Living musical instruments and inanimate body parts?

In the literature about category effects in semantic memory, body parts and musical instruments are often considered atypical, because in cases with a disproportionate impairment of living categories body parts are relatively spared, while musical instruments are often severely defective. In this study the performance of 57 subjects affected by diseases generally associated with lexical-semantic impairment, for the most part Alzheimer's disease and other forms of cortical degeneration, but also herpetic encephalitis and traumatic brain damage are analyzed. The subjects were given a picture naming task tapping eight categories: three living categories (animals, fruits and vegetables) and three non-living categories (tools, furniture and vehicles), plus body parts and musical instruments. On a preliminary analysis at the group level, body parts were the least impaired category and musical instruments the most severely impaired, the six living and non-living categories being intermediate. However, these differences disappeared after covariance for lexical frequency, name agreement and age of acquisition. The relationship between living categories, non-living categories, musical instruments and body parts was investigated by means of a Lisrel model of Confirmatory Factor Analysis. Two latent variables related to living and non-living categories respectively were defined, and it was found that both body parts and musical instruments were significantly related only with non-living categories. The results showed that the definition of the latent variable expressing the substrate of non-living categories was less satisfactory than that expressing the living categories. On this basis, the conclusions of this study appear statistically definite but their psychological interpretation is less straightforward.