Copula omission in the English developing grammar of English/Spanish bilingual children

Abstract

The present study takes as a point of departure Becker's analysis of the copula be in English monolingual data and focuses on the distribution of copula be in the data from two English/Spanish bilingual children. Our data analysis shows that, as in Becker's study, the distribution of copula omission in the bilingual data is determined by the nature of the predicate. However, the omission patterns in our English bilingual data do not coincide with those described by Becker for the English monolingual data, since total omission is very low in our data and there are no significant differences between the stage-level (SL) and the individual-level (IL) predicates. We attribute this to crosslinguistic influence from Spanish, specifically, to the existence of two distinct copulas in Spanish, ser and estar; in particular, we propose that the lexical distinction between these two predicates may trigger the earlier projection of inflection and with it the use of an overt copula in both languages, but specifically in English, and for both SL and IL predicates.

Keywords:

• copula omission
• stage-level predicates
• individual-level predicates
• English/Spanish bilingualism
• crosslinguistic influence
• bilingual acceleration

Acknowledgements

This research has been funded by the Spanish Ministry of Science and Technology [HUM2007-62213] and FEDER [DGICYT #BFF2002-00442] and by the Faculty of Arts of the University of Ottawa and the Social Sciences and Humanities Research Council of Canada [SSHRC #410-2004-2034].

Notes

1. The MLU (Mean Length of Utterance) is derived from two totals: the total number of utterances and the total number of either morphemes (standard MLU) or words (MLUw) for each speaker and in each file/transcript. MLU calculations for the twins were based on word measures (MLUw), while those of the English monolingual children were measured on morphemes (standard MLU). When comparing standard MLU and MLUw values, Malakoff et al. (1999) found that MLU correlates with MLUw at 0.97 for English, and Aguado (1988) found a correlation of 0.99 for Spanish (see MacWhinney 2000, 103).

2. The twins’ average MLU for the study period is slightly below the corresponding MLU in the English monolingual data (three for monolinguals, 2.8 for bilinguals). However, given that twins in general seem to lag behind their singleton counterparts in language development (Dale et al. 1998, among others), we do not consider this to be anomalous. Also, in spite of the fact that the twins’ MLU is slightly lower than that of the monolinguals, their rate of omission is very low (see Section 6). Thus, this reinforces the conclusion that the bilinguals and the monolinguals are very different in this respect and that this may be a consequence of the accelerating effect that Spanish has on their English (see Sections 6 and 7).

3. If there is a bilingual effect, a third possibility would be default retention (as in Pérez-Leroux, Pirvulescu, and Roberge 2009): children in a bilingual context may retain the default grammar of omission (as in root infinitives or functional omission in general) for a longer period of time in either language. This would imply that: (a) English bilingual data will present many more cases of omission than English monolingual data; and (b) this omission will also take place in the Spanish bilingual data. However, as different studies have suggested, instances of copula omission in both monolingual and bilingual Spanish are scant (Bel 2001; Sera 1992; Silva-Corvalán and Montanari 2008).

4. As we indicated in Section 3.2., sessions in the Ferfulice corpus are usually conducted either in English or in Spanish. However, Leo and Simon are Spanish/English bilinguals and may use Spanish words or sentences in an English session, or vice versa. We have used @sp if they use a Spanish word or sentence in an English session when speaking in an English context (i.e. to an English native speaker or person who primarily communicates with them in English). We have used @en if they use an English word or sentence in a Spanish session when speaking in a Spanish context (i.e. to a Spanish native speaker or person who primarily communicates with them in Spanish).

5. An anonymous reviewer wonders how it was decided that examples like this one or the one in (21g) did not have a copula, since in the flow of speech, even if the child had produced a copula, it would have been neutralized by the following ‘s’ in still or sorry. We have followed the way utterances appear in the CHAT transcription file and, although we agree that the reviewer has raised a serious concern here, given the scant number of null be found in the data, a lower number of null cases would only but reinforce the scarcity. Nonetheless, we have reviewed the data to determine whether there were more cases like the ones in (21e) and (21g); namely cases in which a following or a preceding –s might have neutralized the overt copula is. We have found four more cases (3 with a preceding /s/ sound or similar: The suitcase _ heavy; because this _ for my good (wolf); because (it) __ my good (wolf); and one with a following /s/ sound or similar: I _ strong one). These six cases represent 1.3% of the total production and 15.3% of the null be production. We have also reviewed the data for cases of overt copula in which copula is is either followed by a word beginning in an /s/ sound or similar (e.g. it is sheep) or preceded by a word ending in an /s/ sound or similar (e.g. Max is in), since using the same rationale, these cases could be equally problematic. There were 13 examples of this type of null and overt copula is, which represent 2.8% of the total production. There are six null is cases (three per child) and seven overt is cases (three for Simon and four for Leo); these problematic cases balance themselves out. So none of the utterances (overt/null) and neither of the children (Simon/Leo) offers a different picture by this coding of problematic cases. This implies that, if we were to take them out of the counting altogether, the results would not change.

6. The percentages in the tables, both for the English monolingual and for the English bilingual data, were calculated over all the files rather than over each file and then averaging across the files.

7. The number of occurrences corresponding to the monolingual data in Table 15 has been inferred from the information provided by Becker (2004) since she offers the exact number of overt copula utterances but not that of null copula utterances.