Practical Applications

The constraints we have examined in relation to the sequencing of sounds in syllables via sonority have far-reaching implications in many applied situations, such as in normal phonological development, in clinical populations, and in foreign language learning. For example, the process of consonant cluster reduction, which is commonly observed in all three populations mentioned above, is far from being haphazard. A target such as play [ple] is much more likely to be reduced to [pe] than to the alternative [le]. The reason for this is that the former is the more unmarked (more expected) one because it provides a higher jump in sonority from the single onset to the nucleus (in [pe] the sonority index of /p/ = 1, the sonority index of /e/ = 9; thus the resulting sequence is a change from 1 to 9; [le], on the other hand, would result in a change from 6 to 9 in sonority indices). Since a CV sequence is more natural when the contrast between the C and the V is greater, [pe] is the more valuable of the two logical alternatives. Support for the validity of such a constraint is not hard to find in developing phonologies. Chin (1996) observed several children with phonological disorders whose modification of target onset clusters was governed by the principle reviewed above. In one instance, the subject applied consonant cluster reduction to all targets with #sC clusters (e.g. stove [sov], snow [so], etc.) while not reducing other target clusters (e.g. play [ple], brush [bwΛs], etc.). Looking at several examples, Chin concluded that the child applied the reduction process to those targets where the sonority difference between C₁ and C₂ was less than 3 (i.e. stove 3 to 1 =−2, snow 3 to 5 = 2). Targets that had a difference of 3 in sonority from C₁ to C₂ (play 1 to 6 = 5, sleep 3 to 6 = 3, brush 2 to 7 = 5) were not subject to reduction.

Another subject revealed the following patterns:

(a) stop + approximant → stop (twin [dɪn], drum [dΛm], play [pe])

(b) fricative + sonorant → fricative (few [fu], swim [sɪm], shrub [ʃΛb])

(c) fricative + stop → stop (spoon [bun], stove [dov], sky [daɪ])

The child’s modification patterns do not allow us to state whether C₁ or C₂ of the cluster is deleted, as C₁ is deleted in (a) and (b), but C₂ is the one that is deleted in (c). If, however, we analyze the results in terms of sonority rises, we see that the behavior of the subject is very regular in that he follows the path resulting in the greatest jump in sonority from the resulting C to the nucleus V. Thus, while the observed simplification of the target drum is [dΛm], with a movement from 2 to 10 in sonority, the alternative, [ɹ̣Λm], would have resulted in 7 to 10, which is a much smaller jump. Similarly, in sky becoming [daɪ], we have a jump from 2 to 10, while the alternative, [saɪ], would have given a smaller (3 to 10) jump.

Sonority-driven modification of onset clusters has also been observed in aphasic patients. Blumstein (1978), invoking the concept of sonority for the erroneous cluster formation in paraphasias, notes that the addition of, for example, a liquid in an erroneous production goes to the left of a vowel, thus forming the sequence ‘obstruent + liquid + vowel’, which is in accordance with the sonority sequencing principle. Beland et al. (1990) noted that their patient’s deletion of one member of the sequence ‘obstruent + liquid + V’ was always the liquid, resulting in the maximum jump to ‘obstruent + V’. Christman (1992) also found sonority to be influential in the syllable production patterns of jargon aphasics. More recently, Romani and Calabrese (1998) provided further support for sonority-driven patterns by showing their 40-year-old Italian aphasic patient’s modification of #CCV sequences. By producing ‘obstruent + liquid + V’ targets as ‘obstruent + V’, ‘obstruent + /j/ + V’ targets as ‘obstruent + V’, and ‘nasal + /j/ + V’ targets as ‘nasal + V’, the patient deleted the segment of higher sonority in the target cluster and produced a sequence with a maximum rise in sonority from the onset to the nucleus. Telugu-speaking children with prelingual hearing loss have been reported to have greater difficulty in CV syllables in which the sonority jump is small (e.g. ‘glide + V’) than those in which the sonority jump is bigger (e.g. ‘obstruent + V’) (Duggirala Vasanta, personal communication).

children ages 1;9–3;2. The prediction was that when the CCVC targets were reduced to CVC sequences, the deleted member of the double onset would be the one that had a higher sonority index, because its deletion could provide the remaining sequence of the single onset to the nucleus with the maximal rise in sonority. The overall results confirm the predictions in that children preserved the least sonorous consonant member of the onset cluster and created the maximal rise in sonority. Yavas and Someillan (2005) tested the same hypothesis with a group of Spanish–English bilingual children with target #sC sequences. It was hypothesized that among the possible sequences of English #sC clusters (‘/s/ + stop’ sky, ‘/s/ + nasal’ snail, ‘/s + l/’ sleep, ‘/s + w/’ swim), subjects’ success rates would be higher for the targets in which the sonority jump from C₁ to C₂ was higher. The results confirmed the hypothesis over-whelmingly, at least when C₂ was a continuant, in that the easiest target was /sw/ (sonority jump from 3 to 8), followed by /sl/ (sonority jump from 3 to 6). The difference between the clusters in which the C₂ was a non-continuant (i.e. ‘/s/ + stop’ and ‘/s/ + nasal’) did not reveal any significance. Further supporting evidence is provided by Yavas and Barlow (2006), and Yavas (2010).

The ease or difficulty of acquisition of onset clusters invoking sonority indices has also been shown in foreign language phonology. Broselow and Finer (1991) examined the data from 24 native speakers of Korean and 8 native speakers of Japanese with respect to their productions of the target English /pɹ̣, bɹ̣, fɹ̣/, /pj, bj, fj/. The hypothesis was that clusters with a greater sonority jump from C₁ to C₂ (e.g. /pj/ from 1 to 8) would be less problematic than clusters with a smaller sonority difference (e.g. /fɹ̣/ from 3 to 7). In general, the error rates of the subjects supported the predictions.

If maximum rise in sonority is the most unmarked (expected) sequencing from the onset to the nucleus, minimum descent in sonority is the most unmarked (expected) sequencing for a movement from the nucleus to the coda. The reason for this is that the most common (natural/unmarked) syllable type is codaless (CV), where there is no descent in sonority. Thus, when we have a coda, the smaller the descent from the nucleus the more valuable it is. That this principle is at work can be seen in developing phonologies. Ohala (1999) examined the coda cluster modifications of children ages 1;9–3;2 in which CVCC targets were modified to CVC. She hypothesized that the member of the coda cluster to be deleted would be the one that was lower in sonority, so that the remaining higher-sonority item would provide the minimum descent from the nucleus. Thus, for example, a sequence such as [maɹ̣p] was expected to reduce to [maɹ̣] (sonority shift of 10 to 7 from the nucleus to the coda), and not as [map] (from 10 to 1). The results were supportive of the hypothesis in that the expected reductions were made more than 50 percent of the time, and the unexpected modifications totaled only around 16 percent.

Evidence for behavior governed by the principles of sonority also comes from second language phonology data. Hansen (2001) examined the acquisition of English codas by Mandarin speakers. In dealing with three-member target English codas, the learners had the greatest difficulty with clusters that violate the requirement of gradual lowering in sonority from the nucleus to the members of the cluster. Triple codas such as ‘liquid + stop + fricative’ (e.g. /lps/ alps, /ɹ̣dz/ words) and ‘nasal + stop + fricative’ (e.g. /nts/ prints, /ndz/ bands) have sonority descending from V to C₁ to C₂ , but rising from C₂ to C₃ . Learners modified such targets much more frequently than the unmarked codas that follow gradual sonority fall (e.g. /ɹ̣st/ first, /ɹ̣ld/ world). Also significant were the resulting two-member codas when subjects reduced the target CCC# sequence by deleting one of the consonants; all resulting two-member codas obeyed the demands of the Sonority Sequencing Principle in that C₁ was higher in sonority than C₂.

مواضيع ذات صلة

Syllable Weight and Ambisyllabicity

Written Syllabification

Dialectal variations

Triple codas

Double codas

Codas

Triple onsets

Double onsets

Single onsets

English Syllable Phonotactics

Syllabification

Sonority