| Abstract | There is accumulating evidence that speakers recruit inhibitory control to manage the conflicting demands of online language production, e.g., when selecting from among co-activated representations during object naming or when suppressing alternative competing terms in referential language use. However, little is known about the types of conflict resolution mechanisms underlying the production processes. The aim of this research was to assess the relative contribution of various forms of interference arising at different stages of information processing as well as their control to single- and multi-word utterance production.
The systematic review of picture-word interference (PWI) studies (Study 1) was conducted to trace the origins of semantic context effects in order to address the question of whether spoken word production can be seen as a competitive process. The various manipulations of PWI task parameters in the reviewed studies produced a mixture of findings that were either contradictory, unable to discriminate between the rival theories of lexical access, or of questionable validity. Critically, manipulations of distractor format and of whole-part relations with varied association strength produced sufficiently strong evidence to discount post-lexical non-competitive accounts as the dominant explanations for observed interference effects, constraining their locus to early rather than late processing stages. The viability of competitive hypotheses was upheld; however, this is contingent on the relative contribution of pre-lexical processes, which remains to be confirmed by future research.
The relative contribution of different conflict resolution mechanisms (measured by the anti-saccade, arrow flanker and Simon arrow tasks) to object naming under prepotent (the PWI task) and underdetermined competition (picture naming task with name agreement, NA, manipulation) was further investigated in Study 2, while Study 3 extended the notion of separability of the inhibitory processes to grammatical encoding (grammatical voice construction and number agreement computation). In Study 2, only the flanker effect was a significant predictor of the PWI but not NA effect, while the remaining inhibitory measures made no significant contribution to either the PWI or NA effect. Participants with smaller flanker effects, indicative of better resolution of representational conflict, were faster to name objects in the face of competing stimuli. In Study 3, only utterance repairs were reliably predicted by the flanker and anti-saccade effects. Those who resolved representational conflict or inhibited incorrect eye saccades more efficiently were found to self-correct less often during online passive voice construction than those with poorer resolution of inhibition at the representational and motor output level. No association was found between the various inhibitory measures and subject-verb agreement computation.
The negative priming study with novel associations (Study 4) was an attempt at establishing the causal link between inhibition and object naming, and specifically whether inhibition that is ostensibly applied to irrelevant representations spreads to its associatively related nodes. Response times to the associated probe targets that served as distractors in previous prime trials were no different than response times to non-associated probe targets. Possible explanations are discussed for the lack of the associative negative priming effect.
The studies described here implicate two types of interference resolution abilities as potential sources of variability in online production skills, with the underlying assumption that better resolution of conflict at the representational and motor output level translates to faster naming and more fluent speech. There is insufficient evidence to determine whether the representational conflict is lexical or conceptual in nature, or indeed whether it is inhibitory in the strict sense. It also remains to be established whether interference that likely ensues at the response output stage is due to some criterion checking process (self-monitoring), recruitment of an inhibitory mechanism (response blocking) or both. |
|---|