Copula Omission in Down Syndrome

ABSTRACT

This paper explores the nature of copula omission in Cypriot Greek individuals with Down Syndrome (DS). Previous studies on DS have attributed high rates of copula omission to an overall grammatical/inflectional impairment without offering further analysis. In order to identify relevant conditioning factors, we examined copula productions and omissions from spontaneous and elicited experimental tasks under five levels of analysis: categorial type of the predicate (nominal vs. adjectival), aspectual interpretation of the predicate (permanent vs. temporary), and a combination of these first two, as well as subject status (overt vs. covert) and experimental design (spontaneous vs. elicited). Results showed that adults with DS had significantly higher rates of copula omission than TD children. We found subject overtness to be the most reliable predictor of copula omission. A comparison of the two experimental methods of data collection also revealed a significant effect. Following an analysis based on the Unique Checking Constraint, we propose that copula omission is facilitated by the restriction that only one EPP feature (either the one in TP or the one in TopicP) can be checked. We discuss the implications of these results in the context of a distinction across a delayed vs. deviant development of the DS grammar.

Acknowledgments

We are grateful to all our participants and their families for making this research possible. We thank the four anonymous reviewers and the LLD editorial team, especially the Action Editor Prof Caroline Floccia, for all their help and support during the review and proofing process.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Further evidence suggesting the importance of the IL vs. SL predicate distinction comes from studies on bilingual children. Silva-Corvalán and Montanari (2008) reported a delay in the acquisition of estar for their bilingual Spanish- and English-speaking toddler, attributing it to the fact that the child was acquiring English and Spanish simultaneously and thus his performance with estar was affected by his exposure to English. Sera (2008) and Liceras et al. (2012) suggest that this may be due to an overgeneralization of estar, a possibility offered by numerous studies (including Sera, 1992) on the acquisition of Spanish by monolingual children, though Liceras et al. (2012) show that Spanish may positively affect English, pulling the relevant information for the two types of copulas from Spanish when using English copulas.

² # indicates the nonoccurrence of such a sentence.

³ (Cypriot) Greek does not have an infinitive. However, most of the relevant null-subject languages (Italian, Spanish, etc.) do. Nevertheless, children in the OI age-range (up to about 3) do not produce significant numbers of non-finite main verbs when they should be finite. A preliminary analysis for CG couched within the distributed morphology framework (Christodoulou, 2011, 2014; Christodoulou & Wexler, 2016; and a more refined version currently in preparation), allows, as an alternative to non-finite verb production, the use of default inflectional features: present for tense, third for person and singular for number, the latter two also serving as universal defaults (Harley & Ritter, 2002). These occur mostly independently and, less often, as a bundle. The more refined version of this analysis integrates both the OI and NSOI hypotheses.

⁴ Based on experimental data (spontaneous and elicited) as well as numerous diverse syntactic phenomena, Christodoulou (2011) and Christodoulou and Mavrogiorgos (2015) concluded that nominative is the default case for (Cypriot) Greek.

⁵ A number of studies, however, have shown that RIs/OIs can also be found, at a very young age, in some null-subject languages such those of Hebrew (Schaeffer & Ben Shalom, (2004) (2004)—a partial null-subject language according to Shlonsky (2009), Japanese (Murasugi, 2015), Russian (Bar-Shalom & Snyder, 1998; Gangarina, 2002; Gülzow & Gagarina, 2006), also a partial null-subject language, and monolingual and bilingual Spanish and Catalan (Bel, 2001; Liceras, Bel & Perales, 2006; Liceras & Fernández Fuertes, 2019). These studies show that RIs typically appear before the age of two and stop being produced by or shortly after the children’s second birthday.

⁶ Further considerations of Minimize Violations (i.e., choosing the representation with the smallest number of constraint violations) allow neither feature to be omitted, and UCC to be violated, yielding a sentence with both AGR and T, a possibility in the OI stage (Wexler, 1998). This is why it is called the Optional Infinitive stage: children often do produce a correct finite form.

⁷ An anonymous reviewer points out that nominal predicates are canonically considered to be Individual-Level (Carlson, 1977). We acknowledge that this may be the case for English, on which Carlson’s and Becker’s work were based. However, as shown in the examples in Table 3, those environments do exist in CG. Nominal predicates that have a temporal (i.e. time) interpretation and specifically a reference to a time of the day, especially when not habitual, are by definition, temporary. In English a different syntactic environment would be used to express the spontaneously produced example given in Table 3, i.e., instead of saying It’s night, one would say It’s dark or It’s getting dark.

⁸ Past research reported that 7 or 8 years is at the upper limit of the mental age achieved by individuals with DS (Stoel-Gammon, 2001, p. 96), with a reported range of 2;5 to 7;4 (chronological age: 7;3 to 41;10 (Schaner-Wolles, 2004, p. 108)). Following results from these previous studies it was decided that control participants should match approximately the highest mental age reported, since the individuals with DS participating in this study were adults. Hence, TD children aged 7 to 8 were recruited. Under the assumption that adults with DS have reached the highest level of grammatical acquisition that their abilities allow, we wanted to compare them with children who were at the latest stages of language acquisition and potentially matched the DS population’s mental age as well. Based on the dearth of information available in the literature at the time of data collection, we believed this to be the best possible option.

⁹ No refusals were recorded. In the two instances that participants seemed fatigued, we discontinued testing and resumed on a different day.

¹⁰ Predicates were categorized as IL vs. SL based on the aspectual interpretation of the predicate. Predicates were categorized as NP, AP and PP based on the head of the phrase following the copula. PP and AP predicates typically carried a temporary meaning, though we did have quite a few APs with an IL interpretation. Our copula structures included a variety of subject types: overt (simple and conjoined DPs) and covert subjects. See Table 3 for examples.

¹¹ A breakdown of copula omissions based on the inflectional features marked on the copula was also pursued for the DS group. Results revealed a marginal effect for person (F(2, 38) = 3.32, p = .047, η² =.149) and number (F(2,35) = 3.38, p = .045, η² =.162) but not for tense (F(1,27) = 0.38, p = .541, η² =.014). Scheffe post-hoc analyses conducted for person showed that third person copulas were omitted at marginally higher rates than first person (p = .050) but not second person (p = .694). Copulas inflected with the singular/plural value were as likely to be omitted as copulas inflected with plural (p = .976). The same was true for copulas inflected with the singular value (p = .078), as the comparison also revealed a non-significant or marginal difference. See Table A4 in the Appendix for raw numbers and means.

¹² We traced the participants’ performance across a number of other word categories to examine whether they exhibited similar difficulties and explored any potential correlations across those categories/omissions (Table A5, Appendix). Omission of auxiliaries presented the most interesting results of all the categories. Results revealed that auxiliary omission by the DS group was even higher than copula omission (overall omission: M_omi = 57.5, omission in elicited imitation: M_omi = 63.6, SD = 25.4, omission in spontaneous environments: M_omi = 49.4, SD = 39.8). We did find most omissions with the imitation production task, but surprisingly the participant’s production of auxiliaries was at chance level with spontaneous productions as well (Table A6, Appendix). One factor that might explain these results is that the auxiliary may undergo a phonological reduction and oftentimes just the [n] or not even a very audible production might mark its presence. Given its close relation to the copula, we traced the participants’ performance in relation to subject production. As with copula omissions, there was a very clear preference for auxiliary omission with overt subjects (M_omi = 76, SD = 32.3), while auxiliary omission was at chance level with covert subjects (M_omi = 50, SD = 25.3) (Table A7, Appendix). Additionally, not a single case of simultaneous auxiliary and subject omission was recorded. For specifics on the status of the auxiliary in CG and how it functions, we refer the reader to Christodoulou (2011).

¹³ A UCC-based analysis is further supported by results from our small dataset on auxiliaries, which show an even stronger preference for auxiliary omission with overt subjects (M_omi = 76.2, 31/42) and an overall preference of producing copulas with covert subjects (M_omi = 66.3, 61/92). Though one may argue that this is to be expected, as CG is a null-subject variety, we note that this may be true for utterances including first person, and perhaps second person, but not for third person, with which most of these spontaneous utterances were inflected.

¹⁴ In a recent study, drawing from syntactic observations and experimental evidence (2–3-year-old elicited and spontaneous productions), Christodoulou (in preparation) also argues that the underlying word order in CG is VSO.

¹⁵ The logic here is similar to proposals of Wexler (in lectures; 2004) pertaining to wh- movement, supporting that wh- feature must check two EPP features, parallel to the double checking of EPP with a pre-verbal subject under TopicP. This demand of an EPP feature could add an extra checking that would make it even more difficult to satisfy the UCC without an omission, leading to high percentages of OI’s in wh- sentences, or copula omission, in this case.

¹⁶ An anonymous reviewer suggested that temporal anchoring may be a plausible explanation for the phenomenon of copula omission in individuals with DS. In fact, Christodoulou and Wiltschko (2012) presented a brief discussion supporting temporal (INLF) anchoring as a possible analysis for the problematic use of temporal elements (i.e., omission of morphological marking, including the omission of INLF elements, an inflectional affix, copula, or auxiliary (p. 137)) in their examination of subjunctive sentences produced by the same group of DS individuals as in the current study. They suggested that their data indicated that it is not INFL (i.e., the function), which is impaired, but rather, on occasion, the association of INFL with its substantive content (i.e., the morphophonological form), resulting in infrequent omission or commission errors with temporal elements. We do acknowledge that temporal anchoring, as described in this work, is indeed a plausible analysis for problematic use of temporal elements. However, Christodoulou and Wiltschko (2012) did not conduct a thorough investigation of the potential contributing factors of copula omission, as we have done in the current research (i.e., the five levels of analysis plus the supplemental analyses included in the appendix). Based on our full pattern of results, we propose that the clear preference of copula omission with overt as opposed to covert subjects shows that, for copula omission specifically, there might be something additional or more specific going on beyond simply an occasional failure of the association of INFL with its content. Therefore, we maintain that a UCC-based analysis provides a more complete explanation for the high rates of copula omission in individuals with DS.

Additional information

Funding

This work was supported by the Canadian Department of Foreign Affairs and International Trade, Canadian Commonwealth Scholarship Program [2005-2010]; Marie Curie Actions Career Integration Grant of the European Union’s Seventh Framework Programme FP7/2007-2013/ under REA grant agreement [Marie Curie Actions Career Integration Grant of the European Union’s Seventh Framework Programme FP7/2007-2013/ under REA grant agreement PCIG11-GA-2012-322005].