Sparing of number words in oral production

• ¹ IRCSS San Camillo Hospital, Italy
• ² University of Padua, Italy

Phonological errors seldom affect number words (Bencini et al., 2011). Current views suggest that this effect arises at the post-lexical (buffer) level: the basic phonological building blocks of ordinary words are phonemes, but the building blocks of number words are whole words. Dotan and Friedman (2010) suggested that the same effect holds for function words and morphological affixes. Case report. P.M. is a 77-year-old, Italian speaking, right-handed man, with 13 years of education. He had a left hemisphere stroke which resulted in conduction aphasia. Speech production was well articulated but affected by severe phonological disorders. An MRI showed a lesion in the left temporo-parietal areas. Methods. PM was administered: Repetition and Reading aloud of function word categories, cardinal and ordinal numbers in word lists and in sentences. Function words included the most frequent and shortest words in the Italian lexicon. Reading and repetition of number words within culturally transmitted expressions (e.g., “Four wheel drive”). Number words production in response to simple arithmetical operations. Questions requiring a numerical response (e.g., “days in a year?”) Inflected verbs. Reading and repetition of words with an embedded number word. Results. Results for reading aloud and repetition of word categories are shown in Table 1. Cardinal numbers elicited about one fifth of all phonological errors made on function word categories, even short and extremely frequent ones. No phonological errors were detected when PM had to produce the result of arithmetical operations. Cardinal numbers were frequently substituted by other cardinal numbers. Number words in transmitted expressions contained phonological errors in both reading aloud tasks although less than in the other word categories. Phonological errors were counted separately for the root and the bound morpheme on inflected verbs: there was no difference. Words with an embedded number word elicited an equal proportion of phonological errors in the number portion and in the non-number portion. In sentences only number words were spared; free standing function words and bound morphemes were as affected as other word categories. Discussion. These findings seem to set cardinal number words apart in the phonological output buffer from other possible building blocks of preassembled phonological units (like function words and bound morphemes). Building blocks constituted by numbers are more cohesive than the blocks constituted by function words and bound morphemes. Bencini et al. (2011) argued that numbers are recursive and consist of basic lexical units which are then combined following syntactic rules. This property would make number words resistant to damage in the phonological buffer.