2018 Sat Session C 1700

Saturday, November 3, 2018 | Session C, Terrace Lounge | 5pm

Pulling the curtain back: Wh-questions in child Tok Pisin
H. Sato, K. Deen

We present the first ever study of the acquisition of Tok-Pisin wh-questions (subject-, object- and indirect object-WHQs). We show that (i) subject-WHQs are produced more accurately than other questions, (ii) adults produce almost all WHQs in situ, but (iii) children often front WHQs. More interestingly, (iv) children produce an interesting set of resumptive elements, despite no evidence from the input for these. Importantly, these errors evince tendencies observed in other languages across the globe, revealing foundational preferences for question patterns that have as their source something other than the input.

Tok-Pisin is one of the official languages of Papua New Guinea and is its most widely-spoken language. While an English-based creole, Tok-Pisin and English are mutually unintelligible (Holm 2000:96). We conducted two experiments with the same set of participants.

Participants: 23 children were initially tested; 11 were exclude for failure to complete training or responding inappropriately to test items, leaving 12 children (3;1-6;0, mean=4;1), and five adult controls (additional data collection on-going). All participants came from West New-Britain (where English is rare) and are L1 speakers or strongly Tok-Pisin-dominant bilinguals.

Experiment 1: Character selection task. Participants were shown a two-panel set of pictures and asked wh-questions in Tok-Pisin (e.g., 1, object-WHQ), fig 1. Children and adults performed at ceiling, showing good comprehension of Tok-Pisin WHQs. No difference in subject-, object- or indirect-object questions was found.

Experiment 2: elicited production task (Yoshinaga, 1996). Participants were shown pictures like fig.2 and told that their caregiver (present for the experiment) knew the identity of the obscured character – they should ask the caregiver. Participants received 10 subject-, 10 object-, and 5 indirect-object-WHQs.

Results show adults at ceiling, producing WH-in situ at ceiling. Children showed significant deviation from adults (table 1):

  • Overall accuracy was low: target-like subject-WHQs occurred <50%, and object-WHQs and indirect- object-WHQs <30%.
  • The most common error-type was the (non-target-like) fronting of WHQs, accounting for 30% and 45% of object-WHQs and indirect-object-WHQs,

Discussion. Children clearly find S-WHQs the easiest, confirming the well-known subject-extraction advantage. However, two further important points emerge. First, Tok-Pisin children front WHQs, despite WHQs robustly occurring in situ in their environment. None of the adults fronted WHQs; follow-up interviews confirmed fronted WHQs are very unnatural. Second, and importantly, many of the fronted WHQs included a resumptive element at the point of extraction: a pronoun (2), a copy of the fronted wh- element (3), the nominal that was being questioned (4), or some kind of indefinite (5). The spontaneous occurrence of resumptives (and this set of resumptives in particular) is the most significant of our findings, since this clearly is not modeled in the input to children: resumptives never occur in the input since fronting never occurs. This reveals that the fronting strategy for WHQs is something children intuitively consider (even when not present in the input), and when they do so, they naturally mark the position from which the WHQ moved. Implications for the source of fronting/resumption (Universal Grammar, processing, the input) are discussed further in the talk.

References

Holm, John. 2000. An introduction to pidgins and creoles. Cambridge and New York: Cambridge University Press.

Yoshinaga, Naoko. 1996. Wh-questions: a comparative study of their form and acquisition in English and Japanese (Doctoral dissertation).