Models Of Semantic Processing Research Articles

The unification of mind: Integration of hemispheric semantic processing

Seventy-six participants performed a visual half-field lexical decision task at two different stimulus onset asynchronies (50 or 750 ms). Word targets were primed either by a highly associated word (e.g., CLEAN–DIRTY), a weakly associated word (e.g., CLEAN–TIDY), or an unrelated word (e.g., CLEAN–FAMILY) projected to either the same or opposite visual field (VF) as the target. In the short SOA, RVF-left hemisphere primes resulted in high associate priming regardless of target location (ipsilateral or contralateral to the prime) whereas LVF-right hemisphere primes produced both high and low associate priming across both target location conditions. In the long SOA condition, contralateral priming patterns converged, demonstrating only high associate priming in both VF locations. The results of this study demonstrate the critical role of interhemispheric transfer in semantic processing and indicate a need to elaborate current models of semantic processing.

Neuroimaging the semantic system(s)

Background: The organisation of the semantic system is controversially discussed among cognitive scientists. Whereas some authors assume that semantic processing is amodal, others propose modality-specific semantic systems. The amodal account holds that there is a single semantic system independent of the modality of stimulus presentation and nature of the concept. The multi-modal account, on the other hand, postulates that semantic knowledge is represented in modality-specific, distinct semantic systems. Attempts to integrate both accounts resulted in a hybrid model of semantic processing. Although an amodal semantic network is assumed in this view, input-dependent semantic specialisation coexists. Aims: The purpose of this study was to examine contrasting theories of the representation of the semantic knowledge with functional magnetic resonance imaging (fMRI). Methods & Procedures: Nine healthy subjects performed a semantic judgement task on graphemic words and pictures in the experimental condition, as well as a visual identity judgement task on pseudoword and three-dimensional geometric objects in the control condition. Outcomes & Results: A common activation pattern for words and pictures was found in the left inferior frontal gyrus and in the left middle temporal gyrus. Word-specific effects occurred in the left superior parietal, the left inferior occipital, the left anterior cingulate gyrus, and the right inferior frontal gyrus as well as bilaterally in the anteriomedial temporal area. Picture-specific areas involved the left supramarginal gyrus and the left cingulate gyrus. Further activation was found in the right middle frontal gyrus and bilaterally in the fusiform and inferior occipital gyrus. Conclusions: The results can best be accommodated in the framework of an integrative, hybrid model of semantic representation combining features of the amodal and the modality-specific account.

The lateralized Stroop: A meta-analysis and its implications for models of semantic processing

The prevailing theory in the literature concerning the lateralization of Stroop effects involves a speed of processing account (see MacLeod, 1991). Because the left hemisphere (LH) demonstrates an overall advantage relative to the right hemisphere (RH) on most verbal tasks, interference effects are hypothesized to be greater in the LH than in the RH. Inspection of the data, however, reveals that many studies find no difference in magnitude of Stroop interference between the hemispheres. Given findings within the lateralized semantic priming literature, this is not surprising. A meta-analysis on a subset of lateralized Stroop experiments was conducted to determine whether or not the LH produces significantly more interference than the RH in this paradigm. Based on a number of exclusionary criteria, a total of 19 different studies were included, representing a potential 26 effect size estimates of differential interference. The effect size representing interference using congruent conditions as the baseline ( d+=.06) reveals that there is no significant difference between the hemispheres in magnitude of the Stroop interference effect. The LH interference effect was d+=.73, which is significant. Likewise, the RH interference effect, d+=.67, was significant. In summary, while there was no significant difference between the hemispheres, each hemisphere, when examined independently, did exhibit significant within hemisphere interference effects. These findings are presented in light of the lateralized semantic priming literature.

Language Lateralization: Development and Preliminary fMRI Testing of a new Semantic Processing Model

Language Lateralization: Development and Preliminary fMRI Testing of a new Semantic Processing Model

Calling a salad a federation: An investigation of semantic jargon. Part 1—nouns

This paper describes an individual with semantic jargon aphasia (RG). His output was largely composed from real words, but in very anomalous combinations. Preliminary investigations suggested severe word finding and comprehension problems for nouns, which were underpinned by a semantic deficit. Despite this, his speech often featured abstract terms. Subsequent tests showed that his comprehension and production were either unaffected by abstractness or aided by it. For example, when naming from definitions, he performed best when those definitions were abstractly phrased. It was hypothesized that RG's deficit lay in the visual domain of the semantic system. This was supported by a number of tasks showing that his knowledge about the appearance of objects and animals was disproportionately impaired. Naming was also influenced by the semantic characteristics of items, in that objects which are largely defined by function were named well, whereas items defined largely by their appearance were named poorly. RG's performance is interpreted against a distributed model of semantic processing.

Constraining claims about theories of semantic memory: More on unitary versus multiple semantics

Abstract In Caramazza, Hillis, Rapp, and Romani (1990) we proposed that the various patterns of impaired performance that have been used to motivate proposals of modality-specific semantics can, instead, be accommodated by an amodal unitary semantics account—the Organised Unitary Content Hypothesis (OUCH). Chertkow, Bub, and Caplan (1992), based on their study of the performance of seven patients with dementia of the Alzheimer's type, reported data that are supposedly problematic for OUCH. Here we respond by showing that Chertkow and colleagues have misinterpreted our proposal on a number of points and that, in fact, the “problematic” results can be easily accounted for within OUCH. Chertkow et al. made the further claim that their results favour an alternative model of semantic processing, according to which the processing of perceptual-functional semantic information is distinguished from the processing of associative semantic information. We argue here that until Chertkow et al. specify a number of cri...

The multiple semantics hypothesis: Multiple confusions?

Abstract In this paper we discuss the issue of multiple versus unitary semantics. We argue that the notion of multiple semantics (as currently articulated) does not, in fact, represent a theory of semantic organisation but is, instead, an arbitrary conjunction of a set of independent assumptions which are either unmotivated or, if motivated, equally compatible with a unitary semantics hypothesis. Furthermore, the empirical evidence that has been cited as support for this hypothesis is equally compatible with variants of the unitary semantics hypothesis. A model of semantic processing—the Organised Unitary Content Hypothesis (O.U.C.H.)—that is able to account for reported patterns of dissociation of performance is discussed briefly.

Extension of the relational database semantic processing model

A data model consists of three parts: (1) a data definition that represents the information in an understandable manner; (2) a definition of the constraints that must hold for the information to be valid; and (3) a definition of operations that can be performed on the information. Current database management systems do not allow explicit specification of all three parts of the data model. This paper gives an approach that extends current database management systems through a technique called pre-precompilation.

Evoked potentials and the dynamics of language processing

It is observed, in a brief review of studies of evoked potentials and language processing, that for the most part only global and rather static aspects have been investigated. It is argued that the important topic of the dynamics of word recognition has been neglected. The evolution of ‘spreading activation’ models of semantic processing is briefly traced. These postulate pre-conscious word recognition in a semantic network where radiation of excitation between semantically related ‘nodes’ or ‘logogens’ account for important context effects on word recognition. It is suggested that evoked potential studies could be used to test and elaborate this model. An experiment is described in which discrimination between positive and negative instances of primed categories by enhancement of the N1-P2 (N130-P212) amplitude measure supports the notion of early and possibly pre-conscious attainment of word meaning and also indicates enhanced activation from word detection units sensitized by spread of activation from a category prime.