Session Detail

In developing the fifth core knowledge about social partners, race is an important factor biasing children to form social affiliations. The present study explored the development of the race-based social preferences in 3- to 8-year-old Taiwanese children. In Experiment 1, twenty-three 3-4 year-old children viewed three simultaneously presented video clips modeled by a Taiwanese (own race, in-group, high social status), a Southeast Asian (near race, out-group, low social status) and a Caucasian (other race, out-group, high social status) young female smiling at them. Children were instructed to give a toy to their most preferred and their second preferred individuals. Results showed that children preferred to give toys to the Taiwanese actress the most (50%), and there is no difference among the three ethnicities for the least preference. In Experiment 2, twenty-two 5-6 year-old children viewed the same videos and were instructed to choose their most preferred and their second preferred persons as friends. The 5-6 year-olds preferred the Taiwanese actress the most (69%), while both the Caucasian (54%) and the Southeast Asian actresses (43%) were the least preferred. In Experiment 3, twenty-one 7-8 year-old children performed the same task as in Exp.2 and they preferred to choose the Caucasian actress (57%) as friend the most, followed by the Taiwanese actress (38%), and the Southeast Asian actress was the least preferred choice. Combining the results across the three experiments, we found that children preferred the Taiwanese actress the most (53%) and the Southeast Asian actress the least (13%). In sum, our findings suggest that a rudimentary race-based social preference (or prejudice) seems to emerge in the later part of preschool years. These results provide a cross-cultural exploration of when and how Taiwanese children’s social judgments may be influenced by race.

 

The author has been researching narrative generation, especially design and development of an “Integrated Narrative Generation System (INGS)”, which generate story, representation and the other narrative structures. This research is based on the concept of “multiple narrative structures”, which treats narrative generation as the synthesis of micro and macro levels. The INGS is for the micro or personal generation. The macro level is performed through a “Geino Information System (GIS)” by the author. The “geino” means performing arts that includes traditional genres such as kabuki in addition to contemporary genres. The GIS has several INGS systems to circularly produce various types of narratives including scenarios, actual performances and life histories of performers. An important idea is that the GIS multiply stores diverse narrative knowledge to generate diverse narratives dependent on their combinatorial or multiple uses. For example, many narrative stories are created by a kind of inter-textual combination of other stories and episodes. Although the INGS is implemented in an organized system, the GIS is the stage of conceptual design. A current object is to design and develop the GIS. An approach relies on studying kabuki, which is a synthesized drama that incorporates a variety of geino genres, such as dance, noh-kyogen, and ningyo-joururi, into a rich form. The author has analyzed the multiple structures of kabuki regarding fifteen elements including person and story. Knowledge on the analyzed multiplicity in kabuki will be introduced into the INGS and GIS. In this paper, the author analyzes the multiple structures of a “story” using kabuki works based on Inumaru (2005), who is a kabuki researcher, in the framework of the GIS including the INGSs. In particular, the author analyzes a kabuki story based on the relationship with other stories, genres and characters to design the multiple narrative generation mechanism using the systems.

 

An “Integrated Narrative Generation System (INGS)” by the authors synthesizes various mechanisms and knowledge to generate narrative conceptual structures and representations. A basic unit of the narrative conceptual structures is a case structure for an event that has a verb concept and noun concepts provided by the conceptual dictionaries. Various levels of concepts, such as difficult and easy concepts to understand, are mixed in a semantic category in the dictionaries. Therefore, the various levels are frequently mixed in an event through a story generation process. Although a method to solve it is to revise and detail the categorization, this is a difficult semantic problem. The authors presented another method using appearance frequency and co-occurrence of concepts to control the selection in event generation. In a previous research, the authors confirmed that the number of appearance frequency and the degree of comprehensibility are basically proportion. Further, the authors controlled the narrative easiness by the frequency and get concepts having a same degree of comprehensibility. These mechanisms were partially incorporated into the INGS. In addition, each concept uses a corresponding word description in the narrative representation phase. However, diversity of word notation is provided by the word notation dictionaries. On the other hand, the study of narrative generation from a haiku is a recent issue of the authors. This paper conversely aims to generate haikus using the appearance frequency and co-occurrence of concepts and words. A haiku has three parts. This method basically adjusts selecting a concept in each part using the appearance frequency and deciding a concept sequence from a concept to the next concept according to the co-occurrence. For example, using low frequency and co-occurrence may results in the generation of difficult haikus to understand. This study also shows an evaluation for the diverse combinatorial haiku generation.

 

This paper analyzes the relationship between participation and learning represented in ethnographic case studies of ten informants aged 23-59 participating in a common-based peer production site, the FabLab Kamakura community. Digital-based personal fabrication is a new wave culture of mavens, who are devoted to alternatives to mass production, and are on a mission “to make (almost) anything”. FabLab Kamakura is a valuable venue for exchanging information about, for example, digital tools, Arduino, crafts, textiles, and so on. First we frame this work as an effort to think about their participation and learning using the concept of “wildfire activity theory” (Engeström, 2009) and “legitimate peripheral participation (LPP)” from Lave and Wenger (1991). Then we argue an overview of FabLab culture in Japan and at FabLab Kamakura. Using SCAT methodology (Otani, 2011), we group our findings in two different categories: (1) learning through participation in FabLab Kamakura, (2) the visualization of weak ties and mobility through participation in wildfire activities. We conclude that participants at FabLab Kamakura are producing and designing available artifacts for their lives and works, and in doing so, what they are designing is the physical manifestation of their very thoughts.

 

Whether it is sentencing in the courtroom or our moral judgments toward personal behaviors, research has suggested that the intention of the agent is a critical factor. Most of us do not need to, and do not feel that there is a need to argue why unintended harm is more forgivable than deliberate harm, or why attempted murder deserves more severe punishment than unforeseen accidental deaths. The central role of intention, however, was established solely based on evidence from the harm and care domain, and has recently been questioned whether it is as critical in other moral domains. For example, Young & Saxe (2011) found that when it comes to the purity domain (e.g., sexual behaviors), outcome trumps intention in predicting participants’ moral judgments. Here we used a similar set of moral judgment questionnaires on Taiwanese participants to investigate whether there might be a cultural variation in the way people make moral judgments in the harm and purity domains. Our data showed a main effect of intention and outcome, as well as an interaction between moral domain and intention, where intention weighs more heavily in the harm than the purity domain. However, we did not observe an interaction between outcome and moral domain, therefore outcome was not a particularly powerful predictor over intention in the making of purity-related moral judgments in our Taiwanese sample. Together, our results suggest a similarly-weighted role of intention across two cultures, and a cultural distinction in the relationship between outcome and purity-violation events. Implications on the disgust-hypothesis of moral judgment will also be discussed.

 

The trolley dilemma and its many variations have been used as a tool for investigating people’s moral judgment. People are usually willing to pull the switch and sacrifice one person’s life in order to save five others, but such inclination decreases dramatically when the one person has to be physically pushed down the bridge by the main character (i.e., personal force) in order to stop the train (i.e., victim’s body as a means). This combination between personal force and means-to-an-end has been reported to evoke more “immoral” judgments than when each factor is presented alone, an interaction effect to which Greene and colleagues (2009) suggest is necessary in triggering people’s moral alarm. To test this claim, here we used multiple scenarios that have equally dissociable elements of personal force and means-to-an-end in 300 participants. Our results do not support the notion of a magic interaction between personal force and means-to-an-end. Rather, our data strongly support a central role for intention in formulating moral judgments, which is consistent with many previous studies (Mikhail, 2000., Cushman et al., 2006; Hauser et al., 2007). The effect of personal force, on the other hand, varies across scenarios and therefore likely plays a less dominant role than intention. We did not observe an interaction between the two. Together, we conclude that intention alone is self-sufficient in triggering the moral alarm, whereas other moral factors such as personal force may be more context-dependent between scenarios or even cultures.

 

Recently, “programing learning” is considered to be very effective to understand causal relations and problem structures and they have the possibility to generate various learning/education outcomes. This study aimed to inspect effects of the programming learning inserted into the context of science education curriculum for elementary school students. 24 children in the 6th grade participated the programing learning exercises to simulate planetary motion as an advanced program in learning an astronomical body and a constellation. Comparison between the results of pre/post questionnaire surveys made clear the followings: (i) the programming learning was effective in the understanding of the solar system, (ii) the number of children who think the programming to be seemed useful increased after the programming learning, and (iii) the children who had a negative attitude to the programming improved their attitude after they had experience of the programming learning exercise. It was found that the programming learning is a methodology that brings effective learning/educative effects even for the elementary school students.

 

This paper analyzes the relationship between agencies and artifacts represented in ethnographic case studies of ten female informants aged 20-25 participating in the cosplay community. Cosplay is a female-dominated niche subculture of extreme fans and mavens, who are devoted to dressing up as characters from manga, games, and anime. “Cosplayers” are highly conscious of quality standards for costumes, makeup, and accessories. Cosplay events and dedicated SNSs for cosplayers are a valuable venue for exchanging information about costume making. First we frame this work as an effort to think about their agencies using the concept of hybrid collective and activity theory. Then we share an overview of cosplay culture in Japan and our methodologies based on interviews and fieldwork. Using SCAT (Otani, 2011) methodology, we group our findings in two different categories: (1) Cosplayers’ agencies and relationships with others mediated by usage of particular artifacts, (2) Cosplayers agencies visualized through socio-artificial scaffolding and collective achievement. We conclude that cosplayers are producing and standardizing available artifacts for their cosplay objects, and in doing so, they are designing their agencies. We consider that the activities like them are one appearance we can observe in the other our mundane communities not apply only to cosplay one. Not only to cosplay, however we consider that these kinds of activities apply to other mundane communities.

 

In many advanced countries, housing is a property that passed down from generation to generation. However, the average life span of house in Japan is about 30 years, which is shorter than the other countries. Most of houses have no big problem and trouble. 　Since 2009, the spread in long-term high-quality housing is promoted in Japan. The technical development about the structure and equipment is remarkable on the construction field. On the other hand, the housing that have no big problem on structure and equipment is demolished usually in Japan. A building is not always recognized as better long-term assets than a piece of land. In the long run, it is not invested in maintenance in housing sufficiently. Even if housing has a high level technical structure and equipment, the resident of next generation may dislike them, they are destroyed in the near future. The study analyzed about the element of charm in a housing is important to support longevity of housing. 　The purpose of this study is to reveal the physical and psychological element of house that continue to be attractive for a long period of time. We selected NAGANO City for a target place of research, residential area had house built at various age. 　For those who studies architecture, the older houses (traditional housing) are preferred. On the other hand, regardless of the major, the older houses were estimated important and valuable. They are invaluable to maintain something of value at NAGANO City. 　Needless to say, new housing that has a high level technical structure and equipment should be built. We should grasp traditional housing in an area to pass down from generation to generation. Finally, we had better understand the value of area, and attempt to maintain the elements of traditional housing.

 

Understanding how language is acquired by infants has remained to be a challenging task. Previous attempts have all achieved remarkable results that shed light on future directions in the research of infant language acquisition. However, the dynamics of infant language acquisition is a complex process. This process requires the understanding of innate learning mechanisms within infants (which may be responsible for the actions of imitation and social interaction with their surroundings) and the support for development of consciousness. Following Gao et al. (2013; 2008)’s statements of levels of consciousness for language development as theoretical guidelines, our study examines the mechanisms that generate behaviors at different levels of consciousness and their relations to prominent transitions in infant development. The objective of our study is to apply agent-based modeling (Holland, 1995) in exploring the topic of language development in early infants. The theoretical framework of agent-based model (ABM) proposed by Holland (1995) creates a flexible abstraction of the real world and provides an approach in the general study of complex adaptive systems (CAS). ABMs consist of basic computer algorithm units (agents) which are the central modeling focus points. These agents can be either modular or self-contained. An agent is an identifiable, discrete individual with a set of characteristics or attributes, behaviors, and decision-making capability. In this paper, our emphasis and discussions are on the process of language acquisition for infants under the first level of consciousness, and on the expansions of subsequent agent-based models. Results from simulations show that our approach on using agent-based modeling is fair in reflecting the development of pre-linguistic capabilities for infants with minimal consciousness.

 

Infants typically start to produce words that can be perceived as “meaningful” (for instance “ma” or “mama”) at the 12th month period from birth. This feature of “meaningful first words” is seen as a significant phenomenon in infant language acquisition. This shows that an infant has managed a transition from the first level of consciousness (LoC 1, Minimal Consciousness) to the second level of consciousness (LoC 2, Recursive Consciousness). Most importantly, it also signifies that the infant has thorough development within LoC 2. At this level of consciounsess, infants have developed the capacity to engage informational contents within their memory. Nearing to the 12th month period, infants come about speaking their first words. This emergence of first words is a significant achievement in infant development as this implies that an infant has a thorough development through the stage of LoC 2. Furthermore, this phenomenon sets the basis for infants in learning language. In this paper, we follow Gao et al. (2013; 2008)’s statements of levels of consciousness for language development in discussing the behavioral process and mechanism that could be crucial within LoC 2. In this paper, we aim to focus on the underlying mechanisms within LoC 2 that could allow an infant to make sense of information, whereby leading them to speak their first words.

 

The question of whether children with reading problems have deficits in processing musical rhythms has remained unanswered. However, speech prosody may be one of the most important indicators predicting perception of speech phonemes and the sensitivity of the musical rhythm could contribute to the perception of speech prosody. Also, language and music might share brain mechanisms that could be related to basic auditory processing functions. Even though the debate on the association between music and reading is still controversial, interestingly, the relationship between rhythmic signals and phonological cues might be a factor affecting individual’s reading abilities through processing of linguistic prosody. 16 typically developmental children and 16 children with developmental dyslexia were recruited with informed consent in this magnetoencephalographic (MEG) study. Oddball paradigm was adapted for testing the brain responses that were activated by the omissions of the strong or weak beats in the regular rhythmic patterns. The result showed that mismatch negativity (MMNm) and P3a(m) were generated via the irregular patterns of omitted strong beats; however, this effect was only found in the typically reading group but not in the dyslexic group. For between group comparison, there was no significant differential effect in regular rhythms (without omissions), rhythm violations by omitting weak beats, or control rhythms, but there was a significant group effect in irregular patterns produced by omitting strong beats. The results showed that the brain responses to detect irregular rhythmic pattern were larger in typically developing children than in dyslexic children. This study emphasizes the relationship between language and music perception that is in line with the previous studies. Furthermore, at the level of event-related fields, dyslexic children did not differentiate the different rhythmic patterns from each other. This study has showed that dyslexic children have differential neural processing of rhythms that could be one of the underlying factors in dysfluent in reading development.

 

Cognitive development is a trajectory shaped by interactivity between architectural constraints and environmental experience of the developing brain (Westermann et al. 2007). In the domain of language processing, early experience has a long-term impact on later processing. One example of this cognitive footprint is the well-documented phenomenon of age of acquisition (AoA). In lexical processing, early-learned words tend to be processed more quickly and accurately relative to later-learned words (e.g. Juhasz, 2005). Two theoretical explanations emphasising different aspects of developmental changes have been proposed. The representation theory (e.g., Brysbaert & Ghyselinck, 2006) argues that the AoA effect could originate from differences in semantic representations where early-learned words have richer semantic representations. Alternatively, the mapping theory (e.g., Ellis & Lambon Ralph, 2000) proposes that gradual reduction in plasticity for learning mappings between representations over the time course of learning is the key to accounting for the AoA effect. Recently, an emerging view has considered that the two theories may both contribute to the AoA effect. However, the relative contribution of representations and of mapping remains unclear. To explore this, we developed a triangle model of reading including a realistic, cumulative exposure to words during learning to read (Monaghan & Ellis, 2010). Regression analyses on the model’s reading comprehension performance showed that AoA was a reliable predictor. There was a significant interaction between AoA and concreteness, suggesting that AoA operates differentially on concrete and abstract words. Additional analyses of the locus of the effects in the model revealed that the concreteness effect was related to semantic properties of representations, but that AoA was related more to the mappings than to the representations. The model supports the view that changes in both plasticity and representation contribute to the emergence of the AoA effect.

 

Speech sound acquisition is fundamental to spoken language acquisition and is a subject of theoretical debates. The traditional perspectives (e.g., behaviorist theories, scaffolding theory, connectionist view, and dynamic systems theory) suggest that simple input is more important for speech sound acquisition, whereas complexity-based linguistic theories postulate that complex input (Gierut et al., 1987; Gierut et al., 1996) plays a more important role. In the current study, we test these competing sets of theories by comparing the effects of training based on simple and complex stimuli. Complexity is defined by the markedness hierarchy. In our training, the more marked pre-voiced dental-retroflex sounds were considered complex whereas the less marked voiceless dental-retroflex sounds were considered simple. Cantonese-speaking adult subjects were trained for five consecutive sessions on a pseudoword-picture identification task. In order to evaluate their improvement (on trained sounds) and generalization (to untrained sounds), subjects were tested on an AX discrimination task before and after the training. We found that the subjects who were trained on complex sounds (n=15) significantly improved on the perception of both complex and simple stimuli. On the other hand, subjects who were trained on simple sounds (n=15) only improved on perception of simple sounds and did not generalize to the untrained complex sounds. The current findings reveal that complex input induces system-wide changes in the phonological system while simple input leads to limited changes in the phonological system. Overall, our findings suggest that complex input plays a more important role in speech sound acquisition relative to simple input.

 

Human language understanding is conceived to consist of successive stages of lexical, syntactic and semantic processes. Pragmatic and common sense reasoning are then applied to get to the speaker's intended meaning. Rhetorical devices can be considered as a set of tools at the speaker's disposal to guide the latter pragmatic and common sense reasoning of the hearers. Structural inversion is often employed as a rhetorical device to produce a critical punchline with humorous effects when interpreting a text. Paul Krugman once wrote a paper, which concludes with a sentence: This paper is a serious analysis of a ridiculous subject, which is of course the opposite of what is usual in economics. The word 'opposite' triggers structural inversion to produce as implication a critical and humorous message. We present an analysis of structural inversion in terms of situation parsing, inversion and allusion.

 

Horn (2014) cites Hintikka's (1962) ambiguity of negating a 'know' clause between factive presupposition [p & ∼Kap] and non-factive (NF) [∼Kap], adding the NF nonveridical question example like ‘Do you know that he is reliable?’ in English. In contrast, we present a rather unexplored case of the two alternant uses of factive and NF in the positive verb al- 'know' in Korean (K). and similarly bil- in Turkish (T) and tudjia in Hungarian (H). The factive vs. NF distinction is crucially made by the different complement cases of factive ACC vs. NF oblique case (–uro) in K and factive ACC (i) vs. Reportative (diye) in T and similarly definite anaphor vs. oblique anaphor before 'know' in H. Japanese, however, has no NF 'know,' although it has the factive complement case of ACC. The NF (–uro) al- verb is different from other weaker epistemic verbs meaning ‘believe’/‘think’ in that even the NF one tends to require some piece of evidence for JTB (justified, true belief as defined in epistemology), as in T and H. But the evidential justification may turn out to fall short of knowledge, not being true. We conducted experiments to clearly show that the NF (–uro) al- has the relation of neg-raising between the high neg S and the low (complement) neg S, which are truthconditionally equivalent. It implies that this NF verb (–uro) al- is identical in neg-raisability with other weaker epistemic verbs meaning ‘believe’ and ‘think’ in Korean. T and H also reveal neg-raising, as hypothesized. An excerpt from Sejong Corpus indicates that the NF ‘know’ in Korean typically accompanies some piece of evidence that led the speaker to hold a firmer belief than other epistemic verbs meaning ‘believe’/‘think’ in Korean, which can be used with no such evidential justification. T and H also support this characterization of NF 'know' vs. other weaker epistemic attitude verbs. However, the newly discovered commonality of neg-raising between the NF 'know' and the belief type epistemic verbs in all these three languages is the most exciting result we arrived at.

 

In any linguistic context, the two languages of bilingual listeners are active and interact, such that lexical representations in both languages are activated by the spoken input with which they are compatible. Whereas words that overlap segmentally in the two languages compete for activation, it remains unclear whether suprasegmental information further modulates this cross-language competition. This study investigates the effect of stress placement on the processing of English-Spanish cognates by native Spanish speakers with some knowledge of English (in Spain) and intermediate-to-advanced English-speaking second-language learners of Spanish (in the US) using a visual-world eye-tracking experiment in Spanish. In each trial, participants saw a target (asado), one of two competitors (stress match: asados; stress mismatch: asador), and two unrelated distracters, and they heard the target word. Importantly, the experiment included a non-cognate condition (asado-asados-asador) and a cognate condition, where the stress pattern of the English word corresponding to the Spanish competitor in the stress-mismatch condition (inventor) instead matched that of the Spanish target (invento). Second-language proficiency and vocabulary, and inhibitory control were measured. Growth-curve analyses on competitor fixations reveal cognate-status and stress-mismatch effects for native Spanish speakers, indicating that stress constrains lexical access for these participants, similarly for both cognate conditions. For the Spanish learners, results show a greater effect of stress match in the non-cognate condition than in the cognate condition, and a very early advantage for cognate words in the stress match condition (and to less degree in the stress mismatch condition). Results of this study provide further support on the simultaneous activation of the two languages in the bilingual brain. There is clear evidence showing lexical stress can modulate the degree of cross-language activation that bilingual listeners experience.

 

Does the language we speak affect the way we think? This question has been discussed for a long time from different aspects. In this article, the issue is examined with an experiment on how speakers of different languages tend to do different sequencing when it comes to sizes of general objects. An essential difference between the usage of English and Mandarin is the way we sequence the sizes of places or objects. In English, when describing the location of something we may say, for example, ”The pen is inside the trashcan next to the tree at the park.” In Mandarin, however, we would say, “The pen is at the park next to the tree inside the trashcan.” It’s clear that generally English use the sequence from small to big while Mandarin the opposite. Therefore, the experiment is conducted to test if the difference of the languages affects the speakers’ ability to do the two different sequencing. Within the experiment, three nouns were showed as a group to the subjects. Before they saw the nouns, they would first get an instruction of “big to small”, “small to big”, or “repeat”. Therefore, the subjects had to sequence the following group of nouns as the instruction they get or simply repeat the nouns. After completing every sequencing and repetition in their minds, they push a button as reaction. As the result, the experiment shows that English native speakers react more quickly to the sequencing of “small to big”; on the other hand, Mandarin native speakers react more quickly to the sequence “big to small”. To conclude, this study may be of importance as a support of language relativism that the language we speak do shape the way we think.

 

Min is a common local substrate language in Taiwan that substantially molds the phonology of the official language Mandarin. Although many of the Min-affected phonetic variants are of fairly high frequency, our previous research shows that lexical retrieval is more efficient when words are presented in its canonical than its variant form, which directly contradicts the frequency effect commonly found in processing studies. As talker information and visual modes are found to be readily incorporated in speech perception, this study intends to investigate the effect of visual talker information on the perception and representation of some Min-dependent Mandarin phonetic variants in Taiwan, in particular, the deretroflexion rule of /tʂ tʂʰ ʂ/→[ts tsʰ s] and the nasal merger of /in/→[iŋ]. Participants will be first presented with face photographs prototypical of Min and non-Min speakers, and will later be asked to perform a lexical decision task on auditory targets phonetically congruent or incongruent with the facial information. It is hypothesized that lexical retrieval will be facilitated when auditory targets of phonetic forms that are congruent with the facial information is presented.

 

Recent statistical learning studies of English adults and infants suggest the plausibility of cross-situational word learning in much more complex situations with many words, many possible referents, highly ambiguous individual learning trials, and the statistical resolution of the ambiguities only through the accumulation and evaluation of information over many words-referents pairings and many trials. Our study aims to answer the questions: When can Chinese infants learn new words under uncertainty via cross-situational statistics? Does the strength of words-referents association influence on cross-situational word learning? In the study, during the training phase, on each trial, two word forms and two potential referents were presented with no information about which word went with which referent. Although words-referents pairings were ambiguous within individual trials, they were certain across trials. After training, infants were presented with a single word and two potential referents, the cross-trial correct referent and a foil. The longest looking difference (LLD) means the difference of longest looking time between the correct referent and the foil. If infants have calculated the statistics appropriately, LLD should be above 0 significantly after naming (naming effect). In experiment 1, 18 20-month-olds were taught 4 word-referent pairs. There were 12 trials for training. The strength of words-referents association was 6:2 (co-occurrence times of word-referent: co-occurrence times of word-foil)(low). There was no naming effect found. In experiment 2, 17 20-month-olds were taught 5 word-referent pairs. There were 20 trials for training. The strength of words-referents association was 8:2 (high). The naming effects were found. In experiment 3, 15 16-month-olds learned word-referent with the high strength of association. There was no naming effect. The results indicated that: 1) the strength of association of words-referents affects Chinese infants’ cross-situational word learning; 2) only 20-month-old Chinese infants can learn words via the high strength of association (8:2).

 

This research is related to an “Integrated Narrative Generation System (INGS)” by the authors, which automatically generates narrative conceptual structures (story and plot) and the representations. The INGS uses the knowledge of event sequences for generating a story. The authors call this type of event sequence a script. A script means a semantically organized events used for detailing an episode and extending a scene in a story by the corresponding story technique in the INGS. In the previous development, scripts have been stored by hand. However, the INGS currently needs many scripts for diverse and flexible story generation. Firstly, a semi-automatic script composition tool has been presented to easily define and store in the script knowledge base in the INGS. Using this tool, the authors define and store 760 scripts, which include events from 3 to 10, based on the description of event sequences inputted by users. The next method experimented by the author is script acquisition from many novels. In the first step, pairs of two verbs were automatically extracted using bigram. For each verb in the pairs, the authors define a case structure referring to an original sentence, in which the verb is included, using the semi-automatic script composition tool. In this paper, the authors aim to automatically insert the values into the case structures for the acquired verbs to effectively prepare scripts. In particular, the method analyzes the grammatical structure of a sentence in which the extracted verb is included and the elements are inserted into the corresponding place in the case structure. Because the completely correct insertion may be difficult to implement, relatively adequate data will be used actually. Further, the authors show that the acquired and constructed scripts can be effectively utilized in the story generation process in the INGS.

 

The subject of intransitive verbs can have different thematic roles. For example, in “the boy fell”, “the boy” takes a patient role, while in “the boy smiled”, “the boy” takes an agent role. To preserve a consistent mapping between thematic roles and syntactic position (subject in this case), linguists propose that the subject of some intransitive verbs is underlyingly a syntactic object (the Unaccusative Hypothesis, UH). This generates the prediction that the subject of an unaccusative must be mentally reactivated in its original postverbal position. Previous psycholinguistic studies have reported evidence of reactivation (Friedmann et al 2008, Koring et al 2012). However, these studies did not equate the unaccusative and unergative stimuli for imageability, animacy of the subjects, or sentential context, or visual stimuli, resulting in confounds which jeopardize their conclusions. We reexamined UH with two Visual World Paradigm experiments carefully controlled for all the factors above. On each trial, participants (n=40; n=60) saw 4 black-and-white drawings and heard a sentence. In the test condition, but not the control condition, one image was semantically related to subject of the sentence. We measured the proportion of looks to the target image at three time regions after the verb onset and found a robust match-effect (p’s<.05) but no differences between the unaccusative and unergative conditions (i.e. no difference in reactivation). By comparing our analyses with the previous analysis, we found that the effect in the previous studies are not stable due to their choice of statistics. We reconsider the prior experimental findings and conclude that the effects in the previous studies are due to the artifacts of confounds or noise in the data. From a methodological point of view, we show that the growth curve model with no transformation of the distribution is not appropriate for analyzing eye-tracking data.

 

Startling auditory stimulus (SAS, > 120dB) has been considered as a reliable trigger to a rapid release of prepared responses, including upper limb movements (Valls-Solé et al., 1999; Carlsen et al., 2012) and English consonant-vowel syllable production (Stevenson et al., 2014). The reaction time of the SAS-induced response is argued to be too short to involve feedback correction, suggesting that the rapid release is the result of a prepared program being executed with only limited afferent feedback. Chiu and Gick (2014) extended this startle paradigm to Mandarin syllable production, showing that while startle-elicited responses are released at shorter latencies, phonemic tonal contour and formant profiles are preserved. Both English and Mandarin results reveal that pitch height is elevated in startle-induced responses. However, such pitch elevation is less observed in pitch-trained speakers’ responses than in general Mandarin speakers’ responses (Chiu, 2017). In order to maintain a required target pitch level, the responses in Chiu (2017) were pre-specified longer than a syllable duration in connected speech. It is not clear whether a SAS may elicit similar effects to responses of different durations. The current study uses the startle paradigm to tackle this question by comparing general Mandarin speakers and Mandarin speakers with pitch training. Participants were instructed to produce a CV syllable of either 0.5, 1, or 3 seconds. Preliminary results show that with pre-specified duration, SAS-induced responses are not triggered at a latency as short as those in syllable responses with no duration specification. Such absence of rapid release is more robust for speakers with pitch training background. Results suggest that responses with specifically long durations may require additional online adjustment and thus may not be elicited as rapidly. A more general implication is that such prosodic information may not necessarily be specified in the speech plan prior to the production.

 

In a conversation, code-switching (CS) is often observed in bilinguals with the same language backgrounds. CS is the phenomenon of bilinguals alternating between two languages in conversations; this unique phenomenon was investigated by observing CS during behavioral experiments. Two sessions were conducted, assessing ten native Japanese subjects with high English proficiency; they performed deliberate CS conversational tasks using Japanese and English. All subjects were Japanese native speakers who were exposed to English after twelve years of age. They are currently using English in their professions. The first session was the task to name the pictures in either language that was the language used for the question. The other session the subjects had to name the pictures that were not the language used for the question. Both Japanese and English question sentences were timed and equalized for length. Pictures were presented immediately after the question sentences and the subjects’ response times were analyzed. Additionally, the subjects’ answers were analyzed if they were making language errors. The main finding was that their responses in Japanese were more delayed than those delivered in English. This suggests that the subjects’ attention to the second language is so cautious and aimed that they slowed down replying in Japanese as though they neglected to pay attention to their native language. Considering BIA+ model (Dijkstra & van Heuven, 2002) and L2 episodic hypothesis (e.g., Witzel & Forster, 2012), a putative model of late bilingual with high L2 proficiency was attempted to be proposed for CS production.

 

Teacher cognition has its origins in psychology, in the cognitive studies, but soon it was affecting, and being affected by the research developed in other areas. This influence helped the field become multidisciplinary which included studies on linguistics, anthropology, neuroscience, among others. In simple terms, teacher cognition relates to everything that is in the mind of the teachers, such as their beliefs, attitudes, knowledge, thoughts, and decision making. In the last 25 years, educational research has given meaningful attention to teacher cognition, especially in the teacher education spheres. However, a common question emerges in the literature of the field: the clear identification of the constructs of teacher cognition. They have been usually studied in relation to other constructs and such association help the conceptualization of new meanings which add to the general idea of teacher cognition. Therefore, the understanding of teacher cognition becomes more inclusive and constantly renewed. The research about teacher cognition which grounds this oral presentation follows the socio-cultural theories of Vygotsky and the educational principles of Freire. This research also applies core aspects of narrative enquiry and elements of discourse analysis in order to explore main teacher cognition construct. Having as a basis the model of narrativised teacher cognition, such as decision making, the audience is invited to analyze excerpts of teachers' narratives to find specific constructs and the potential relations between them. In the final discussion, it is expected that the audience find similar ways to the identification of some constructs, as well as their complex interrelations in the teachers' narratives.

 

Typical online video learning materials such as MOOCs (massive open online courses) often include teachers’ motion images concurrently with learning contents such as teachers’ speeches and PowerPoint visual slides. Although the teachers’ motion images seem to disrupt learning attitude because of depriving learners’ attention from the learning contents, why are the images used in many learning videos? We consider teachers’ nonverbal signs informed the images essential to learn even in asynchronous online learning as human learning is intrinsically not information-to-learner communication but teacher-to-learner, that is, human-to-human communication. As the first step of verifying the hypothesis, we investigated whether teachers’ motion images would divide learners’ attention from visual slide contents by eye-tracking. Three types of 4-minute video materials were prepared: motion image, still image, and no image. The typical video learning materials were motion images, which consisted of a visual slide, a teacher’s motion image and speech. In the still image materials, the motion images were replaced with the still images of the teacher. In the no image materials, the motion or still images were removed. We requested twelve university students to watch the materials naturally while evaluating their eye-fixations on the images and the slide areas with an eye tracker (Tobii X2-60). The findings show that in the motion image condition the average ratio of fixation frequency on the teacher’s image to the visual slide was 23% to 77%, and the ratio on the teacher’s image was significantly higher than that in the still image condition (10%). The ratio of total fixation time revealed the same pattern. These results suggest that learners watch teachers’ motion images while learning and acquire some information from them to learn effectively and deeply. Our further research aims to verify learning processes from video materials with teachers’ images.

 

We are fond of beautiful scenery, but not all types of scenery are equally preferred. Scene types have an enormous effect on the preferences for landscape in Western adult participants. A recent study using visual signal computational model to predict landscape preference discovered that Taiwanese young adults showed a higher preference for natural scenes (coasts, forests, countrysides, mountains views) than urban scenes (highways, tall buildings, streets, inner cities) (Ho et al., 2015). The present study aimed to explore the landscape preference in Taiwanese school-aged children using the same image database. Forty 5- to 12-year-old Taiwanese children participated the study. Each participant received 80 pictures containing four natural scene types (coasts, forests, countrysides, mountains) and four urban scene types (highways, tall buildings, streets, and inner cities), 10 for each type. There were six different sets of 80 pictures from the 480-picture image database. Each picture was displayed one by one on a touch screen notebook (Asus S551L). The participants were asked to drag each picture to one of the five folders based on their preference: 1) strongly disliked, 2) slightly disliked, 3) neither, 4) slightly liked, and 5) strongly liked. We found that Taiwanese children showed a higher preference for natural scenes (3.812) than urban scenes (3.159, p < .001), and their preference of the coast scenes (4.127) was the highest among all types. We also found that, compared to the adults’ rating scores in Ho et al. (2015), children tended to rate higher for both natural and urban scenes. In sum, the present study revealed that, like adults, Taiwanese children exhibited a stronger preference for natural scenes than urban scenes, which supports the prospect-refuge theory that natural scenes simultaneously provide abundance and a sense of security to meet human needs.

 

We have implemented three outline tools in an academic writing course for the first-year students at a university in Japan, where students learn how to write academic essays of 1500-2000 words length. First, mind map is introduced to help learners collect their own thoughts and pieces of reference information, and determine their opinion statements. Then, an outline map is introduced; using it learners extract materials that are relevant to their main statement and also explicitly classify the elements into opinion, evidence, and reason. Finally, our outlining scheme, called "detailed outline," is introduced to reorganize the resulting elements as a draft of text. The scheme of detailed outline is designed following the Rhetorical Structure Theory in linguistics (Mann and Thompson, 1988) aiming to facilitate thoroughly covering necessary elements and structurally organizing them to produce cohesive and coherent texts. To evaluate their effect on the learners, we conducted a questionnaire based on the scale for "feeling of difficulty" in writing composition (Kishi et al., 2012) in the beginning and the end of the semester. Paired t-test on the answers collected from 235 learners showed that the questions corresponding to the three factors, i.e., (a) text organization, (b) audience awareness, and (c) way of thinking factors, were answered significantly more positively in the end of semester. Since all of our outline tools visualize relations of elements, we speculate that introducing them helps learners produce and arrange their idea and consequently better understand the importance of cohesiveness and coherence in writing this length of texts.

 

In an academic writing course for the first-year students at a university in Japan, we have introduced outline tools along with a three-step text production process, assuming opinion statement texts of 1500-2000 word length. First, mind map is introduced to help learners collect their own thoughts and pieces of reference information, and determine their opinion statements. Then, an outline map is introduced; using it learners extract materials that are relevant to their main statement and also explicitly classify the elements into opinion, evidence, and reason. Finally, our outlining scheme, called "detailed outline," is introduced to reorganize the resulting elements as a draft of text. The scheme of detailed outline is designed following the Rhetorical Structure Theory in linguistics (Mann and Thompson, 1988) aiming to facilitate thoroughly covering necessary elements and structurally organizing them to produce cohesive and coherent texts. To help learners learn to produce such texts, in addition to the above three tools, we use four means for giving learners feedback: (a) reflection sheet, (b) peer reviewing, (c) rubric, and (d) personal instruction. A questionnaire was conducted after the semester that asked the usefulness of each of our seven means in a 7-point Likert scale. A principal factor analysis for the answers from 198 learners revealed clear distinction between two major factors: "process-aiding tools" and "way of feedback". A cluster analysis based on the factor scores resulted four learner clusters. One-way ANOVA showed that the factor scores of each cluster were significantly different, indicating that these two factors have sufficient power in discriminating the adaptability of learners to the examined means.