2018 Alternates 6424

2018 Alternates |Saturday, November 3, 2018 | Poster Session II, Metcalf Small | 3:15pm

Acquisition of Ellipsis: Phonology or Syntax?
Y. Fujiwara, H. Shimada

Introduction

The goal of this study is to see which operation children apply to ellipsis phenomena such that they require a linguistic antecedent, phonological or syntactic deletion. In Universal Grammar, both types are allowed. For example, deletion of initial materials in English colloquial speech

(1) is not derived by syntactic deletion but by phonological deletion since syntactic operations cannot be applied to a non-constituent or a part of a word.[1] However, ellipsis phenomena requiring a linguistic antecedent is not derived just by phonological deletion.[2] As in (2), ellipsis rescues island-violations. Moreover, it cancels the polarity sensitivities of polarity items.[3] The elided NPI in (3a) need not be licensed by negation and the covert PPI in (4a) takes scope under negation. The differences between the overt and covert elements in (2-4) would not be expected if deletion here were a mere phonological operation.

Previous studies
Acquisition studies on ellipsis have been investigated mainly in Japanese, which is a so-called radical pro-drop language. Although an empty pronoun pro can occupy null positions in Japanese, many studies have reported that four- to five-year-old children can apply ellipsis to interpret a null position.[4,5] They claim that children can access interpretations derived only by ellipsis such as the quantificational reading in (5a), which cannot be derived by pro (cf. 5b).

However, they cannot tell whether children apply purely phonological deletion or syntactic deletion to a null position because the quantificational reading in (5a) can also be obtained with its overt counterpart (6).

Research Question
This study investigates whether children are sensitive to different interpretations between overt and covert conjunction in Japanese. The conjunction -mo-mo behaves as a PPI in (7b),[6] while deletion cancels its PPI property (cf. 7c).[7] Thus, it takes scope under negation when it elides.

Experiment
To address this question, we tested (7b/c) with 13 adults and 17 children (age 5;5-6;4, Mean 5;9). We also examined (8b) as control to see whether they can interpret a null position using ellipsis. In this experiment, a puppet is asked a question like (7a/8a) after a story such as (7d/8c). The participant’s task is to judge whether puppet’s answer like (7b,c/8b) matches the story. There were three trials for each item.

In (8c), the puppet answer (8b) to (8a) is true under the quantificational reading, which is derived by ellipsis, whereas it is false under the pronoun-reading. The children’s acceptance rate was 84.3% (43/51). The adults completely accepted the control trials (8b). This suggests that they can use ellipsis. However, there were two children who did not accept them at all. Thus, we excluded them from the analysis of the target trials.

(9) summarizes the results of the target trials (7). In (7d), (7b) is false and (7c) true. The 15 screened children rejected (7b) at 80%, while they accepted (7c) at 75.6%. This contrast is significant by Wilcoxon Signed-Rank Test (Z=2.88, two-tailed p=.0004). The adults completely rejected (7b) and accepted (7c). This indicates that children’s ellipsis occurs not in phonology but in syntax like adults.

References

[1] Napoli (1982) Initial Material Deletion in English. Glossa 16(1): 85–111. [2] Merchant (2001) The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis. [3] Sag (1976) Deletion and Logical Form. Ph.D, MIT. [4] Sugisaki (2007) The Configurationality Parameter in the Minimalist Program: A View from Child Japanese. BUCLD 31. [5] Otaki (2014) Ellipsis of Arguments: Its Acquisition and Theoretical Implications. Ph.D. UConn. [6] Goro (2007) Language-Specific Constraints on Scope Interpretation in First Language Acquisition. Ph.D. Maryland. [7] Funakoshi (2013) Disjunction and Object Drop in Japanese. TAW 4.