Day 2 - November 8

The syntactic characteristics, such as word order, play a crucial role in determining the complexity of a sentence. In languages where the subject (S) typically precedes the object (O), there is substantial evidence showing a preference for SO word order over OS word order. This study explores how this SO preference persists, even in languages with significantly different typological features, like Truku, an Austronesian language. In Truku, the Verb-Object-Subject (VOS) word order is the canonical and syntactically basic structure, while the Subject-Verb-Object (SVO) order is derived and syntactically more complex. Investigating word order preferences in Truku is essential as it helps to assess whether these widely observed processing preferences are rooted in the linguistic system or stem from broader cognitive properties. The symmetrical voice system in Truku serves as an ideal context to distinguish the effects of grammatical functions from thematic roles. The syntactic complexity hypothesis suggests that the derived SVO word order in Truku should incur a higher processing cost, while the saliency hypothesis predicts a preference for word orders where the agent precedes the theme. Our auditory comprehension experiment revealed that native speakers of Truku favored the OS word order. This finding suggests that the frequently observed SO preference is not a universal linguistic feature. Additionally, the absence of a strong agent-before-theme preference indicates a possible relationship between a language’s voice system and the significance of saliency factors.

Most of our knowledge on children’s language acquisition comes from studies on only a few of the world’s languages (Kidd & Garcia, 2022). As this reliance on a small, unrepresentative sample of languages skews theoretical development, conducting research on understudied languages is an important priority. In this talk, I will focus on the acquisition and processing of Tagalog’s symmetrical voice system, a typologically rare feature where there is more than one basic transitive construction (i.e., equally marked by voice morphology, without any demotion of an argument), providing an important testing ground for accounts of thematic role assignment and online sentence parsing. I will present the results of our corpus analyses and field experiments; which support experience-based accounts of language acquisition. 

It has been suggested (yet debated) that regularly inflected words are decomposed into roots and affixes while irregularly inflected words are stored in the lexicon as a whole (Pinker, 1999, among others). Japanese verbal inflectional suffixes seem to be productive and verbs are said to be quite regular except for a couple of truly irregular verbs (McCawley, 1968). Despite these observations, Vance (1987, 1991), Klafehn (2003) failed to provide evidence for the regularity in off-line production-type tasks. We conducted a masked priming experiment to examine whether or not the processing of inflected verbs in Japanese exhibits the effects of decomposition. Contrary to what has been suggested, all three types of regular verbs showed priming effects when the prime was truly grammatically inflected form. We discuss their implications concerning the kinds of morphemes attached to these verbs. 

A wealth of {psycho/neuro}linguistic evidence has shown that recognition of morphologically complex words involves an early, form-based decomposition procedure, where words like teacher are segmented into smaller meaningful units {teach}+{-er} called morphemes. However, the field currently faces a serious limitation—a great majority of what we know is based on English morphology and related Indo-European languages (Cayado & Rastle, in prep), which limits our current understanding of how early, form-based decomposition takes place and what linguistic factors affect this procedure. Existing models of morphological processing variously assume that positional information, phonological/orthographic changes, phonological variability, and consistency of orthographic/phonological forms of morphemes play a critical role during this decomposition stage. In this talk, I will present a series of behavioural and magnetoencephalography (MEG) experiments that systematically test how these linguistic properties affect morphological decomposition. First, I exploit Tagalog infixation and pseudo-infixation to investigate the role of positional information of affixes in morphological processing (Cayado et al., 2023; in prep). Second, I also take advantage of morphophonological changes and variability in Tagalog [that is, nasal assimilation and substitution] to test how morphophonological changes that obscure the boundary between prefix and stem, as well as the variability of their application, modulate the way morphologically complex words are decomposed (Cayado et al., 2024, 2025). Third, I also use reduplication in Tagalog to test how mismatch in orthographic and phonological forms affect morpheme recognition and activation (Wray et al., 2022; Cayado et al., in prep). Finally, I will discuss how data from Tagalog challenge existing models of visual word recognition.

Formosan languages are quite diverse regarding the morphosyntactic and semantic properties of temporal-related markers. In this talk, we will present results of acceptability judgment experiments on the phenomenon of ‘temporal concord’ from three Formosan languages: Paiwan, and Truku Seediq. These three languages were selected based on the different temporal markers they exhibit: (a) perfective aspect and irrealis mood preverbal clitics for Paiwan, and (b) perfective aspect and irrealis mood auxiliary verbs for Truku Seediq. The results from these two experiments will allow us to propose a preliminary typology of the temporal concord processing of Formosan languages taking into account morphosyntactic and semantic considerations and in the light of experiments conducted on other languages.

Acceptability judgment studies focusing on temporal concord are still sparse, and the experiments focusing on aspect marking only rely on a very small number of languages, thus raising the question of the representativity of the results. In this study, we investigate the temporal concord phenomenon through web-based acceptability judgment experiments with two markers in Tagalog: (a) perfective aspect, related to past time interpretation, and (b) contemplative/perspective aspect, and we further examine task-related effects. Finally, we will compare the patterns of results from our Tagalog experiments with the ones form other languages, and especially Formosan languages.