Oral reading of words and sentences: Investigating the source of context effects

Background: Aphasic speakers' oral reading of isolated words has been widely studied, yet little is known about how contextual information influences reading ability when words are placed in sentences. Embedding words in sentences or texts has been argued to provide syntactic and/or semantic constraints on oral word reading that are unavailable in word lists, but no details are available about how such constraints might operate. The few studies that have actually compared sentence vs list reading have yielded conflicting results, and fail to show any consistent reading pattern that would predict when and why a reading advantage would arise from sentence context. Aims: We attempted to identify the pattern of aphasia/alexia that is associated with a sentence (vs list) reading advantage, and to assess the contribution of language and reading functions that might influence this context effect. Methods & Procedures: Five fluent, aphasic patients read the same 100 words (nouns, verbs, adjectives, and function words) in lists and in sentences. Patients were selected to have relatively intact sentence production ability, which has been argued to be an important element when sentence reading is superior to list reading. Outcomes & Results: Only two of the five patients showed a significant advantage for reading words in sentences, and additional analyses were carried out in an attempt to distinguish the reading patterns of the patients who showed the effect from the patterns of those who did not. The context effect exhibited by the two patients did not appear to be related to semantic influences, or to a cumulative build-up of contextual cues across words in sentences. Rather, the one finding that distinguished the two patients with significantly better reading of words in sentences compared to lists from the other patients was improved sensitivity in sentences to target words' grammatical class. Conclusions: The results indicate that good sentence production abilities are not a strong predictor that sentence reading will be enhanced relative to list reading. We propose that the sentence context effects in the two patients arose from their ability to extract information from sentences about targets' grammatical class, an ability that was not apparently shared by the other patients. Knowledge of grammatical class was argued to improve patients' reading by combining with other influences (e.g., orthographic structure, imageability) to substantially reduce the number of candidate word responses. Further specification of the source of these effects will help to identify patients who are most likely to benefit from sentence- and text-level reading treatments.