Final Report Summary - DEV LANG COMPRHNSN (Development of spoken language comprehension processes in a verb-final language: Incremental interpretation of case marking cues in Turkish speaking children)
Since early 1960s, cognitive scientists have explored how adults interpret spoken sentences as they hear the utterances unfold in time. This work has produced a rich and coherent description of the moment-to-moment processes that are involved in converting an acoustic signal into a conceptual representation of sentence meaning. In contrast, there is relatively little work on how this system develops. We know that, by four or five years of age, children appear to effortlessly understand most of what we say to them. But, until recently we have not known much about how they do this, largely because the techniques that are used to study language comprehension in adults are not suitable for use in young children. In the past fifteen years, new techniques have been developed that allow us examine moment-to-moment (i.e. online) comprehension in young children. Visual world eye-tracking is one of such techniques that allows moment-by-moment analysis of spoken language understanding. The linking assumption underlying this methodology is that the position of the eye provides an index for the level of attention, which is in a dynamic relationship with the visual context and spoken linguistic input.
The technique is well suited for testing children because it has high temporal resolution and it is very simple to implement. It takes a spontaneous behaviour (i.e. fixations, saccades, or pupil size) as its measure so it does not require the participant to perform a secondary task. The present project has been the first study to use the visual world eye-tracking paradigm with Turkish-speaking children.
Our aim in this project was to test whether children can interpret morphosyntactic cues incrementally. This method allows us to detect the point at which a particular cue is interpreted by using natural sentences, without recourse to extra tactics needed for other paradigms like replacing the verb with a non-sense word.
In this project we conducted a series of experimental studies that investigate how a young child acquiring a verb-final language (e.g. Turkish or German) interpret spoken utterances involving various structures. In the first three experiments, we explored whether 4-year-old children acquiring Turkish or German are able to interpret case marking cues (nominative vs. accusative) in spoken utterances that locate the verb as the sentence-final element (Subject Object Verb vs. Object Subject Verb). Previous studies have shown that children cannot interpret case markers in object-initial sentences until late in childhood. This was especially confirmed in many studies with German speaking children until who could not interpret case independent of the verb until age 4 and independent of word order until age 6 or 7. This has been related to late abstraction of conceptual roles (Agent/Doer, Patient/Undergoer) and their mapping onto grammatical roles (Subject/Object) (Tomasello, 2000) or late maturation of related brain regions (Friederici, 2012). Our studies have revealed for the first time that both Turkish-speaking and German-speaking 4-year-olds are able to interpret case markers independent of the verb or word order information. Our studies are crucial in showing that children are already adult-like at age 4 in terms of incremental and predictive interpretation of case markers for thematic role assignment.
We conducted three more experiments with Turkish-speaking children to address whether they are able to interpret the contrast between the accusative and the ablative case, which is a less reliable case marker. More specifically, we aimed to find out whether preschoolers are able to interpret the meaning restrictions imposed by the accusative and the ablative case on the part-whole interpretation. Ablative case marks partitivity in consumption verb in Turkish (I ate cake-ABL: I ate from the cake) whereas the accusative case indicates that the whole is consumed (I ate cake-ACC: I ate up the cake). In the first of these experiments, we found that children do not assign partitive meaning to the ablative case. In two follow-up experiments, we discovered that children are more likely to assign a partitive interpretation to the ablative case when their attention is attracted to the quantity in the consumption or to the part-whole relationship with a priming task explicitly testing the quantifiers 'some' vs. 'all'.
In the next two studies, we explored whether children interpret verbal morphemes incrementally, independent of the identity of the nouns and independent of the case marking cues by young children. We tested the interpretation of relativizing morphemes (subject and object relativizing morpheme) and passivizing morpheme. These studies revealed for the first time that Turkish-speaking preschoolers are able to interpret verbal morphemes even when there are no extra cues provided by the nouns or the case markers. This finding is crucial because verbal morphemes are considered even less reliable in verb-final languages, as they tend to appear sentence-finally. According to early-cue-first perspectives, children rely on early appearing cues and ignore the late-coming ones during the course of spoken language interpretation, which is claimed to be a pattern that results in late acquisition of verbal morphemes in verb-final languages that locate the verbal morphemes in sentence-final positions (Lidz et al., 2003; Trueswell, et al., 2012). Our study has been the first study to challenge this perspective.
The research questions addressed in this project are at the intersection of multiple disciplines. The research questions here are informed by theoretical linguistics, computational linguistics, and cognitive science (e.g. ‘What is the role of the verb and morphosyntactic markers in the interpretation of a sentence?’, ‘Do typologically distinct languages follow different processing patterns, or is there a universal parsing mechanism?’) The findings usually have implications for developmental psychology, early childhood education and clinical linguistics (e.g. ‘When do children become adult like in their interpretation of an utterance?’, ‘How does speed of language processing influence learning?’, ‘What does typical language processing look like?’). Furthermore, computational fields like machine translation, speech-to-text conversion, and human-machine interaction highly benefit from studies investigating the nature of spoken language processing across languages.
The present project is innovative in several respects. First, it employs an emerging paradigm for studying children’s language comprehension. This is the first time child language processing is investigated using the visual-world paradigm in Turkish. Second, the vast majority of the work to date has been conducted in English. To date, there are only two published studies that have looked at the development of sentence processing in verb final case marked language (Choi &Trueswell, 2010; Özge, Marinis, & Zeyrek, 2015). The present research contributes to our understanding of how syntactic and semantic processing develops in verb-final languages. Third, the present project addresses core and unresolved theoretical issues in language development. For example, there are still open questions as to the types of information used during online language comprehension, whether the same information sources are employed across languages and how these sources are acquired. The present research explores these issues and challenges three long-standing acquisition hypotheses (mentioned above).
The present research increases our knowledge of how the spoken language interpretation abilities develop in head-final languages. The findings reported in this project have crucial implications for other European languages with verb-final features (e.g. German or Dutch) and with rich case system (Serbian-Croatian, Finnish, Hungarian, or German). The present findings about monolingual language acquisition and processing offer a solid baseline to understand how bilingual acquisition or acquisition/language processing in impaired populations differ from typical norms. Understanding how language comprehension typically develops in typically developing monolingual children acquiring a verb-final language could inform the development of tools for identifying children who have language delays. This leads to more effective therapy and education; hence contribute to the economic and social well being of the family and the society both at monolingual and multilingual communities. Hence, the findings from the present project could be of use to cognitive scientists, theoretical linguists, computational linguists, developmental psychologists, educators, educational therapists, clinical scientists, parents, and policy makers.
The technique is well suited for testing children because it has high temporal resolution and it is very simple to implement. It takes a spontaneous behaviour (i.e. fixations, saccades, or pupil size) as its measure so it does not require the participant to perform a secondary task. The present project has been the first study to use the visual world eye-tracking paradigm with Turkish-speaking children.
Our aim in this project was to test whether children can interpret morphosyntactic cues incrementally. This method allows us to detect the point at which a particular cue is interpreted by using natural sentences, without recourse to extra tactics needed for other paradigms like replacing the verb with a non-sense word.
In this project we conducted a series of experimental studies that investigate how a young child acquiring a verb-final language (e.g. Turkish or German) interpret spoken utterances involving various structures. In the first three experiments, we explored whether 4-year-old children acquiring Turkish or German are able to interpret case marking cues (nominative vs. accusative) in spoken utterances that locate the verb as the sentence-final element (Subject Object Verb vs. Object Subject Verb). Previous studies have shown that children cannot interpret case markers in object-initial sentences until late in childhood. This was especially confirmed in many studies with German speaking children until who could not interpret case independent of the verb until age 4 and independent of word order until age 6 or 7. This has been related to late abstraction of conceptual roles (Agent/Doer, Patient/Undergoer) and their mapping onto grammatical roles (Subject/Object) (Tomasello, 2000) or late maturation of related brain regions (Friederici, 2012). Our studies have revealed for the first time that both Turkish-speaking and German-speaking 4-year-olds are able to interpret case markers independent of the verb or word order information. Our studies are crucial in showing that children are already adult-like at age 4 in terms of incremental and predictive interpretation of case markers for thematic role assignment.
We conducted three more experiments with Turkish-speaking children to address whether they are able to interpret the contrast between the accusative and the ablative case, which is a less reliable case marker. More specifically, we aimed to find out whether preschoolers are able to interpret the meaning restrictions imposed by the accusative and the ablative case on the part-whole interpretation. Ablative case marks partitivity in consumption verb in Turkish (I ate cake-ABL: I ate from the cake) whereas the accusative case indicates that the whole is consumed (I ate cake-ACC: I ate up the cake). In the first of these experiments, we found that children do not assign partitive meaning to the ablative case. In two follow-up experiments, we discovered that children are more likely to assign a partitive interpretation to the ablative case when their attention is attracted to the quantity in the consumption or to the part-whole relationship with a priming task explicitly testing the quantifiers 'some' vs. 'all'.
In the next two studies, we explored whether children interpret verbal morphemes incrementally, independent of the identity of the nouns and independent of the case marking cues by young children. We tested the interpretation of relativizing morphemes (subject and object relativizing morpheme) and passivizing morpheme. These studies revealed for the first time that Turkish-speaking preschoolers are able to interpret verbal morphemes even when there are no extra cues provided by the nouns or the case markers. This finding is crucial because verbal morphemes are considered even less reliable in verb-final languages, as they tend to appear sentence-finally. According to early-cue-first perspectives, children rely on early appearing cues and ignore the late-coming ones during the course of spoken language interpretation, which is claimed to be a pattern that results in late acquisition of verbal morphemes in verb-final languages that locate the verbal morphemes in sentence-final positions (Lidz et al., 2003; Trueswell, et al., 2012). Our study has been the first study to challenge this perspective.
The research questions addressed in this project are at the intersection of multiple disciplines. The research questions here are informed by theoretical linguistics, computational linguistics, and cognitive science (e.g. ‘What is the role of the verb and morphosyntactic markers in the interpretation of a sentence?’, ‘Do typologically distinct languages follow different processing patterns, or is there a universal parsing mechanism?’) The findings usually have implications for developmental psychology, early childhood education and clinical linguistics (e.g. ‘When do children become adult like in their interpretation of an utterance?’, ‘How does speed of language processing influence learning?’, ‘What does typical language processing look like?’). Furthermore, computational fields like machine translation, speech-to-text conversion, and human-machine interaction highly benefit from studies investigating the nature of spoken language processing across languages.
The present project is innovative in several respects. First, it employs an emerging paradigm for studying children’s language comprehension. This is the first time child language processing is investigated using the visual-world paradigm in Turkish. Second, the vast majority of the work to date has been conducted in English. To date, there are only two published studies that have looked at the development of sentence processing in verb final case marked language (Choi &Trueswell, 2010; Özge, Marinis, & Zeyrek, 2015). The present research contributes to our understanding of how syntactic and semantic processing develops in verb-final languages. Third, the present project addresses core and unresolved theoretical issues in language development. For example, there are still open questions as to the types of information used during online language comprehension, whether the same information sources are employed across languages and how these sources are acquired. The present research explores these issues and challenges three long-standing acquisition hypotheses (mentioned above).
The present research increases our knowledge of how the spoken language interpretation abilities develop in head-final languages. The findings reported in this project have crucial implications for other European languages with verb-final features (e.g. German or Dutch) and with rich case system (Serbian-Croatian, Finnish, Hungarian, or German). The present findings about monolingual language acquisition and processing offer a solid baseline to understand how bilingual acquisition or acquisition/language processing in impaired populations differ from typical norms. Understanding how language comprehension typically develops in typically developing monolingual children acquiring a verb-final language could inform the development of tools for identifying children who have language delays. This leads to more effective therapy and education; hence contribute to the economic and social well being of the family and the society both at monolingual and multilingual communities. Hence, the findings from the present project could be of use to cognitive scientists, theoretical linguists, computational linguists, developmental psychologists, educators, educational therapists, clinical scientists, parents, and policy makers.