Visual Heuristics for Verb Production: Testing a Deep-Learning Model With Experiments in Japanese.

Tense/aspect morphology on verbs is often thought to depend on event features like telicity, but it is not known how speakers identify these features in visual scenes. To examine this question, we asked Japanese speakers to describe computer-generated animations of simple actions with variation in visual features related to telicity. Experiments with adults and children found that they could use goal information in the animations to select appropriate past and progressive verb forms. They also produced a large number of different verb forms. To explain these findings, a deep-learning model of verb production from visual input was created that could produce a human-like distribution of verb forms. It was able to use visual cues to select appropriate tense/aspect morphology. The model predicted that video duration would be related to verb complexity, and past tense production would increase when it received the endpoint as input. These predictions were confirmed in a third study with Japanese adults. This work suggests that verb production could be tightly linked to visual heuristics that support the understanding of events.

Learning conversational dependency: Children's response using un in Japanese.

This study investigates how Japanese-speaking children learn interactional dependencies in conversations that determine the use of un, a token typically used as a positive response for yes-no questions, backchannel, and acknowledgement. We hypothesise that children learn to produce un appropriately by recognising different types of cues occurring in the immediately preceding turns. We built a set of generalised linear models on the longitudinal conversation data from seven children aged 1 to 5 years and their caregivers. Our models revealed that children not only increased their un production, but also learned to attend relevant cues in the preceding turns to understand when to respond by producing un. Children increasingly produced un when their interlocutors asked a yes-no question or signalled the continuation of their own speech. These results illustrate how children learn the probabilistic dependency between adjacent turns, and become able to participate in conversational interactions.

Testing a computational model of causative overgeneralizations: Child judgment and production data from English, Hebrew, Hindi, Japanese and K'iche'.

How do language learners avoid the production of verb argument structure overgeneralization errors ( *The clown laughed the man c.f. The clown made the man laugh), while retaining the ability to apply such generalizations productively when appropriate? This question has long been seen as one that is both particularly central to acquisition research and particularly challenging. Focussing on causative overgeneralization errors of this type, a previous study reported a computational model that learns, on the basis of corpus data and human-derived verb-semantic-feature ratings, to predict adults' by-verb preferences for less- versus more-transparent causative forms (e.g., * The clown laughed the man vs The clown made the man laugh) across English, Hebrew, Hindi, Japanese and K'iche Mayan. Here, we tested the ability of this model (and an expanded version with multiple hidden layers) to explain binary grammaticality judgment data from children aged 4;0-5;0, and elicited-production data from children aged 4;0-5;0 and 5;6-6;6 ( N=48 per language). In general, the model successfully simulated both children's judgment and production data, with correlations of r=0.5-0.6 and r=0.75-0.85, respectively, and also generalized to unseen verbs. Importantly, learners of all five languages showed some evidence of making the types of overgeneralization errors - in both judgments and production - previously observed in naturalistic studies of English (e.g., *I'm dancing it). Together with previous findings, the present study demonstrates that a simple learning model can explain (a) adults' continuous judgment data, (b) children's binary judgment data and (c) children's production data (with no training of these datasets), and therefore constitutes a plausible mechanistic account of the acquisition of verbs' argument structure restrictions.

The crosslinguistic acquisition of sentence structure: Computational modeling and grammaticality judgments from adult and child speakers of English, Japanese, Hindi, Hebrew and K'iche'.

This preregistered study tested three theoretical proposals for how children form productive yet restricted linguistic generalizations, avoiding errors such as *The clown laughed the man, across three age groups (5-6 years, 9-10 years, adults) and five languages (English, Japanese, Hindi, Hebrew and K'iche'). Participants rated, on a five-point scale, correct and ungrammatical sentences describing events of causation (e.g., *Someone laughed the man; Someone made the man laugh; Someone broke the truck; ?Someone made the truck break). The verb-semantics hypothesis predicts that, for all languages, by-verb differences in acceptability ratings will be predicted by the extent to which the causing and caused event (e.g., amusing and laughing) merge conceptually into a single event (as rated by separate groups of adult participants). The entrenchment and preemption hypotheses predict, for all languages, that by-verb differences in acceptability ratings will be predicted by, respectively, the verb's relative overall frequency, and frequency in nearly-synonymous constructions (e.g., X made Y laugh for *Someone laughed the man). Analysis using mixed effects models revealed that entrenchment/preemption effects (which could not be distinguished due to collinearity) were observed for all age groups and all languages except K'iche', which suffered from a thin corpus and showed only preemption sporadically. All languages showed effects of event-merge semantics, except K'iche' which showed only effects of supplementary semantic predictors. We end by presenting a computational model which successfully simulates this pattern of results in a single discriminative-learning mechanism, achieving by-verb correlations of around r = 0.75 with human judgment data.

Testing an input-based account of children's errors with inflectional morphology: an elicited production study of Japanese.

This study tested the claim of input-based accounts of language acquisition that children's inflectional errors reflect competition between different forms of the same verb in memory. In order to distinguish this claim from the claim that inflectional errors reflect the use of a morphosyntactic default, we focused on the Japanese verb system, which shows substantial by-verb variation in the frequency distribution of past and nonpast forms. 22 children aged 3;2-5;8 (Study 1) and 26 children aged 2;7-4;11 (Study 2) completed elicited production studies designed to elicit past and nonpast forms of 20 verbs (past-biased and nonpast-biased). Children made errors in both directions, using past forms in nonpast contexts, and vice versa, with the likelihood of each determined by the frequency bias of the two forms in the input language, even after controlling for telicity. This bi-directional pattern provides particularly direct evidence for the role of frequency-sensitive competition between stored forms.

Disentangling Effects of Input Frequency and Morphophonological Complexity on Children's Acquisition of Verb Inflection: An Elicited Production Study of Japanese.

This study aims to disentangle the often-confounded effects of input frequency and morphophonological complexity in the acquisition of inflection, by focusing on simple and complex verb forms in Japanese. Study 1 tested 28 children aged 3;3-4;3 on stative (complex) and simple past forms, and Study 2 tested 30 children aged 3;5-5;3 on completive (complex) and simple past forms, with both studies using a production priming paradigm. Mixed effects models for children's responses were built to test the prediction that children's verb use is explained by the relative bias in input frequency between the two inflectional forms. Although Study 1 did not show a significant effect of input bias (apparently due to problems with item selection), Study 2, which corrected for this problem, yielded the predicted relationship. These findings suggest that input frequency effects, at the level of different inflectional forms of the same verb stem, hold even after controlling for morphophonological complexity.

Domain-general mechanisms for speech segmentation: The role of duration information in language learning.

Speech segmentation is supported by multiple sources of information that may either inform language processing specifically, or serve learning more broadly. The Iambic/Trochaic Law (ITL), where increased duration indicates the end of a group and increased emphasis indicates the beginning of a group, has been proposed as a domain-general mechanism that also applies to language. However, language background has been suggested to modulate use of the ITL, meaning that these perceptual grouping preferences may instead be a consequence of language exposure. To distinguish between these accounts, we exposed native-English and native-Japanese listeners to sequences of speech (Experiment 1) and nonspeech stimuli (Experiment 2), and examined segmentation using a 2AFC task. Duration was manipulated over 3 conditions: sequences contained either an initial-item duration increase, or a final-item duration increase, or items of uniform duration. In Experiment 1, language background did not affect the use of duration as a cue for segmenting speech in a structured artificial language. In Experiment 2, the same results were found for grouping structured sequences of visual shapes. The results are consistent with proposals that duration information draws upon a domain-general mechanism that can apply to the special case of language acquisition. (PsycINFO Database Record

Comparing generativist and constructivist accounts of the use of the past tense form in early child Japanese.

The present study investigated children's early use of verb inflection in Japanese by comparing a generativist account, which predicts that the past tense will have a special default-like status for the child during the early stages, with a constructivist input-driven account, which assumes that children's acquisition and use of inflectional forms reflects verb-specific distributional patterns in their input. Analysis of naturalistic data from four Japanese children aged 1;5 to 2;10 showed that there was substantial by-verb variation in the use of inflectional forms from the earliest stages of verb use, and no general preference for past tense forms. Correlational and partial correlational analyses showed that it was possible to predict the proportional frequency with which the child produced verbs in past tense versus other inflectional forms on the basis of differences in the proportional frequency with which the verb occurred in past tense form in the child's input, even after controlling for differences in the rate at which verbs occurred in past tense form in input averaged across the caregivers of the other children in the sample. When taken together, these results count against the idea that the past tense has a special default-like status in early child Japanese, and in favour of a constructivist input-driven account of children's early use of verb inflection.

Regional cerebral blood flow in a patient with Nasu-Hakola disease.

We report a functional neuroimaging study of a 43-year-old woman with Nasu-Hakola disease (NHD). Regional cerebral blood flow (rCBF) images were measured with technetium-99m ethyl cysteinate dimer single photon emission computed tomography (SPECT). rCBF was decreased in the bilateral frontal lobes and thalamus. This finding was consistent with the known underlying neuropathology in patients with NHD. Brain SPECT is useful for demonstrating the pathophysiologic brain region in patients with NHD.

A rice semi-dwarf gene, Tan-Ginbozu (D35), encodes the gibberellin biosynthesis enzyme, ent-kaurene oxidase.

A rice (Oryza sativa L.) semi-dwarf cultivar, Tan-Ginbozu (d35Tan-Ginbozu), contributed to the increase in crop productivity in Japan in the 1950s. Previous studies suggested that the semi-dwarf stature of d35Tan-Ginbozu is caused by a defective early step of gibberellin biosynthesis, which is catalyzed by ent-kaurene oxidase (KO). To study the molecular characteristics of d35Tan-Ginbozu, we isolated 5 KO-like (KOL) genes from the rice genome, which encoded proteins highly homologous to Arabidopsis and pumpkin KOs. The genes (OsKOL1 to 5) were arranged as tandem repeats in the same direction within a 120 kb sequence. Expression analysis revealed that OsKOL2 and OsKOL4 were actively transcribed in various organs, while OsKOL1 and OsKOL5 were expressed only at low levels; OsKOL3 may be a pseudogene. Sequence analysis and complementation experiments demonstrated that OsKOL2 corresponds to D35. Homozygote with null alleles of D35 showed a severe dwarf phenotype; therefore, d35Tan-Ginbozu is a weak allele of D35. Introduction of OsKOL4 into d35Tan-Ginbozu did not rescue its dwarf phenotype, indicating that OsKOL4 is not involved in GA biosynthesis. OsKOL4 and OsKOL5 are likely to take part in phytoalexin biosynthesis, because their expression was promoted by UV irradiation and/or elicitor treatment. Comparing d35Tan-Ginbozu with other high yielding cultivars, we discuss strategies to produce culm architectures suitable for high crop yield by decreasing GA levels.

An overview of gibberellin metabolism enzyme genes and their related mutants in rice.

To enhance our understanding of GA metabolism in rice (Oryza sativa), we intensively screened and identified 29 candidate genes encoding the following GA metabolic enzymes using all available rice DNA databases: ent-copalyl diphosphate synthase (CPS), ent-kaurene synthase (KS), ent-kaurene oxidase (KO), ent-kaurenoic acid oxidase (KAO), GA 20-oxidase (GA20ox), GA 3-oxidase (GA3ox), and GA 2-oxidase (GA2ox). In contrast to the Arabidopsis genome, multiple CPS-like, KS-like, and KO-like genes were identified in the rice genome, most of which are contiguously arranged. We also identified 18 GA-deficient rice mutants at six different loci from rice mutant collections. Based on the mutant and expression analyses, we demonstrated that the enzymes catalyzing the early steps in the GA biosynthetic pathway (i.e. CPS, KS, KO, and KAO) are mainly encoded by single genes, while those for later steps (i.e. GA20ox, GA3ox, and GA2ox) are encoded by gene families. The remaining CPS-like, KS-like, and KO-like genes were likely to be involved in the biosynthesis of diterpene phytoalexins rather than GAs because the expression of two CPS-like and three KS-like genes (OsCPS2, OsCPS4, OsKS4, OsKS7, and OsKS8) were increased by UV irradiation, and four of these genes (OsCPS2, OsCPS4, OsKS4, and OsKS7) were also induced by an elicitor treatment.

Changes across age groups in self-choice elaboration and incidental memory.

This study investigated differences in the self-choice elaboration and an experimenter-provided elaboration on incidental memory of 7- to 12-yr.-olds. In a self-choice elaboration condition 34 second and 25 sixth graders were asked to choose one of the two sentence frames into which each target could fit more congruously, whereas in an experimenter-provided elaboration they were asked to judge the congruity of each target to each frame. In free recall, sixth graders recalled targets in bizarre sentence frames better than second graders for self-choice elaboration condition. An age difference was not found for the experimenter-provided elaboration. In cued recall self-choice elaboration led to better performance of sixth graders for recalling targets than an experimenter-provided elaboration in both bizarre and common sentence frames. However, the different types of elaboration did not alter the recall of second graders. These results were interpreted as showing that the effectiveness of a self-choice elaboration depends on the subjects' age and the type of sentence.