On the typology of information integration processes
In my speech, I will try to show, using Vallduvì's (1992) sentence model, that the interactants of a discourse can generally integrate knowledge into their memory store in two ways (L and E integration). The choice of integration method depends on whether the linking element that connects to the informational status quo is conceptualised as a quasi circumstantial element referred to deictically ("by locality"), or as an anaphorically referable object ("by entity"). Subsequently, it will be shown in a comparison of data from various languages that the preferred linking method appears to be parametrised differently in each language. Finally, a preliminary typology of information integration will be suggested.
The contents of my contribution take the following order:
1. Locality vs. entity
The morpho-syntactic properties of a lexeme are shaped in part by a number of cognitive parametrisations of a reference object. The common characteristic of these parameters seems to be their expression of something akin to a degree of perceptual manageability of a reference object. Ikegami (2000) subsumed this intuition into the paired term "Tokoro-Mono" (place-object), where the core parameter of distinction is the degree of boundedness of the object in question. This dichotomy seems to be logically connected to the Nominal Mapping Parameter put forward by Chierchia (1998).
2. E and L Links
Vallduvì’s approach (Vallduvì 1992) intended to avoid the difficulties that can arise during the information processing of focus information. Starting from the Tokoro-Mono dichotomy, I will attempt to demonstrate that Vallduvì's link concept is based on ambiguous reference, oscillating between the referred item itself and its contextual localisation.
However, as the latter is deictic, i.e. spatial or temporal, and the former is identified by its quality, the result is two basic methods of cognitive integration of information: by locality and by entity.
3. Attempting a typology
Extending Chierchia’s findings to links in general, we can propose a parameterisation for informational integration. One prediction hereof is the deictically licensed possibility for some languages to transgress syntactic requirements such as the theta criterion and the case filter.
Since there appears to be a correlation between the tolerance of languages towards such transgressions and the preferred method of linking, any typology of informational integration would, at the same time, make certain predictions about the syntactic idiosyncrasies of the language in question.
Ikegami, Yoshihiko (2000): Nihongoron-e no Shotai (Invitation to a discourse on the Japanese language), Kodansha
Vallduvì, E. (1992): The informational component, New York: Garland
Chierchia, G. (1998): Reference to kinds across languages. Natural Language Semantics 6:339-405.