2018 Sat Session C 1130

Saturday, November 3, 2018 | Session C, Terrace Louge | 11:30am

Reflexives and Word Order in Adult and Child Tagalog
I. Bondoc, K. Deen, E. Or

Tagalog, a symmetrical-voice language, has two canonical transitive patterns (Agent Voice, AV and Patient Voice, PV), both exhibiting variable word order. We explore the role of voice, case, and word order in the interpretation of Tagalog reflexive sentences by Tagalog-speaking adults and children, finding that (i) adults prefer patient-voice (PV) verbs with non-pivot-marked reflexives, (ii) children show the same (weaker) preference, though in agent voice (AV), they prefer the agent-reflexive word order. Our results show that children’s preferences are an attempt to reconcile (i) the surface linear order of arguments, (ii) the underlying structural positions of arguments in each voice, and (iii) Principle A of the binding theory.

The voice system in Tagalog picks out one argument as the pivot (typically marked with ang). Ang-marked nominals are syntactically privileged, allowing extraction of various kinds. In AV, the agent is the pivot; the patient is marked as a non-pivot, typically with nang. In PV, on the other hand, the patient is ang-marked (and thus privileged), and the agent is nang-marked, and thus not privileged. A priori, in sentences involving reflexives, it is not clear whether the reflexive must be ang-marked or nang-marked, nor whether this changes based upon the voice of the verb. We investigate this issue with adults and children in two experiments testing the 8 patterns in (1). Patterns marked ungrammatical were judged so by the first author, but these judgments have never been experimentally verified. These patterns cross voice (PV/AV), word order (lexical agent- reflexive and reflexive-lexical agent) and case marking (pivot/non-pivot). Note, in principle, all these patterns may be possible.

Experiment 1: Sentence Completion Task. 28 Tagalog-speaking adults and 15 children (3;10- 6;10) saw pictures depicting reflexive actions (e.g., a girl spraying herself with water), heard the verb (5 in AV; 5 in PV, counterbalanced, semi-randomized; 10 control non-reflexive transitive pictures) and completed the sentence. The results (Table 1) reveal: (i) adults rarely produce ungrammatical patterns, vastly preferring (a), (b), (e), and (f); (ii) adults and children strongly prefer verb-agent-reflexive order; (iii) children produce some errors, though none of the ungrammatical forms in (1).

Experiment 2: Felicity Judgment Task. 24 adults and 11 children (3;10-6;10) saw pictures depicting reflexive actions, heard two sentences and selected the better match for the picture.

Results (Table 2):

• Adults and children preferred grammatical sentences (a>c, e>h).
• Adults prefer verb-agent-reflexive order, but only in PV; also a strong preference for PV
• Children prefer verb-agent-reflexive regardless of voice. No preference for PV or

In sum, Tagalog adults strongly prefer (e): PV with lexical-agent before reflexive. Children show no preference for either voice, but in both voices, prefer the lexical-agent preceding the reflexive argument. Moreover, the case patterns that children produce/prefer reveal that the reflexive is structurally lower than the agent. These two facts together indicate an isomorphic preference that aligns the surface linear order with the underlying position of nominals (agent structurally c- commanding patient), revealing a faithfulness to Principle A, in what is otherwise a remarkably complicated set of available patterns.