Task Effects in the Interpretation of Pronouns

Abstract

Children acquiring a range of languages have difficulties in the interpretation of personal pronouns. Ongoing debates in the relevant literature concern the extent to which different pronoun types are subject to this phenomenon, as well as the role of methodology in relevant research. In this study, we use two different experimental tasks to examine the interpretation of pronominal elements in Greek, a language that has both strong pronouns and pronominal clitics. Results reveal a complete absence of problems in the interpretation of clitics irrespective of task. However, there were task effects in the comprehension of strong pronouns in both children and adults. Findings indicate that the phenomenon is intrinsically linked to the properties of the elements in question. Analysis of the task effects highlights how different methodological manipulations can shed light on different properties of the elements studied.

Notes

¹ Note that there are different views on the type of problems children may have with coreference. An early attempt toward the interpretation of the phenomenon has been the suggestion that children lack a pragmatic principle that regulates local coreference, called Principle P (Chien & Wexler 1990). A related account has attributed the phenomenon to children’s processing limitations, which prevent them from applying Rule I (Grodzinsky & Reinhart 1993). According to the latter view, Rule I places a burden on working memory, as it requires the simultaneous examination of two alternative interpretations. For populations with processing limitations, including typically developing children, this results in failure to apply the rule.

² Note that clitics can also be bound in discourse (not only in syntax; see Baauw et al. 1997).

³ Greek examples are not given in strict phonetic or phonemic transcription; they are loosely transcribed to the extent that serves the purposes of the present study.

⁴ This observation has been linked to an important additional property that sets Greek strong pronouns apart from typical strong pronouns (in addition to resistance to local accidental coreference). Greek strong pronouns can refer to nonhuman entities (Varlokosta 2000), despite a robust cross-linguistic generalization (Cardinaletti & Starke 1999b) that strong pronouns cannot have nonhuman referents. The only exceptions to this cross-linguistic generalization, including Greek afton, have demonstrative morphology (Cardinaletti & Starke 1999b). These properties give Greek afton membership to a special pronoun category, which consists of demonstrative pronouns that follow Principle B (B-demonstratives; Cardinaletti & Starke 1999b). According to Cardinalletti & Starke (1999b), a language may have a demonstrative pronoun following Principle B instead of Principle C, if there is no corresponding personal pronoun in the language in question. These elements are used as strong pronouns (i.e., in contexts requiring strong pronouns), can take nonhuman referents, and resist local accidental coreference. (Note that we continue referring to the element afton as a strong pronoun throughout the article, to avoid confusion.)

⁵ It is worth pointing out that several studies examining atypical Greek-speaking populations, such as children with Specific Language Impairment, have shown that the comprehension and production of these elements tend to pose difficulties to these populations (e.g., Chondrogianni, Marinis & Edwards 2010; Stavrakaki & van der Lely 2010; Tsimpli & Stavrakaki 1999; but see Manika, Varlokosta & Wexler 2011; Varlokosta 2002; Varlokosta, Konstantzou & Nerantzini 2014; Varlokosta & Nerantzini 2012, for opposite findings). This is in contrast with findings in other languages showing that although the production of these items in children with Specific Language Impairment is problematic, their comprehension is well preserved (e.g., French: Jakubowicz et al. 1998).

⁶ This experimental paradigm is also known as picture verification task (Conroy et al. 2009).

⁷ On the reflexive conditions, adults gave target responses 100% of the time (control and test conditions), while the corresponding percentages for the child group were 96% and 99% respectively.

FIGURE 3 Percentage of target responses on each of the control and test conditions: clitic (CL) and strong pronoun (PRON). Child group.

FIGURE 4 Percentage of target responses on each of the control and test conditions: clitic (CL) and strong pronoun (PRON). Adult group.

⁸ Responses on the reflexive condition were at ceiling (95% target). In the adult group, the percentage for reflexives was 100%.

⁹ We are grateful to the anonymous reviewers for their helpful suggestions regarding task differences.

¹⁰ We thank an anonymous reviewer for his input regarding the nature of the mechanisms in question.