Origins of Nonword Phonological Errors in Aphasic Picture Naming

Abstract

A recent theory of lexical access in picture naming maintains that all nonword errors are generated during the retrieval of phonemic segments from the lexicon (Dell, Schwartz, Martin, Saffran, & Gagnon, 1997b). This theory is challenged by "dual origin" theories that postulate a second, post-lexical mechanism, whose disruption gives rise to "phonemic paraphasias" bearing close resemblance to the target. We tested the dual origin theory in a corpus of 457 nonword errors drawn from 18 subjects with fluent aphasia. The corpus was divided into two parts, based on degree of phonological overlap between error and target, and these parts were separately examined for proposed diagnostic characteristics of the postlexical error mechanism: serial order effects across the word, sensitivity to target length, and insensitivity to target frequency. Results did not support the dual origin theory but were consistent with a single, lexical origin account in which segment retrieval operates from left to right, rather than in parallel. Findings from this study also shed new light on how individual differences in the severity of the retrieval deficit modulate the expression of phonological errors in relation to target characteristics.

Origin of phoneme substitution and phoneme movement errors in aphasia

“…Indeed, positive as well as negative results have been reported with regard to lexical frequency effects on phonological errors (see Pate, Saffran and Martin, 1987, Gordon, 2002, compared to Best, 1996, Schwartz et al, 2004, for opposite results). As argued in the introduction, following the predictions of multiple origins of phonological errors, absence of an effect of lexical factors such as frequency rather favours a post-lexical locus of impairment in both patients.…”

“…On the one hand, studies in the framework of the double or multiple origin have attempted to demonstrate that distinct factors affect different kinds of errors, for instance neologisms and milder phonological errors, or errors produced by fluent and dysfluent aphasic patients (Valdois, 1990;Smith, 1994, 1995;Kohn, Melvold and Shipper, 1998;Wilshire and McCarthy, 1996;Romani and Calabrese, 1998;Romani and Galluzzi, 2005). On the other hand, studies in the framework of interactive activation models have sought to show that all kinds of errors are affected by the same lexical and phonological variables (Wilshire, 2002;Schwartz, Wilshire, Gagnon, and Polansky, 2004;Olson, Romani and Halloran, 2007).…”

“…Against multiple origins of phonological errors, a single account for all kinds of errors has been proposed in the framework of connectionist models (Dell, Schwartz, Martin, Saffran and Gagnon, 1997;Foygel and Dell, 2000;Schwartz, Wilshire, Gagnon and Polansky, 2004). In these models, all phonological errors are generated by mis-selection of phonemes.…”

Contextual phonological errors and omission of obligatory liaison as a window into a reduced span of phonological encoding

“…Different proposals provide an explanation for non-contextual segmental errors. In the context of connectionist models of language production it has been suggested that phoneme mis-selection is due to noise in the connection between lexical and phonological nodes or throughout the lexical-semantic and lexical-phonological connections (Dell et al, 1997;Foygel & Dell, 2000;Schwartz, Wilshire, Gagnon and Polansky, 2004). Alternatively, segmental errors are thought to arise because of a default mechanism attributing segments when phonological information is missing or cannot be completely retrieved (Butterworth, 1992;Kohn & Smith, 1995).…”

Analysis and interpretation of serial position data

“…This type of analysis has been widely applied in studies of errors made by aphasic patients in spelling and in speech (e.g. Buchwald & Rapp, 2006;Schwartz, Wilshire, Gagnon, & Polansky, 2004;Caramazza, Papagno, & Ruml, 2000;Ward & Romani, 1998b;Gagnon & Schwartz, 1997;Kay & Hanley, 1991;Neils, Roeltgen, & Greer, 1995; Serial position data 4 Papagno & Girelli, 2005;Cipolotti, Bird, Glasspool, & Shallice, 2004;Buchwald & Rapp, 2004;Cotelli, Abutalebi, Zorzi, & Cappa, 2003;Croisile & Hibert, 1998). Recently, Mactynger and Shallice (2009) showed that there are some systematic distortions of the serial position curve that the Wing and Baddeley method can introduce (see also accompanying response Wing & Baddeley, 2009), and they suggested an alternative method that we have also used and in a study of spelling errors made by deaf participants, and speech errors made by aphasic patients (Olson, 1995;Olson & Caramazza, 1999).…”