6. Discusión y Cconclusiones
2.3. En el marco de la juventud
Now that we have an idea of the baseline rates of occurrence and have excluded invariant tokens from the data, we can begin to examine what might account for the variation that does exist. One possible contributor to the overall variation is that individual speakers, or speakers at different ages choose between the variants for different reasons. Data in this study can be viewed across three age spans due to its cross-sectional and longitudinal design (see Chapter 4, §4.5.4 for details). As children grow older, they are exposed to more schooling, and therefore more SAE. The
6 In fact, the importation of L1 use of ‘got’ into the SCHOOL context results may itself constitute a camouflaged form.
As noted in footnote 1, it is used in constructions for which the target SAE form is ‘has/have got’ (or possibly ‘has/have’). The use of ‘got’ therefore results in clauses that are ungrammatical from the perspective of ‘standard’ Australian English (e.g. ‘I got no paper’, ‘They got too many’) but only because they are missing the auxiliary ‘has/have’, which is often contracted to the subject making it potentially hard to detect. Further, the absence of the auxiliary doesn’t pose a big threat to communication, particularly in the kinds of contexts children are using this construction. V 24% V 'got' 10% V punctual 32% Ving 13% Vbat 21% HOME V 34% V 'got' 10% V punctual 20% Ving 36% SCHOOL
prediction, then, is that older children will lead any overall change between HOME and SCHOOL data sets. It might be possible to observe such changes in use of the variable
over time as evidence of a growing separation between L1 and L2 systems. In exploring this possibility, this section will address Research Question 1c): What speaker-related factors (specifically age) best account for the variation in choice between the main forms in each context?
One possible effect of age on the data is a well-documented trend for early learners to rely heavily on uninflected verb forms. Tense and aspectual information is conveyed through pragmatic/contextual and then lexical means (or possibly not at all for L1 learners), prior to the acquisition of verbal inflections. In studies of first language acquisition this has been called the ‘optional infinitive stage’ (e.g. Wexler 1994), and descriptions of adult second language have termed the same (at least surface) phenomenon a feature of the ‘basic variety’ (e.g. Klein & Perdue 1997). The same pattern of uninflected verb forms is found in the L2 speech of children, but naturally there is speculation regarding the underlying mechanisms depending on the age of the child (i.e. ‘pre-inflection’ per the L1 developmental stage or a case of ‘missing’
inflections per normal adult L2 acquisition or other possibilities e.g. Haznedar & Shwartz 1997; Gavruseva 2002; 2004).
Table 6-8: Distribution of V and Ving per AGE, SCHOOL context [not significant χ2(2,N=315)=4.8897, p>.05] V Ving Total % N % N 5;0-5;11 49 24 51 25 49 6;0-6;11 43 58 57 76 134 7;0-7;11 57 75 43 57 132 Total 50 157 50 158 315
Note: excluded tokens (for which specific age bracket unknown) [N=6]
Figure 6-4: Distribution of V and Ving per AGE, SCHOOL context (data from Table 6-8)
While it is not the aim of this thesis to test these particular competing accounts, it is of interest to see if children in this study go through a period of reliance on the V form in their SCHOOL language data (they are past the ‘early’ phase where this would be a
relevant consideration of their L1). The distribution of SCHOOL context data per age is
presented in Table 6-8 (and graphed in Figure 6-4). It shows that children in this corpus do use inflected Ving forms from the earliest age (and recall that these were made only a few months into the first pre-schooling exposure for the youngest children). Further, the proportional use of Ving versus V does not indicate over-reliance on V in the early age bracket. In fact, the use of V and Ving does not differ significantly across age. It may be that the period of reliance on V forms is very brief for these children, and so it wasn’t captured by the schedule of recordings used. It may also be the case that children make early use of the overlap between their HOME language and what they hear their teachers
speaking, and just continue to use their HOME language in SCHOOL contexts. We already
know that some accommodation to L2 patterns have been made: Vbat is used minimally in SCHOOL contexts [N=15] in the earliest age group and abandoned by the age of 6, as
noted in the above section (§6.2). This suggests that increasing exposure to SAE leads to a suppression of non-SAE forms. The fact that Ving is not similarly relegated supports this. This is potentially an example of where the acquisition of a closely-related language may differ from typical L2 acquisition. We can further examine the extent to which the uninterrupted use of Ving in SCHOOL contexts constitutes a continuation of HOME
language use, or adoption of L2 norms, by comparing the variable constraints on each system - and this will be done in the following section (§6.5).
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5;0-5;11 6;0-6;11 7;0-7;11 V Ving
Table 6-9: Distribution of V, Ving and Vbat per AGE, HOME context. [not significant χ2(4,
N=656)=3.3082, p>.05]
V Ving Vbat Total
% N % N % N
5;0-5;11 42 98 21 50 37 87 235
6;0-6;11 43 108 25 64 32 81 253
7;0-7;11 43 72 19 32 38 64 168
Total 42 278 22 146 35 232 656
Note: excluded tokens (for which specific age bracket unknown) [N=11]
Figure 6-5: Distribution of V, Ving and Vbat per AGE, HOME context (data from Table 6-9)
Before that I will make a few observations about the HOME language data presented in
Table 6-9 (and graphed in Figure 6-5). This distribution reveals that there are no
obvious age-related changes in the incidence of each variant over time. The significance of age on the choice of variant was tested and there was no difference attributable to age. Age can therefore be excluded as a contributor to the variation between V, Ving and Vbat in the HOME context7.
The second speaker-related factor that is of interest is the individual contribution of each participant. In order to check that no one speaker is skewing the data in terms of use of a particular variant (e.g. in case only some speakers use Ving in the SCHOOL
context), the rates of use of each variant have been graphed and presented in Figure 6-6 and Figure 6-7. There are no significant differences between speakers in either setting, meaning that the variation between speaker is not significantly contributing to the overall variation between the choice of verb forms.
7 A further investigation could be made by taking the data at each age level and conducting separate variable rule
analyses (of the kind to follow in the proceeding section) on each. This would determine if the same set of factors is conditioning the variation between the variants at each age level. Unfortunately there are not enough tokens at each age level to facilitate this kind of exploration.
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5;0-5;11 6;0-6;11 7;0-7;11 V Ving Vbat
Figure 6-6: Rates of use of V, Ving and Vbat by SPEAKER, HOME context. [Not significant χ2(10,N=656)=10.460, p>.05. Speakers for whom fewer than 15 tokens recorded are
excluded, N=11]
Figure 6-7: Rates of use of V and Ving by SPEAKER, SCHOOL context. [Not significant χ2(6,N=306)=8.97, p>.05. Speakers for whom fewer than 15 tokens recorded are
excluded, N=15]