[Disturbances of word finding of compound nouns (nomina composita) in aphasics (author's transl)]

The production of nominal compounds in aphasia

This study reports the difficulties of aphasic patients in producing nominal compounds in German. In a picture naming task with semantically relatively transparent targets it was found that the word frequency of the components determined the accuracy of the patients' performance. A second picture naming task using relatively opaque targets, and a further naming task in which patients were confronted with aurally presented object paraphrases using very opaque targets, are reported. The error patterns of all three tasks give evidence for morpheme-based mis-productions (e.g., componential substitutions), however, in declining proportions with decreasing semantic transparency. Access to final (or basic) components was clearly superior to initial (or determinative) components suggesting position-specific access routines.

Finnish compound structure: experiments with a morphologically impaired patient

The present study addresses two major aspects of compound processing, namely the issues of access code and structural effects (headedness), with an extensively studied deep dyslexic patient, HH, who has previously been shown to suffer from morphological impairment. In oral reading of compounds, HH preserves their morphological structure and, at the same time, shows additional processing load in increased errors compared to monomorphemic and derived words. Our data suggest that semantically transparent Finnish compounds are decomposed into their constituents at the level of lexical access. Furthermore, there is a processing asymmetry between the constituents, since several qualitative features of HH's reading performance suggest that it is the first constituent which primarily serves as the initial access code.

The processing of compound words: a study in aphasia

The naming difficulties of an aphasic patient (MB) are almost exclusively confined to compound words. The paraphasias in response to compounds always reflect the compound structure of the targets. Frequent semantically adequate substitution errors suggest the separate processing of single components at the lemma level. MB's specific difficulties probably arise when two lemmas are defined by one lexical entry, as we suggest for compounds. The compound structure, as well as the positions of the single lemmas within the compound, would be specified before lemmas are accessed. The lack of a positional effect in substitution errors suggests a parallel activation of the two components.