The Influence of Aphasia Type and Severity on Sentence Comprehension after Left Hemisphere Stroke.

The subgroups were small. The severe aphasia group had only 8 people after the severe and very severe categories were merged, and the type analysis rested on 24 people with anomic aphasia, 18 with Broca's, 10 with latent aphasia and just 6 with conduction aphasia. The 3 people with Wernicke's aphasia and 3 with transcortical motor aphasia were left out entirely. The type and severity analyses drew on largely the same people, so the anomic group and the mild group were mostly the same participants. The two sets of comparisons are not independent of each other. Everybody found the hardest sentences, the passive object-relatives, hard. Accuracy sat near chance in every group including the older controls, so that condition tells you little about aphasia specifically. Participants came from two studies and completed different numbers of trials, though the task itself was identical and performance did not differ between the two platforms. Only sentence comprehension was measured. There were no independent tests of attention or working memory, so the authors' explanation that limited processing resources caused the difficulties is untested. Everyone was at least six months past their stroke and the sample was mostly mild to moderate, so the results say nothing about the early weeks after a stroke or about people with the most severe aphasia. Nobody received treatment. This is a cross-sectional comparison, so it cannot show that any therapy improves sentence comprehension.

The work was funded by a 2023 ASHFoundation New Investigators Research Grant and a National Institutes of Health grant (R21DC021481), both awarded to the lead author. The authors declare no conflicts of interest.

This was not a treatment study, so nothing here shows that any particular therapy improves sentence comprehension. And do not assume that a client with latent aphasia or a mild anomic profile has intact comprehension because they pass a standard language test: in this study both groups were still less accurate than controls on the object-relative sentences, even though the latent group matched controls on the simpler ones.