Making sense out of jargon: a neurolinguistic and computational account of jargon aphasia.

Abstract

To identify the cognitive and neuroanatomic bases of neologistic jargon aphasia with spared comprehension and production of written words. Detailed analysis of performance across experiments of naming, reading, writing, repetition, and word/picture matching by a 68-year-old woman (J.B.N.) served to identify which cognitive mechanisms underlying naming and word comprehension were impaired. J.B.N.'s impairments were then simulated by selectively "lesioning" a computer model of word production that has semantic, word form, and subword phonologic levels of representation (described by Dell in 1986). In comprehension experiments, J.B.N. made far more errors with spoken word input than with written word or picture input (chi-square = 40-59; df = 1; p < 0.0001) despite intact auditory discrimination. In naming experiments (with picture, definition, or tactile input), J.B.N. made far more errors in spoken output relative to written output (chi-square = 14-56; df = 1; p < 0.0001). These selective impairments of spoken word processing were simulated by reducing connection strength between word-level and subword-level phonologic units but maintaining full connection strength between word-level and semantic units in Dell's model. The simulation yielded a distribution of error types that was nearly identical to that of J.B.N., and her CT and MRI scans showed a small subarachnoid hemorrhage in the left sylvian fissure without infarct. Cerebral angiogram showed focal vasospasm in sylvian branches of the left middle cerebral artery. Focal left perisylvian dysfunction can result in a highly selective "disconnection" between word-level and subword-level phonologic representations manifest as neologistic jargon aphasia with intact understanding and production of written words.