Abstract
Background: Disordered discourse in cases of senile dementia of Alzheimer's type (DAT) has mainly been described in conversation and picture description tasks. The referential communication task provides researchers and clinicians with new insights on the nature of these disorders.
Aims: To study to what extent persons suffering from DAT can benefit from shared experience through trial repetition to achieve common reference.
Methods & Procedures: Thirteen persons suffering from DAT at minimal or mild stage (MMSE score range = 18–27) were compared with 13 healthy elderly adults (64–86 years) in neuropsychological assessment of executive functions and in a referential communication paradigm. To study how the two partners achieve mutual understanding by progressively elaborating a common ground, the task was repeated three times.
Outcomes & Results: Persons with DAT produced a larger number of words than control participants and they benefited from the task repetition. However, they were less able to take into account previously shared information, used no definite referential expressions and were more idiosyncratic in their descriptions of the referent. This decline of communicative effectiveness was found not to relate closely to executive deficits.
Conclusions: Collaborative exchanges allow healthy elderly persons to ground reference in common experience. This process is severely disturbed in persons with DAT, in relation to poor memory of preceding episodes or to other cognitive impairments.
Notes
1. Means and standard deviations used for the transformations were estimated from the performance of 45 healthy adults aged 48–93 years (Berrewaerts Citation2002).
2. That correction was brought because we found an R2 = 0.99 in one person with DAT showing a systematic increase of word number from trial to trial. The corrected value was close to 1 in six persons with DAT (range 0.83–1.00) and close to zero (range –0.01 to –0.13) in five other persons. There was one intermediate value (0.60). Correlation coefficients were computed separately by groups because due to group differences in Mattis and composite executive Z‐scores, the repetition effect correlated moderately (r = 0.40 and 0.43, respectively, both p<0.05) with these scores in the whole sample (n = 26).