The central issue that the dissertation addresses is: In what form is language available to the language user? One of the claims is that: clauses stage events (or actions) like a camera staging a film/ frame. The utterance/ understanding of a sentence is a spectacle. A major part of the thesis, therefore, is concerned with the presentational aspect of a clause.
Another aspect is the connection that a language user makes with the staged spectacle. We capture this through the notions of accommodation and field which constitutes a modified DRT. This is the formal tool that is used to capture this connection. The main tension of the thesis is of using a particular formal method to capture concepts that lie beyond the boundaries of the formalism.The answer to the question: How does the user get a grip on the presénted clause? has been the major thread of discovery in this dissertation -- the notion of salience. We claim that salience is a general cognitive apparatus through which the user computes the clause and thus gets a grip on it. We capture salience of clause through various asymmetries that are part and parcel of a clause -- like topic/ focus, AGRs/ AGRo, etc. Our claims are the following in this regard:
(i) asymmetries can be subsumed under a general notion of a new versus given opposition
(ii) asymmetries are reflected at each level of abstraction
Transitivity is the clearest of the asymmetries which represents the cognitive/ perceptive notion of salience. Psycholinguistic evidence show that for a child, the basic conceptual structure is that 'persons perform actions and things are affected by actions'. We read this as transitivity. We construct a syntactic account of transitivity where certain syntactic configurations and operations reflect extra-sentential notions like staging, scening, and event. Regarding (ii) above, a slow reading of the dissertation displays a general narrowing down of the scope from discourse structure to clausal structure to phrasal structure (chapters 1-4). Crucially, the claim is that at each level, the complexities involved is but a refraction of the complexities at a higher level. For the purpose of this dissertation these complexities are asymmetries. In the introduction (chapter 1), we elaborate the interconnections that obtain between various asymmetries and the given/ new distinction. We further discuss the syntactic impact that such interconnections may have on concepts like Staging, Scening and Event which together define the consequences of a clause in the totality of a discourse.
We extend DRT to a camera angle view of discourse (chapter 2). Discourse, in this model, is to be seen in terms of photographs. Language understanding takes place through the camera lens. We propose that field is the theoretical construct to capture a camera angle view. A camera which is sensitive to changes in the scene and records or arranges the snaps in an album. There is a universal set of fields which are part of human language processing/ understanding. Implementation of the theory demands a field to file mapping in terms of a salience gradient called the zoom potential. Zoom potential is calculated in terms of the transitivity of the clause. Transitivity is further reduced to predication and agreement. Connections between the head and the tail of asymmetries are established through agreement. The notion of field, we claim, will lead to a more efficient correspondence between the Kamp (1981) and the Heim (1982) versions of DRT.
In chapter 3 we discuss the notion of agreement as much as it bears upon our agenda. Agreement for our purpose serves the goal of identifying the participants for evaluating syntactic transitivity and therefore, ultimately, salience the major thrust of this study. Agreement provides finer details in a particular subroutine of an algorithm that we presented in the previous chapter.
The bulk of the chapter is devoted to the thesis that the object relation is more important; we try to see this in the light of a more general term like landmark. In this chapter, unergative clauses are first shown (section 3.2) to consistently contain a deep object position. The following sections discuss ergatives, transitives unaccusatives to argue that all of them have an object at some level of derivation. This discussion also includes revisions of the SplitVP Hypothesis and the Obligatory Case Parameter.
In section 3.5 we present our analysis of the phenomenon of longdistance agreement in Hindi, based on Watanabe's (1993) ThreeLayered Case Theory and claim that the analysis has an advantage over existing analyses in terms of the data that it covers as also the computational edge that it packages. In our terms an (actually) Lrelated position can be detected if we are able to track the different features like Fs (section 3.5 and 3.6 ) located/created during the derivation. This makes the task of producing a list of the typology of positions (in terms of the A/A' distinction, for example) easier. This is claimed to be the computational advantage of the present theoretical account.
In chapter 4 the phenomenon of (Noun) Classification in Bangla (and Hindi, to some extent) is discussed in conjunction with our drive towards discovering newer asymmetries down the clause highway. The inner stories of strength resolution of B(adge) and D(eclension) are revealed in order to flesh out the relevant phrase picture as much as it contributes to the clause picture. Definiteness, in this connection, seems to correlate strongly with the new/given distinction (section 4.1). Thus, the classifier as a cognitive category and its definiteness import is presented in section 4.3.
In section 4.10 we discuss PrincipleBased Parsing (PBP) in connection with the Bangla classifier system and show that a PBP approach along with a strong KB will give us the right results as far as the DPs in Bangla/Hindi are concerned. We propose (section 4.10.2.1) that Frames are phrase level computational variants of the thematic concept of scening which we claimed (section 1.4.3) determines the modality aspects of a clause and thus the parsing technique that we suggest enables a computation of scenes. Lastly (4.10.3), we propose a KB called WISE which solves certain residual problems of Bangla nominal syntax.