Abstract
Background
The importance of syllables for speech production is recognized, but what kind of units syllables are is controversial. Some argue that they are content units or chunks, with each syllable having its own independent representations (e.g., Ferrand et al., 1997; Levelt et al., 1999), others that they are only structural units (e.g., Romani et al., 2011). Effects of syllable frequency have been used to argue for the first position, but results have been inconsistent because of the difficulty of separating syllable and phoneme frequency (e.g., Cholin et al., 2006; Laganaro, 2008).
Method
We looked at phoneme and syllable frequency in a corpus of errors made by aphasic patients in word repetition, word reading and picture naming. The patients were subdivided into an apraxic (N=11) and a phonological group (N=10), based on a number of characteristics including high rates of phonetic errors and phonological simplifications in the apraxic group, but not in the phonological group (e.g., Galluzzi et al., 2015).
We considered: 1. average phoneme and syllable frequency of target words and corresponding errors; 2. number of errors which increased vs. decreased frequency; 3. frequency of real errors compared to frequency of pseudo-errors obtained by recombining consonant phonemes produced in errors --across the whole group of patients-- with new targets.
Results
The apraxic group showed positive effects of phoneme frequency. Errors had a higher frequency than their targets (see Table 1) and more errors increased than decreased frequency (1353 vs. 954; 2=35.2; p<.001). The real errors had significant higher phoneme frequency than the pseudo-errors. Effects of syllable frequency, instead, were absent or in the wrong direction (N errors increasing vs. decreasing frequency 1062 vs. 1248; 2=7.5; p<.001).
The phonological group showed positive effects of phoneme frequency, but paradoxical effects of syllable frequency. When we only considered single consonant substitutions, errors had a significant lower syllable frequency in repetition and reading (see Table 1). When all types of errors were considered, this effect was significant across all tasks. Fewer errors increased rather than decreased syllable frequency (across all tasks: 410 vs. 570; 2=13.2; p<.001). Real errors had significant lower syllable frequency than the pseudo-errors.
Across groups, pseudo-errors showed evidence of decreasing syllable frequency compared to targets.
Conclusions
Positive effects of phoneme frequency in apraxic patients are expected. Frequent phonemes will be more practised and easier to articulate. In contrast, there were no or paradoxical effects of syllable frequency, especially in the phonological group (see also Stemberger, 1991). This likely due to two facts: 1. the phonological patients had no preference to produce phonemes of higher frequency; 2.when phonemes are semi-randomly substituted, the outcome is likely to be of lower frequency because real words will use relatively frequent syllables. These results do not support the hypothesis that syllable units are stored and strung together in production. If this were the case, syllables of higher frequency should be produced in error more often than syllables of lower frequency, which we did not find.
The importance of syllables for speech production is recognized, but what kind of units syllables are is controversial. Some argue that they are content units or chunks, with each syllable having its own independent representations (e.g., Ferrand et al., 1997; Levelt et al., 1999), others that they are only structural units (e.g., Romani et al., 2011). Effects of syllable frequency have been used to argue for the first position, but results have been inconsistent because of the difficulty of separating syllable and phoneme frequency (e.g., Cholin et al., 2006; Laganaro, 2008).
Method
We looked at phoneme and syllable frequency in a corpus of errors made by aphasic patients in word repetition, word reading and picture naming. The patients were subdivided into an apraxic (N=11) and a phonological group (N=10), based on a number of characteristics including high rates of phonetic errors and phonological simplifications in the apraxic group, but not in the phonological group (e.g., Galluzzi et al., 2015).
We considered: 1. average phoneme and syllable frequency of target words and corresponding errors; 2. number of errors which increased vs. decreased frequency; 3. frequency of real errors compared to frequency of pseudo-errors obtained by recombining consonant phonemes produced in errors --across the whole group of patients-- with new targets.
Results
The apraxic group showed positive effects of phoneme frequency. Errors had a higher frequency than their targets (see Table 1) and more errors increased than decreased frequency (1353 vs. 954; 2=35.2; p<.001). The real errors had significant higher phoneme frequency than the pseudo-errors. Effects of syllable frequency, instead, were absent or in the wrong direction (N errors increasing vs. decreasing frequency 1062 vs. 1248; 2=7.5; p<.001).
The phonological group showed positive effects of phoneme frequency, but paradoxical effects of syllable frequency. When we only considered single consonant substitutions, errors had a significant lower syllable frequency in repetition and reading (see Table 1). When all types of errors were considered, this effect was significant across all tasks. Fewer errors increased rather than decreased syllable frequency (across all tasks: 410 vs. 570; 2=13.2; p<.001). Real errors had significant lower syllable frequency than the pseudo-errors.
Across groups, pseudo-errors showed evidence of decreasing syllable frequency compared to targets.
Conclusions
Positive effects of phoneme frequency in apraxic patients are expected. Frequent phonemes will be more practised and easier to articulate. In contrast, there were no or paradoxical effects of syllable frequency, especially in the phonological group (see also Stemberger, 1991). This likely due to two facts: 1. the phonological patients had no preference to produce phonemes of higher frequency; 2.when phonemes are semi-randomly substituted, the outcome is likely to be of lower frequency because real words will use relatively frequent syllables. These results do not support the hypothesis that syllable units are stored and strung together in production. If this were the case, syllables of higher frequency should be produced in error more often than syllables of lower frequency, which we did not find.
| Original language | English |
|---|---|
| Journal | Frontiers in Psychology |
| DOIs | |
| Publication status | Published - 15 Aug 2016 |
| Event | 54th Annual Academy of Aphasia Meeting - Llandudno, United Kingdom Duration: 16 Oct 2016 → 18 Oct 2016 |