relevance,which are not only too general but also too weak to regulate the real inferential process.And this leads immediately to the theory's interpretative subjectivity and inadequacy in its explanatory power.What's more,this defect also results indirectly in the randomness of the theory's application. Due to the lack of precise and clear restraints,the presumption of optimal relevance (POR) and the relevance-theoretic comprehension procedure(RTCP) have been employed subjectively by different users to meet different ends.Covert shifts in the reference of certain key notions are made,consciously or not,by RT followers in their studies from time to time.All the above induce to a certain degree the unjustified omnipotence and non-falsifiability of RT.The present research takes as its study subjects the core claims of RT---POR and RTCP,the inferential mechanism derived from the former.Through adapting the research method of optimality theory(OT) in phonetics to the pragmatic field,it aims at constructing a highly-constrained relevance-theoretic inferential model based on an amended version of the original RTCP.It is expected that not only can RTCP's maneuverability and RT's explanatory power be promoted in this way,but also RT's applicability should be extended into an even larger range of linguistic phenomena owing to the newly proposed relevance-theoretic model.Most integrated studies of RT with other inferential theories are conducted for the sake of their mutual complementarity.All of such studies exploit each theory's strengths while avoiding its weakness.Such studies cannot solve or remove the inherent problems with those theories fundamentally.Integrated studies towards the enrichment and refinement of RT are barely seen in a real sense.Compared with existing studies, our research on theoretical combination features a unique quality:instead of adopting the evading strategy as others do,it faces and confronts relevance-based theoretical problems in a direct way.Through locating then correcting or amending RT's internal defects,it seeks to improve the theory in a substantial way.Since the combination of RT with the other theory ---OT is mainly conducted here for the refinement of RT, relevance and optimality in the inferential model to-be-presented do not share equal status.It would be wiser to regard the model as optimality-based inferential model for relevance-theoretic verbal comprehension(ROIM).Generally speaking,there are two principal issues to be settled for the comprehensive presentation of ROIM,the construction of relevance-concerned premises for optimality-based inferential model and the construction of optimality-based inferential model for relevance-theoretic verbal comprehension.To be specific,the present study first addresses the issue of high-subjectivity and weak-maneuverability in RT.It attempts to settle these general problems in the specific operating level of RT.And during this process three sources of the theory's presented imperfectness are found:unconformity between the expectation of relevance(ER) and speaker's intention,simplification on defining implicit meaning,and contradictions inside RTCP.While the first problem reflects the mismatch between RT's elaboration on comprehension from the hearer's perspective and its inheritance of Gricean speaker-centered notion---implicature,the second one shows the lack of consideration in RT as to the properties of implicature.Actually,implicature forms a cognitive category with its own prototypicality,which is caused by various degrees of the interpreter's certainty towards an implicature.As to the contradictions inside RTCP, they have a common root,viz.the vagueness and ambiguity in defining the reference of certain key notions such as relevance,the communicator's abilities and preferences,etc.Since the simplification or confusion in RT's key concepts is a main source of the problems with the theory,it is necessary to delve into these concepts and provide them with a refined analysis.Being part and parcel of the model-forming process,we call this part of the present research the construction of relevance-related premises for an optimality inferential model,where three decisive factors in a relevance-based inferential process are expounded,which are ER---the dynamic factor determining inferential directions,cognitive effects---the ultimate goal of pragmatic inference,and contextual assumptions---the materials employed in pragmatic inference.Special attention is given to the core functioning of these notions that is ignored or misunderstood by RT,i.e.the dynamicity of ER,re-categorization of cognitive effects, and factors other than accessibility that may affect the selection of contextual assumptions.Through a thorough inspection of these notions,the functions performed by them during the inferential process and their operating mechanisms,are revealed.In this way not only the general framework for pragmatic inference takes shape to a certain extent,but the attempt to solve the targeted problems of RT at the specific operational level also gets fully prepared.Based on the above preparations,amendment to RT is conducted in the following three aspects:expansion of the original inferential range by viewing communication from the interpreter's perspective,enrichment of the extant POR by extending each of its two sub-presumptions into several maxims and corollaries at different directions so as to prevent the confusion and contradiction suffered by the original POR,and adjustment of the relationship between POR and RTCP from that of deriving and being derived to that of being dominated and dominating.The above also form the fundamental notions of our optimality-based inferential model for relevance-theoretic verbal comprehension.In consideration of RT's subjectivity in application and weakness in its constraining capacity,the relevance-theoretic inferential model here is built in the framework of optimality theory(OT),which enjoys high objectivity and maneuverability thanks to its employment of an evaluating system made of a strict hierarchy of constraints according to their respective significance and suspendibility. However,since a conflict exists between the dynamicity of online inference and staticity of original OT,certain adaptation has to be made on the latter before it can be integrated into the construction of inferential model.To meet the current need,a specialized operating mechanism for the generator(Gen.) in an OT system is proposed, and the constraints in its corresponding evaluator(Eval.)are classified according to features such as dynamicity or relativity.Moreover,special emphases are laid on the interaction between Gen.and Eval.,on the candidates' sequential entrance into the latter from the former,and also on the non-uniqueness of the optimal candidates surviving the evaluation.After the adaptation is completed,optimality-based inferential model for relevance-theoretic verbal comprehension is constructed according to the following steps:first,to construct in Gen.a hierarchical sequence of constraints on the selection of contextual assumptions to be employed in the inferential process;then to work out in Eval.an ordered set of static constraints according to their respective relativity for the verification of candidates produced by Gen.;thirdly,to figure out the transforming way of the Evaluator's dynamic constraint during the inferential process,and its interaction with other static constraints in Eval.;finally,to explain the overall interaction between Gen.and Eval..The Generator's inner operational mechanism explains the sequential generation of cognitive effect candidates;the linear sequence of static constraints in Eval.decides the selection of optimal candidates,and it also classifies those candidates according to the certainty of speaker's intention embodied in them.As to the third step in constructing the inferential model,it reflects the online satisfaction of the constantly changing ER.Last but not least,the interaction between Gen.and Eval.(the influence exerted by evaluation outcomes on the Generator's further production of candidates,to be specific),presents the various termination ways of pragmatic inference.After the step-by-step construction,a simplified diagram is given to display the integrated functioning of all these steps in an interpreter-centered inferential process.As the last major part of the present research,the newly constructed inferential model is employed in the analysis of certain linguistic phenomena,such as online lexical replacing or intensifying which can not be explained or thoroughly explained by original RT,and argument constructions which have never been touched by RT before. Not only is the optimality-based relevance-theoretic model verified through this way, but studies on the target phenomena can also benefit from such discussions.To be specific,since the newly proposed inferential model contains the amendment on the nature of cognitive effects,its explanatory power in the field of lexical pragmatics surpasses that of the original.Linguistic phenomena such as ad hoc concepts undergoing online lexical substituting and a special case of lexical broadening,which used to be difficult to explain in the framework of RT,are easily analyzed and interpreted in our inferential model.In addition,the amended RT and its newly proposed inferential model offer a new analytical approach---linear analysis to the study of argument construction as well,which besides updating the previous explanation on constructional prototypicality,also provides a mechanism for the description and interpretation of a neglected phenomenon in the field of construction grammar---a single structure representing different constructions or various combinations of constructions.