acquisition of the voiceless alveolar lateral fricative, which showed a wide range of ... (see Table 1). Welsh lacks voiced fricatives at the alveolar, and postalveolar ..... of the clicks and ejectives (as well as palatal plosives) appear among the.
The acquisition of the lateral fricative in Welsh-English bilinguals MARTIN J . BALL, NICOLE MÜLLER, AND SIÂN MUNRO
Abstract This article reports on a study into the acquisition of Welsh phonology in bilingual Welsh-English children. It concentrates on the acquisition of the voiceless alveolar lateral fricative, which showed a wide range of substitutions, and differential patterns of development dependent on the dominant language of the subjects, and the position within the word of the target. Various substitution strategies are described, including fricatives, and fricatives plus lateral. It is noted that in these two categories, velar fricatives become increasingly common in later age ranges, especially with the English dominant speakers. It is argued that this is because, unlike front fricatives, the velar fricative shares several important acoustic characertistics with the lateral fricative: namely, a formant-like structure and the same top limit to the noise spectra.
1. Introduction In a study of the acquisition of Welsh and English phonology in bilingual children of both Welsh and English dominant backgrounds (Ball, Munro, Miiller, and Duckworth 1998; Munro, Ball, Muller, and Duckworth, in preparation), developmental patterns for the two languages were described. Of particular interest were the trends evident for the two groups of children in respect of the lateral fricative of Welsh. Not only were there noticeable differences in patterns of acquisition between the Welsh-dominant and the English-dominant children, but this sound showed a great amount of variation, with 20 different variants attested in the data. In this account, these results will be examined to see whether this variation is amenable to analysis in terms of both language dominance and developmental stage, and whether a phonetic account can explain why certain variant forms preMultilingua 20-3 (2001), 269-284
061-8507/01/0020-0269 © Walter de Gruyter, Berlin
270
M. J. Ball, N. Müller,
The acquisition of the lateral fricative in Welsh-English
and S. Munro
dominate in the data. In order to do this, we need first to provide some background to the phonology of Welsh. Welsh is a Celtic language spoken by about half a million speakers throughout Wales, nearly all of whom are bilingual with English. The phonological system of Welsh consonants (described in Ball and Jones 1984) differs from that of English in the fricative sub-system in particular (see Table 1). Welsh lacks voiced fricatives at the alveolar, and postalveolar places of articulation, but unlike English has a voiceless fricative at the velar place. Welsh also has a voiceless alveolar lateral fricative, which can be found at all places within the word (though most .commonly word-initial). Examples include lliw [*iu], 'color'; llawn [iaun], 'full'; gallu [gat], 'to be able'; callach [katex], 'wiser'; pell [pet], 'far'; Hall [la*], 'other'. The vowel system also differs from that of English (and there are major regional differences with vowels), but this is not discussed here.
bilabial nasal plosive
labiodental
Table 1. Welsh
consonant
alveolar
alveola r lateral
dental
m P
t
b
system postalveolar
palata 1
glottal
velar
rj
n d
k
g
affricate f
fricative liquid glide
V
9
8
*
s r
w
X
1 i
h
Notes 1. ITJ is peripheral in these accents, as it may occur in borrowings. 2. In formal speech, a distinction between hi and hi may be found in mutation contexts. 3. hi is a trill in most contexts; l\l is clear in all contexts (though dark in northern accents). The lateral fricative enters into a system of morpho-syntactically triggered phonological changes, called 'consonant mutation' (see Ball and Miiller 1992). The mutation known as 'lenition', or 'soft mutation' (treiglad meddal) involves the change lil to III in word-initial position. In the data reported below, all instances of potential mutated forms have been ignored, and only instances where this change would not be triggered are included. /!/ can also occur in syllable-final clusters, but singleton usage only is considered in this account.
2. The study In seeking to describe the phonological development of Welsh-speaking children, the project team described in Ball et al. (1998) and Munro et al. (in
bilinguals
271
preparation) considered Welsh within a bilingual context as so many of the children acquiring Welsh are, at the same time, also mastering English. In South Wales (where the project was based) it would have been difficult to find sufficient numbers of monolingual Welsh-speaking children but, even if this had been possible, it would have been unrealistic to ignore the predominant bilingual milieu. The aims of the project, therefore, were first to provide profiles of phonological acquisition which are typical enough to be generalizable to other regions of Wales, and to present methods of analysis which are sufficiently illuminating to be utilized in similar studies on children in Wales. Subjects were divided into five age ranges with the aim to have 20 subjects in each group so thatvgroups would be large enough to allow generalizations to be made from results. Subjects were also divided, by means of a language background questionnaire (based on that of Baker 1992), into Welsh-dominant bilinguals and English-dominant bilinguals. Only Welshdominant bilinguals were found for the youngest age group. The make-up of the five groups is shown in Table 2. The sample collected from the subjects consisted of both elicited single words and words in context, and the sample size was 50 words per language at the single word level and the same items and others within connected speech. High quality audio recordings were made of the sample for each subject, which were transcribed by members of the project team using the International Phonetic Alphabet and, where needed, the extensions to the IPA (Duckworth et al. 1990). Interscorer reliability measures exceeded 90 percent in virtually all cases. Table 2.
Subjects used in the project Welsh
Group
Age range
dominant
English
dominant females
males
females
males
Group A
2;6-3;0
3
3
0
Group B
3;0-3;6
4
5
5
3
Group C
3;6-4;0
3
7
5
4
Group D
4;0 - 4;6
5
5
7
4
Group E
4;6 - 5;0
5
4
5
6
0
Ball et al. (1998) and Munro et al. (in preparation) discuss the full results in detail, but of main interest here is the acquisition of the Welsh voiceless alveolar lateral fricative - /*/. This is a comparatively common sound in Welsh, occurring initially, medially, and finally in the word, though is of course absent from English. It is also comparatively rare in the languages of the world (Ladefoged and Maddieson 1996) and, of course, in those languages that have been investigated in terms of phonological acquisition (though we return to a study of the acquisition of Xhosa below).
272
M. / Ball, N. Mutter, and S. Munro
The acquisition of the lateral fricative
in Welsh-English
bilinguals
213
However, it was not the cross-linguistic rarity of the sound that prompted this analysis, but rather the large amount of variability in its realization evidenced in the data, and the reasons why particular substitutions may have been chosen. The results of the study for lil and a detailed analysis of them are presented in the following section.
realizations per group can act as a basic metric of development for the lil. Table 4 shows the numbers of different realizations per group, and which forms actually occurred for each group. It is clear from this table that the amount of variability, as well as the rate of acquisition differs between the two dominance groupings, and that this variability does generally decrease over time (though note the increase in number of realizations in Group D English dominant).
3. The acquisition of lil
In the following sections, we examine the English dominant and Welsh dominant subjects separately, but also comparing across the two groups. Also, we maintain the distinction between the three word positions (initial, medial, and final), as it is clear that different substitution strategies and developmental patterns occur at these places.
The rate of acquisition of lil is shown in Table 3, where percentage correct scores for the lateral fricative are given for each group of subjects, and for the three positions in the word (initial, medial and final). It will be seen that only the Welsh-dominant group score 100 percent correct for the sound, and this is only achieved by the oldest subjects (4;6 to 5;0). The English dominant group only reach around three-quarters correct at this final stage (and then only in non word-final position). It is also noticeable that for all subjects, lil in word-final position is more difficult to produce correctly than elsewhere in the word, with word-initial position seeming easiest. Table 3. Acquisition of /V
Groups Group A
Group B
Group C
Group D
Group E
(percentages)
Position
Welsh dominant %
I M F I M F I M F I M F I M F
69% 83% 69% 47% 46% 20% 72% 80% 67% 75% 52% 50% 100% 100% 100%
English dominant
Table 4. Realizations
Welsh dominant 0,
Group B
i, s,
Group C
i,
9,
x,
1,
xl, çl, kl, kl (8)
0,
Group D
t, s,
9,
x, k, si, 91, xl (8)
0,
Group E
However, these numerical results do of course mask the patterns of substitution that occur in the data. Whereas other sounds investigated in the project (such as velars and consonant clusters) show relatively straightforward patterns of substitution (similar in many respects to those of other languages investigated), lil had the following realizations (though many of these occurred only once or twice): [ 0 , f, v, 9, 8, s, i , / j , c, x, 1, t, k, g, fl, 61, si, x l , tk, st] ( 0 stands for deletion). Many of these realizations were restricted to certain groups of subjects, and, indeed, the numbers of
English dominant
Group A
i, s, 9,
9,
x, k,
1,
si, kl ( 9 )
x, 1, si, xl, kl, tl ( 9 )
correct lil
46% 12% 0% 51% 42% 5% 68% 57% 21% 81% 67% 50%
of target /V
t(D
i, s, e, f, v, x, ç, k, g, 1, h, si, 81, xl, tl, tk (16) i, s, 9, f, J, x, k, 1, si, fl, 91, xl (13) i, s, 9, 8, v, / j , x, t, 1, h, si, 91, xl, kl, st (17)
J,
i, 8, x, fl, xl (5)
3.1 English dominant subjects In this section we examine the usage of the four groups of English dominant subjects: Groups B, C, D and E. It is immediately clear that for most of these groups, target lil in initial position is realized successfully more often than in medial and, especially, final position. Indeed, only Group E reaches 50 percent correct lil in word-final position. Table 5 illustrates the categories of substitutions found for the subject groups in different word positions; raw scores rather than percentages are used due to low numbers in some cells. For ease of illustration, we have established categories of realizations: L L - [i], F - any fricative, FL - any fricative plus [I], and O - other (mainly stops, or stop plus [1], plus deletions). The occurrence of [1] for lil was rare, and so did not warrant a separate category. It should be remembered that group scores hide the fact that in many instances individual group members either used a wide range of substitutions and few target pronunciations, or were very high percentage target users (albeit this differed from place to place within the word). There were numerous examples of the velar fricative and the velar plus [1]. Fricatives close to the target place of articulation were expected (e.g. [s,
274
The acquisition
M. J. Ball, N. Millier, and S. Munro
6, fj), and indeed were to be found. However, as [x] is a less obvious substitution (from an articulatory viewpoint), we have separated [x] and [xl] into the categories X and X L . Table 5. Variant usage (numbers). English dominant
B
C
D
E
I M F I M F I M F I M F
LL 16 2 0 20 5 1 50 12 6 34 10 7
F 12 3 3 7 1 6 4 1 9 0 0 2
X 1 1 0 4 2 0 1 1 0 2 0 0
FL 2 4 7 1 .0 2 6 3 6 4 1 5
subjects XL 0 2 0 2 2 0 5 3 0 2 4 0
0 4 4 2 5 2 9 7 1 8 0 0 0
3.1.1 Word initial usage. For Groups B and C, use of [i] is about the 50 percent level, though this improves considerably in Groups D and E to twothirds and four-fifths respectively. The most commonly occurring category of substitution is the fricative group, but it should be noted that the velar fricative becomes more popular throughout the age range, such that only velar fricatives are found here in Group E. Fricative plus lateral clusters increase somewhat with Groups C and D , and are on a par with fricatives in Group E. Again, the use of the velar fricative increases in fricative plus lateral clusters, at least through the first three age groups. The 'Other' category with these subjects consists mainly of stops (the most common stops are [k] and [g]), [1], and clusters such as [kl]. These last are fairly common attempts at [i] by non-Welsh speakers, so may have been heard by the English-dominant children from adults. The overall trend in initial position is a gradual increase of target [1]; and with substitutions, a gradual move away from front fricatives towards the velar fricative, either singly or in a cluster with [1] (the clusters gradually becoming as common as single fricatives). 3.1.2 Word medial usage. As we can see from Table 5, numbers of potential IV targets were much fewer in medial than in initial position (reflecting to some extent the patterns of frequency in the adult phonology). Correct use of the target increases from a very low point in Group B to just under and just over 50 percent in Groups C and D respectively to about twothirds in Group E. Substitutions are split more or less evenly between fricatives and fricative plus lateral. As with initial position, the numbers of velars found in
of the lateral fricative in Welsh-English
bilinguals
275
these categories increase across the age groups, such that with Group E only velars are found in fricative and fricative plus lateral substitutions. The 'Other' category reduces across the age ranges, and is absent in Group E. Typical realizations for this category include [k], [g] and [1] as singletons, and [tl], [kl] and [ik] as clusters. The trend in word medial position across the age ranges was a gradual increase in target pronunciations coupled with a move towards velar or velar plus lateral substitutions. These substitutions still accounted for half the realizations by Group E. 3.1.3 Word final usage. Word final patterns are characterzsed by a generally low target usage, higher use of the 'Other' category than elsewhere in the word, and total absence of fricative plus lateral clusters. Use of [1] is absent in Group B , negligible in Group C, reaches to just over a fifth in Group D and a half in Group E. Substitutions are predominantly in the fricative category (with velar fricatives common for all but Group C). Interestingly, the 'Other' category, strong for both Groups C and D , is predominantly [1] (though [k], [t], [J] and [st] also appear). It could be that [1] is the positional variant of fricative plus lateral in word final position for these speakers. The very few deletions (three in all) for the English dominant speakers occur in word final position. The trend in word final position, therefore, is a very gradual increase in target pronunciations, with velar fricatives as the main alternative together with a certain amount of use of the non-fricative lateral.
3.2 Welsh dominant subjects In this section we examine the usage patterns of Groups A to E of the Welsh dominant speakers. Again, it is clear that [1] use is strong in word initial position with word medial usage similarly strong. Table 6 also demonstrates a rather erratic developmental pattern (Group B , for example, generally show higher substitution rates than Group A). This may, of course, be due to the relatively small number of subjects in the groups, but does also reflect a tendency for advances in phonological development to be followed by short periods of apparent increased variability, perhaps because 'advanced' forms are reduced to the existing system for a period before the system can develop to encompass them (see Ingram 1986, 1989). 3.2.1 Word initial usage. Group A shows over two-thirds use of target [i] in word-initial position, though this drops to just under half for Group B , climbing again to around three-quarters for both Groups C and D , and 100% for Group E. The dominant substitution patterns in Groups A to D involve
276
The acquisition
M. / Ball, N. Midler, and S. Munro
the use of fricatives (fricative plus lateral is virtually absent for these subjects). The usual fricative used by Group A is [s]; in Group B [s] is the normal fricative though use of [x] is also found. In Group C [x] is the only fricative found, though in contrast to the English dominant subjects, [x] usage recedes with Group D (where [s] and [9] are also found). The 'Other' category includes a few examples of [1], and clusters like [tl], [kl] and [ki], as well as a surprising number of deletions in Group A. Table 6.
A
B
C
D
E
Variant usage (numbers). LL
F
I M F I M F
33 10 11 26 11 4
12 1 3 19 6 14
I M F I M F I M F
48 20 18 44 12 10 35 10 14
0 0 2 9
3 6 0 0 0
Welsh dominant X 2 0 0 0 4 0 0 0 0 0 5 0 0 0 0
subjects
.FL
XL
0 0 1 6
0 0 0 0 1 0 1
2 2 15 4 7 6 1 2 0 0 0
1 0 0 2 0 0 0 0
0 1 0 1 4 0 0 3 0 0 0 0 2 0 0 0
Overall, the developmental pattern with these subjects differs from the English dominant speakers through the overall higher use of the target form, the lack of the fricative plus lateral substitution category, and the earlier peaking' of velar substitutions. 3.2.2 Word medial usage. Again, there is no smooth progression in the acquisition of [i] in word medial position. Group A speakers use the target form over 80 percent of the time, but this drops to under half in Group B; goes back up to 80% in Group C, but dips back to just over half for Group D, before reaching 100% with Group E. It is difficult to say whether this pattern is due to subject numbers, the tendency for variability after initial success referred to earlier, or whether it is linked to the rise and then fall of velar forms that we refer to below. Unlike word initial and word final position, these subjects do make use of the fricative plus lateral as well as the fricative category in their substitutions, although it only occurs in Groups B, C, and D. [s] is the usual fricative substitution in Groups A and B, [x] only is found in Group C, but [9] is the usual fricative for Group D. As well as [xl], [si] and [91] are the common fricative plus lateral substitutions. The overall trend for this word position would appear to show an increase in velar usage up to the middle
II
of the lateral fricative in Welsh-English
bilinguals
277
age range, followed by its decrease, and competing fricative and fricative plus lateral strategies. However, we must stress that the number of tokens in this word position is not large, and so we cannot be as confident in our conclusions as we were for word initial position. 3.2.3 Word final usage. As with the previous two word positions, there is a decrease in target pronunciations between Group A, with over two-thirds use of [i], and Group B with only a fifth. This rises in Group C to two-thirds once more, with Group D at 50%, and Group E 100%. The reasons mooted above for this pattern can be applied here as well. Substitutions in word final position are similar to the English dominant speakers in that the fricative plus lateral category is not made use of, but differ in that the 'Other' category is virtually absent in the Welsh dominant subjects. This means that nearly all substitutions are of fricatives: and again, we see an increase in velar fricatives up to Group C, after which they are less important. The fricatives [s] and [9] are the dominant substitutions in Groups A, B, and D, while [x] is dominant in Group C. Overall, then, this word position shows movement between target lateral fricative, and fricative substitutions, with the velar fricative 'peaking' in the middle age grouping.
3.4 Discussion We do not intend to consider the differences between the Welsh and English dominant subjects in any detail in this section. The fact that the Welsh dominant subjects achieved higher rates of target pronunciations sooner than the English dominant is predictable, and the differential substitution patterns have been discussed in detail above. We have also noted the overall number of different realizations found with each subject group, and how this number alters across the age ranges. 1
More interesting, however, are the possible motivations for the choice of substitution categories. In particular, we will examine the choice between a fricative and a fricative plus lateral, and that between a front fricative and a velar. At first sight, it might appear that subjects who do not have consistent control over a lateral fricative realization of target N are resorting to a variety of articulatorily related alternatives. So, we could claim that by producing [s] or [9] (the two commonest fricative substitutions), subjects are reproducing the fricative part of lil at more or less the correct place of articulation. Further, when they produce fricative plus lateral combinations, they are demonstrating a deconstruction of the target N into its fricative and lateral components; and, again, we could argue that this is a strategy to simplify the target articulation.
278
M. J. Ball, N. Muller, and S. Munro
The acquisition
However, this does not seem to account for the common usage of [x] and [xl] which, for the English dominant subjects at least, seems to increase over the age ranges. It appears to us that an alternative, acoustic, explanation accounts for the developmental patterns more convincingly than an ease of articulation proposal. We would argue that the children are attempting to produce a target form that has both friction (with noise in a particular frequency range), and a clearly defined formant pattern. Early attempts to reproduce this concentrate on the noise component, whereas later attempts try to marry the noise component with formants. To examine this idea in more detail, we need to look at the acoustic structure of Welsh fricatives. 3.4.1 Noise spectra of Welsh fricatives. Ball and Williams (2001) report on an acoustic analysis of the voiceless fricatives of Welsh, they note that the acoustic characteristics of the front fricatives fit in well with the values reported in Strevens (1960), which are reproduced below as Table 7.
of the lateral fricative
in Welsh-English
bilinguals
279
together with values for /s/. This figure clearly demonstrates that Ixl has formant-like peaks of energy (not found with /s/, which has random peaks of energy across the noise spectrum), similar to the formant-like structures found with lil. (Ball and Williams [2001] note that the formants of lil are very similar to those found with III). Furthermore, while the noise spectrum of Ixl is much more diffuse than that of lil, the two sounds have the same top limit (just over 7000 Hz), and the bottom limit of the noise spectrum for lil coincides with the top formant-like energy peak of Ixl, and the lower limit of the noise spectrum of Ixl is close to the middle formant of lil. On the other hand, Is/ has a higher upper limit for its noise spectrum, has no clear formant-like energy peaks, and the lower limit of its noise spectrum does not coincide with any prominent acoustic event for lil. Hz 8000 6000
Table 7. Fricative spectra after Strevens (I960) (all measurements
m [fi [e] [s]
m
[ç] [x] [xl [h]
lowest frequency 1600/1650 1500/1700 1400/2000 3500 1600/2500 2800/3600 1200/1500 700/1200 4000/7000
peaks of energy 18/2000,40/4500,5500 1900,4000. 5000 2000+ 1000 apart random, over 1000 apart random, over 1000 apart at 1000 intervals 2000, 3500 formant-like, 1000, 2400 formant-like, 1000, 1700
in Hz)
highest frequency 6500 7500 72/8000 8000+ 7000 60/7200 50/7500 60/7000 6500
However, as noted in Ball (1984) and Ball and Williams (2001) the fricative represented by in Welsh orthography may be either velar or uvular (to some extent this may be regionally determined), so in Ball and Williams (2001) it was decided to examine the acoustic characteristics of this sound separately, together with those of lil. In Table 8 below we report the results (examining southern speech only to fit into the target accent of the children described above). The results for Ixl suggest that this sound is velar for the southern variety reported here, as the values are mostly similar to those reported by Strevens (1961) for the velar fricative. Table 8. Fricative spectra of M and /V in southern speech (all measurements lowest frequency 1367 5567
[X]
M
peaks of energy/formants 3634, 5900 234, 1500, 2400
in Hz)
highest frequency 7334 7334
What Tables 7 and 8 make clear is that there is considerable similarity between the acoustic make-up of /x7 and lil, and we show this in Figure 1,
*
4000 2000 0
[s]
[x]
Figure 1. Acoustic characteristics
[i] of[s],
[x] and [i]
The three fricatives Is/, IQI and Ixl are all part of the phonological system of Welsh consonants. It is not surprising, therefore, that [s] as a default fricative and [6] as its articulatory neighbor, are early substitutions for lil. However, speakers soon move towards [x] (which, unlike the velar stops, is virtually completely acquired in age-range A by both Welsh dominant and English dominant subjects), and we claim this is for acoustic reasons, in that [x] is perceptually closer to [i] than are the other fricatives. An alternative strategy to produce a fricative with formant-like structure is to deconstruct the target into its component acoustic parts. So we see the use of fricative plus lateral, giving us a noise component and accurate formant values. Many of the subject groupings, however, show a move to [xl] as the favored fricative plus lateral combination and, again, we would argue as we did above that this is for acoustic reasons. Fricative plus lateral combinations do not occur frequently with the Welsh dominant speakers; and these speakers also move away from singleton velar usage. We argue that this is because their ability to produce target [i] is stronger, and occurs earlier than that of the English dominant subjects. Therefore, they do not need to develop the cluster strategy and do not need to maintain the velar 2
280
The acquisition of the lateral fricative in Welsh-English
M. J. Ball, N. Muller, and S. Munro
bilinguals
281
strategy beyond about age 4;0.
Acknowledgement
3.4.2 Acquisition of lateral fricatives in Xhosa. As noted above, Xhosa is one of the few languages other than Welsh having a lateral fricative that has been investigated in terms of phonological acquisition. Xhosa (like Zulu) of the Nguni branch of Southern Bantu has both a voiced and a voiceless lateral fricative ( % if). Mowrer and Burger (1991) looked at phonological development in Xhosa- and English-speaking children in the age range 2;66;0. It was hoped that substitution patterns used by the Xhosa-speaking children for the lateral fricatives might show similarities to those described above, and support the analysis proposed. However, Mowrer and Burger found that the majority of their subjects had correctly mastered both /tj/ and lil by the age of 2;6 (with a few examples of [6] for [i] recorded). The early age of acquisition could well be explained by internal phonological pressures of the Xhosa sound system. Xhosa has a comparatively large number of consonant phonemes, with a range of clicks, and some ejective and implosive phonemes. Certainly some of the clicks and ejectives (as well as palatal plosives) appear among the later acquired sounds, so lateral fricatives may be early because both laterals and many fricatives are also acquired early according to Mowrer and Burger. This study, therefore, does not throw any further light on the Welsh results. We must await further studies of languages which do have lateral fricatives.
We would like to acknowledge the help given by Martin Duckworth for his work on the initial data collection.
4. Conclusion In this study we have attempted to demonstrate that much of the apparent widescale variability in the acquisition of the Welsh lateral fricative can be accounted for if we examine the acoustic characteristics of the variants concerned. Differences in rate of acquisition and amount of variability are clearly linked to the dominant language of the subjects, but the use of velar and velar plus lateral substitutions appears to derive from their acoustic similarity with the target lateral fricative. These results clearly suggest that perception rather than production is the dominant feature in the acquisition of the Welsh lateral fricative: it will be illuminating to see whether similar patterns are found in other languages with lateral fricatives. University of Louisiana at Lafayette University of Wales Institute, Cardiff
Appendix - Raw scores Table I. B2f 133 B13
Group B (English
dominant)
Initial 44 il t l . v l,g l,sl 1
Medial x 1, xl 1 i 1 91 1, xl 1
Final k 1 x 1,9 1 s2
tl,s4,fl,h 1
x2 x2
B8m Bll B12
tl,x2, e l tl,s2,kl,g 1
x 1 x 1 H , s 2 , tl l,tk 1
B14
tl.sl.kl
s 1, k 1, g 1,
Çl x 1
B16
s l,g 1
x 1
1 1
Table II. C6f CI C12 C13 C8m C16 C17 C18 C19
Group C (English
Initial il, 91 2,9 1 i 1,1 1,si 1 i 4 i 1,93 I I,si, fl 1 i l,13,xl xl 1,1 l , k 1 i 5 i 5
Table III. D9f Dll D13 D18 D6m D7 D14 D15 D19 D20 D21
Initial i4 i 11 * i,m i 2, xl 5 i 6 i l,x6, h 2 i 6 1 4 • 5 i 2, kl 3, si 1 i 2, kl 2,1 2
dominant)
Medial 9 1 i 1 i 1 91 1 12 1 1 xl 1 i 1,91 1 i 1, xl 1 Group D (English Medial i 2 i 3 i 2 xl 1,91 1 i 1 i l,x 1 i 1 i 1 xl2 kl 1 x2, v 1
Final f1 9 1 x 1 92 18,02 1 1 _ i 1 x 1 dominant) Final 1 2 i 3, 8 2, 9 1 9 2, J 1 x2 i 1 x l,s 1 12, st 1 9 2, 0 1 x3 13 1 1
282
M. J. Ball, N. Muller, and S. Munro
Table IV. E7 f E9 E10 Ell E13 E15 E8 m E12 E16 E17 E18
Initial 15 13 15 14 14 x3 fl 2, x 1 14 13 14 12, xl 2
The acquisition of the lateral fricative in Welsh-English
Group E (English
dominant)
Medial 11 12 1 1 11 1 1 xl 1 x1 11 ll,xl2 12 xl 1
Table VIII. Final 1 1 61 1 1 42
Dl f D2 D5 D8 D17 D3 m D4 DIO D12 D16
e
l x1 x2 12 1 1 x1 x1
Group D (Welsh
Initial 14 13 17 14 1 10 x6 14,92 s7 13 19
Al f A4 A5 A7 A2m A3
Initial 14, 0 3, s 1. si 1 1 4 , 0 2, kl 1 17 ill s 4,0 1 17,1 1,0 1
Table VI. Bl f B5 B6 BIO B17 B4m B7 B9 B15
CI f C2 C5 CIO Cll C14 C15 C4m C9 C20
Medial 1 1 13 12 12 s1 12 Group B (Welsh
Group C (Welsh
Initial 17 I2,x7,kl 1 x 3 , xl l . k l 1 15, kt 1 16 17 15 17,1 1 14, x 5 15
Final 12 12 12 11 11 x2 92 s3 k2 12, 9 1
Group E (Welsh
dominant)
dominant) Final k 1 •3 12, 9 1 12 s2 14, x 1
E5 f E6 E14 E19 El m E2 E3 E4 E20
dominant)
Medial s2 s 1, si 2 1 3 1 2 1 5 x2 xl 1 1 1, s 3 si 2
Initial 1 3, s 7 t 4, s 2, tl 1 1 6 1 4,1 1 1 5 x6 1 4, kl2 s5 s5
Table VII.
Group A (Welsh
283
dominant)
Medial 13 1 1 1 1 1 1 12 x 1, xl2 93 si 3 912 14
Table IX. Table V.
bilinguals
Final s4 s4 1 2, 9 1 1 2, 9 1 x1 x1 s2 92
dominant)
Medial 13 x2 14 13 12 13, çl 1 13 x2 12
Initial 14 13 15 15 13 13 14 13 15
Medial 11 11 11 11 11 11 12 11 11
Final 13 11 11 11 12 12 12 1 1 11
Notes 1. 2.
We also do not comment on sex differences between the subjects. There is some evidence that girls produce more correct target pronunciations than boys, but numbers are insufficient to discuss this in detail. It should be noted, however, that /0/ is acquired consistently comparatively late in both Welsh and English by both major groupings of subjects.
References Final 13.x 1 11, x 3 x1 13 11,92 1 1 13 14 x2 12
Baker, C. 1992 Attitudes and Language. Clevedon: Multilingual Matters. Ball, M. J. 1984 Phonetics for phonology. In Ball, M. J. and G. E . Jones (eds.), Welsh Pho_ , nology. Cardiff: University of Wales Press, 5-39. Ball, M. J. and G. E. Jones (eds.) 1984 Welsh Phonology. Cardiff: University of Wales Press. Ball, M. J. and B. Williams 2001 Welsh Phonetics. Lewiston, NY: Edwin Mellen Press. Ball, M. J. and N. Miiller 1992 Mutation in Welsh. London: Routledge. Ball, M. J., S. Munro, N. Mailer, and M. Duckworth 1998 Establishing phonological norms in bilingual language acquisition. First Asia-Pacific Conference on Speech, Language and Hearing, University of Hong Kong. '
284
M. J. Ball, N. Mutter, and S. Munro
Duckworth, M., G. Allen, W. Hardcastle, and M. J. Ball 1990 Extensions to the International Phonetic Alphabet for the transcription of atypical speech. Clinical Linguistics and Phonetics 4, 273-80. Ingram, D. 1986 Phonological development: Production. In Fletcher, P. and M. Garman (eds.), Language Acquisition. 2nd edition. Cambridge: Cambridge University Press, 223-239. 1989 First Language Acquisition. Cambridge: Cambridge University Press. Ladefoged, P. and I. Maddieson 1996 The Sounds of the World's Languages. Oxford: Blackwell. Mowrer, D. and S. Burger 1991 A comparative analysis of phonological acquisition of consonants in the speech of 2 1/2-6-year-old Xhosa- and English-speaking children. Clinical Linguistics and Phonetics 5, 139-164. Strevens, P. 1960 Spectra of fricative noise in human speech. Language and Speech 3 , 3 2 ^ 9 .
