Spatial, verbal processing and syntactic decoding

For the Token Test to be a valid measure of syntactic decoding competence, performance must be substantially independent of capabilities in the domain to which its items refer (spatial arrangements of tokens). McNicol & Morice (1985) found that scores on their new syntactically more complex version of the Token Test loaded on a visual-spatial as well as a language factor. This result is pertinent to a contentious issue concerning the extent to which the encoding of verbal and visual-spatial information is domain specific or dependent upon common central control,. If the encoding is domain specific, then a score on the Token Test may reflect either a single language deficit or multiple language, visual-spatial or other deficits according to the relative importance of such specific skills for task requirements. However, if the encoding is dependent upon central control, then a general cognitive deficit may underlie both linguistic and visual-spatial processing. This thesis reviewed clinical literature pertaining to the Token Test, particularly that which established its claim as a specific test of subtle receptive impairment. The findings of subsequent studies which posited specific and general non-verbal factors as critical for the test's discriminatory sensitivity were also examined. Literature which conceptualized the nature and processing of mental representations from an information processing viewpoint was reviewed. The dual and common-code or propositional positions represented by Paivio (1969) & Pylyshyn (1973) were outlined together with the findings of studies which have tested their alternative hypotheses. The possibility for resolution of these opposing views offered by the notion of working memory was considered, though with the accompanying emergence of fresh controversy between hypotheses of central and distributed processing control. An experiment was undertaken which involved two levels of syntactic complexity in the form of embedded and non-embedded clauses being manipulated with the presence or absence of inherently spatial referents. These were within the syntactic stimuli and in two corresponding interfering tasks. Inteference logic (Brooks, 1968) was used to determine whether the verbal and spatial processing requirements were parallel or single channel. Additive factor logic (Sternberg, 1969) was used to determine whether the syntactic decoding occurred within a same or separate stage from either the verbal or combined verbal and spatial processing demands of the syntactic task. Unexpectedly, the results showed an inconsistent complexity effect and, contrary to the Brooks (1968) prediction, there was a like modal facilitation trend of verbal task performance by the verbal interfering task. Design weaknesses were suggested to account for these outcomes. With these reservations, the data were tentatively suggested to most closely resemble the central processing model. This assumed that decoding of syntax occurred together with that of other verbal and spatial information. The central processing model was consistent with theories which posited either a common abstract propositional code (Pylyshyn, 1973) or central executive control (Baddeley, 1981). The implication of the findings for Token Test performance was that spatial ability and other general processing skills may well influence Token Test results. A need for further investigation incorporating methodological refinements was suggested, to enable stronger conclusions to be drawn.