Case study: Where do aphasic perseverations come from?

Tess Ackerman, Andrew W. Ellis

Research output: Contribution to journalArticlepeer-review


Background: Perseverations are common in the speech of people with aphasia but the phenomenon has been the subject of relatively little investigation. We had the opportunity to study an aphasic patient who produced a large number of perseverations in naming, reading, and repetition tasks.

Aims: To gain a better understanding of the origins and causes of perseverative errors in a man with aphasia ( MM) through a detailed analysis of his errors across naming, reading, and repetition, combined with a thorough assessment of his language disorder.

Methods & Procedures: MM was given a cognitive neuropsychological assessment of his language-processing system. He was then asked to name 140 black and white drawings of objects and to repeat and read aloud the same 140 object names. Each of these tasks was done on two occasions.

Outcomes & Results: MM showed major semantic impairment combined with some phonological and orthographic deficits. He was better at reading object names than repeating them, and worst at object naming. Analysis of his errors showed differences between naming, repeating, and reading. Whole-word perseverations were most common in object naming, where they were predominantly unrelated to the target items.

Conclusions: MM's aphasia and perseverations are discussed in terms of the theory put forward by Martin and Dell ( 2007, this issue). We endorse the proposition that perseverations should be understood within the wider context of aphasic breakdown, and that no special mechanisms may be required to explain perseverative errors. But the lack of any influence of lexical responses like word frequency on MM's performance, and the fact that perseverations mostly resulted in unrelated errors, are problematic for the Martin and Dell framework. They suggest that MM's perseverations mostly occurred when his semantic and phonological system were deprived of any useable input, with the target being unable to influence the response. Under such circumstances, either no response was available, or MM made an unconstrained and therefore unrelated response, in which case the endogenously driven reactivation of recent responses ( whole words or fragments) was likely to fill the void with a perseverative error.

Original languageEnglish
Pages (from-to)1018-1038
Number of pages21
Issue number10-11
Publication statusPublished - 2007


  • SET

Cite this