Abstract
Studies of agrammatism usually rely on selecting ‘agrammatic’ speakers by using a pre-determined definition. The grammatical disruptions demonstrated by these selected subjects are then examined and accepted as the basis for identifying further agrammatic subjects. Such an approach is not only tautologous but also entails a certain degree of subjectivity in identifying agrammatism in the first place. Subjects who do not fit in well with the pre-selected definition would be excluded from the studies and the general picture of agrammatism then becomes incomplete. The present study reports the grammatical ability of 30 Cantonese-speaking aphasic subjects using a cluster analysis. As the subjects were not selected on the basis of their ‘grammatical abilities’, the results can be interpreted as more representative of the possible pattern of grammatical disruption in Cantonese. The results have shown that there are at least two patterns of agrammatism, which differ in the extent and magnitude of grammatical disruptions. One pattern demonstrated disruption of grammatical structures across the board whilst the second pattern showed the vulnerability of prepositional co verbs and utterance-final particles. There was also an atypical case who had no difficulty with closed class words in general but the production of aspect markers was disrupted.