Syntactic Skills Research Articles

Maternal Question Use Relates to Syntactic Skills in 5- to 7-Year-Old Children.

This study examined how mothers' question-asking behavior relates to their child's syntactic skills. One important aspect of maternal question-asking behavior is the use of complex questions when speaking with children. These questions can differ based on both their purpose and structure. The purpose may be to seek out information, to teach, or to get a simple yes/no response. Questions may even be rhetorical, with no answer intended at all. Structurally, questions can include a wh-word (who, what, when, where, why, and how) or not; however, these wh-questions are important because they elicit utterances from the child and support vocabulary development. Despite wh-questions eliciting a response from children, it remains unknown how these questions relate to children's syntactic skills. Thirty-four mother-child dyads participated in a 15-min seminaturalistic play session. Children were between the ages of 5 and 7 years (M = 6.26 years, SD = 1.04 years; 20 girls/14 boys). The Diagnostic Evaluation of Language Variation (DELV) assessment was used to measure syntactic skills in children. Using the Systematic Analysis of Language Transcripts, questions were categorized based on structure (wh-questions vs. non-wh-questions) and purpose (information-seeking, pedagogical, or yes/no and rhetorical questions). A repeated-measures analysis of covariance and a linear regression model were implemented to address the frequency of different questions asked by mothers, as well as what types of questions are most related to children's concurrent syntactic skills. When controlling for total maternal utterances, results revealed that non-wh-questions and rhetorical/yes and no questions were the most frequent types of questions produced by mothers, in terms of structure and purpose, respectively. However, wh-questions were predominantly information-seeking questions. This is important, as the use of information-seeking wh-questions was positively associated with children's syntactic skills, as measured by the DELV, and resulted in children producing longer utterances in response to these questions, as determined by child mean length of utterance in words. Taken together, these findings suggest maternal use of wh-questions aids syntactic skills in children ages 5-7 years, likely because they require a more syntactically complex response on the child's behalf. https://doi.org/10.23641/asha.27276891.

Narrative production and executive functions in post-stroke agrammatic aphasia

ABSTRACT Background Narrative production has been widely characterized as providing an ecologically valid way to assess language skills in post-stroke aphasia. Although narrative tasks have been instrumental in delineating patterns of lexical and syntactic production in individuals with agrammatic aphasia, our knowledge of how narrative skills are affected in agrammatic aphasia is still limited. Aims The study’s aims were to (a) compare narrative performance between individuals with agrammatic aphasia and a language-unimpaired group, (b) investigate the contribution of lexical and syntactic skills in each group’s narrative organizational skills, and (c) explore the effects of executive functions on each group’s narrative performance. Methods & Procedures The study included 14 individuals with agrammatic aphasia and 14 age- and education-matched language unimpaired individuals as the control group. Both groups told the Cinderella story, and their narrative production was analyzed in terms of microstructure (lexical diversity, syntactic complexity, clause length) and macrostructure (story grammar, story structure complexity). Each group’s executive functions were evaluated through the One-Touch Stockings of Cambridge test that assesses working memory, and the Intra-Extra Dimensional Set Shift test that assesses cognitive flexibility. Outcomes & Results Regarding narrative microstructure, the individuals with agrammatism scored significantly lower than controls in syntactic complexity and clause length, but not in lexical diversity, but they performed lower than controls in both narrative macrostructural measurements. Also, the agrammatic group scored lower than controls in both executive function tasks. A series of linear regression models showed that microstructural skills significantly affected narrative macrostructural abilities in the cohort with agrammatic aphasia, while controls’ macrostructure was affected by both narrative microstructure and executive function skills. Conclusions The individuals with agrammatic aphasia exhibited impairment in both low-level language features of narrative production, such as syntactic complexity and narrative length, and global measures of narrative organization. Their macrostructural performance critically relied on their language resources, while controls’ narrative macrostructure seemed to also draw on domain-general cognitive skills. These findings suggest that both low-level language and high-level discourse organizational skills are vulnerable in narrative production in agrammatic aphasia, and that macrostructural skills are mainly related to the individuals’ microstructural skills.

A Study on the Using of Endings (“Eomi”) of 3- to 5-Year-Old Children in Spontaneous Language Samples

Objectives: Considering that language development is the development of language use ability, endings, which are essential for sentence construction, can be said to be a good indicator of language development as well as grammatical development. This study attempted to systematically and comprehensively examine the overall and subcategory use of endings in spontaneous speech for children aged 3 to 5 whose syntactic skills are rapidly developing. Methods: Language samples containing 50 utterances were collected from a total of 150 children aged 3 to 5 years (50 in each age group) through a semi-structured procedure. ‘Eomi’s are classified into four subcategories: prefinal Eomi (PE), Sentence-closing Eomi (SE), Connecting Eomi (CE), Transformative Eomi (TE). The number of total Eomi (NTE) use, the number of different Eomi (NDE) of overall Eomi & all subcategories of Eomi were analyzed. Results: The NTE & NDE of overall Eomi and all subcategories except SE significantly increased between the 3-year-old group and two older groups, but not between two older groups. Regression analysis demonstrated that the NDE of final Eomi and the NTE of CE significantly explained chronological age in the 3-year-old group. Conclusion: It was confirmed that NDE along with NTE of overall and all subcategories of Eomi consistently increased actively between the ages of 3 and 4. Also NDE along with NTE of Eomi should be considered in language assessment and intervention.

Examining Syntactic Errors in Essay Writing Among EFL Students: A Case Study of ESEF Ibn Tofail University

This article presents a thorough examination of error patterns within essays written by students enrolled at Ibn Tofail University, focusing on four major types of errors: Omission Addition, Mis-ordering, and Mis-formation. Utilizing data collected through essays and coded through content analysis. The data collected were analysed through SPSS, descriptive analysis, the study investigates the prevalence, distribution, and characteristics of these errors across a diverse range of essays. Through frequency distributions and skewness analysis, significant perspectives into error distributions and patterns emerged, revealing varying degrees of writing proficiency among the sample. The findings showed that the most frequent syntactic errors in students’ writings are addition errors and the most infrequent ones are mis-ordering. Professors may need to focus more attention on addressing and correcting addition errors, as they appear to be more prevalent and may have a greater impact on the overall clarity and coherence of students’ writing. Meanwhile, less emphasis may be needed on Mis-ordering errors, although they should still be addressed to ensure students develop strong syntactic skills and produce well-structured compositions.

A Meta-Analysis of the Relation Between Syntactic Skills and Reading Comprehension: A Cross-Linguistic and Developmental Investigation

Theories of reading comprehension have widely predicted a role for syntactic skills, or the ability to understand and manipulate the structure of a sentence. Yet, these theories are based primarily on English, leaving open the question of whether this remains true across typologically different languages such as English versus Chinese. There are substantial differences in the sentence structures of Chinese versus English, making the comparison of the two particularly interesting. We conducted a meta-analysis contrasting the relation between syntactic skills and reading comprehension in first language readers of English versus Chinese. We test the influence of languages as well as the influence of the grade and the metrics on the magnitude of this relation. We identified 59 studies published between 1986 and 2021, generating 234 effect sizes involving 15,212 participants from kindergarten to high school and above. The magnitude of effects was remarkably similar for studies of English (r = .54) and Chinese (r = .54) readers, with similarities at key developmental points and syntactic tasks. There was also some evidence of modulation by grade levels and the nature of syntactic tasks. These findings confirm theory-based predictions of the importance of syntactic skills to reading comprehension. Extending these predictions, demonstrating these effects for both English and Chinese suggests a universal influence of syntactic skills on reading comprehension.

Hubungan Kemampuan Sintaksis dengan Reading Comprehension pada Anak Gangguan Pendengaran di Surakarta

Abstract
 Background: Children with hearing impairment delays in aspects of language, one of which is written language. One of the most difficult written languages to mastered is reading comprehension. Low reading comprehension ability is often associated with low syntax ability. Sentences contained in a reading are quite complex which requires syntactical skills in reading comprehension. Objective: This research aim for knowing exists relationship between syntax ability and reading comprehension of hearing impairment children in Surakarta. Method: This research use quantitative research with cross-sectional data collection. Technique sample used on this research is purposive sampling with 30 respondens. The statistical test used is the Spearman Rank test. Results: The results of this study indicate a significance value (ρ) of 0.007 so that there is syntactic ability with reading comprehension in children with hearing loss in Surakarta. The strength indicates a correlation coefficient (r) of 0.481 which means that having a moderate category of correlation strength. Conclusion: There is relationship syntax ability with reading comprehension of hearing impairment children in Surakarta.
 Keywords : Syntax Ability, Reading Comprehension, Hearing Impairment Children

Individual differences in neural markers of beat processing relate to spoken grammar skills in six-year-old children

Based on the idea that neural entrainment establishes regular attentional fluctuations that facilitate hierarchical processing in both music and language, we hypothesized that individual differences in syntactic (grammatical) skills will be partly explained by patterns of neural responses to musical rhythm. To test this hypothesis, we recorded neural activity using electroencephalography (EEG) while children (N = 25) listened passively to rhythmic patterns that induced different beat percepts. Analysis of evoked beta and gamma activity revealed that individual differences in the magnitude of neural responses to rhythm explained variance in six-year-olds’ expressive grammar abilities, beyond and complementarily to their performance in a behavioral rhythm perception task. These results reinforce the idea that mechanisms of neural beat entrainment may be a shared neural resource supporting hierarchical processing across music and language and suggest a relevant marker of the relationship between rhythm processing and grammar abilities in elementary-school-age children, previously observed only behaviorally.

Investigating the Relevance of the NCCE Curriculum in Teaching Pre-Service Students the Syntactic Structure of Arabic Sentences

This research paper investigates the relevance of the National Commission for Colleges of Education (NCCE) curriculum in teaching pre-service students the syntactic structure of Arabic sentences. The study aims to assess the effectiveness of the current curriculum in developing students' understanding and proficiency in Arabic syntax. Through a mixed-methods approach, including surveys, classroom observations, and student assessments, data is collected to evaluate the alignment between the curriculum and students' learning outcomes. The findings reveal both strengths and areas for improvement in the NCCE curriculum's coverage of Arabic sentence syntax. The study concludes with recommendations for curriculum enhancements and instructional strategies to better support pre-service students' acquisition of syntactic skills in Arabic language learning.

Linking Language to Action: Enhancing Preschoolers' Communicative Abilities Within Language Stimulation.

Children's vocabulary and syntactic skills vary upon school entry in depth and breadth, persistently influencing academic performance, including reading. Enhancing early communicative abilities through multisensory, playful, and conversational experiences is essential and will benefit children's school readiness. This study investigated whether a language-to-action link created during language stimulation, which combines multisensory input, play, and conversation using clay, improves preschoolers' communicative abilities in terms of vocabulary, syntactic, and pragmatic language abilities more than traditional toy-based language stimulation. Language skills were examined in a pre- to posttest design in which 43 typically developing participants, ages 3-5 years, were randomly assigned to clay-based (n = 24) or traditional play-based (n = 19) language stimulation for 8 weeks. Receptive and expressive vocabulary knowledge for items introduced in the language stimulation program, mean length of utterance (MLU), and conversational initiations improved for participants in the clay condition, whereas significant language skill growth was not observed for participants in the traditional play-based stimulation condition with toys. A language-to-action link is created when children engage with open-ended materials, such as clay, as they craft target objects hands on and step by step, affording additional opportunities for language input and output. Results preliminarily suggest that using open-ended materials may enhance children's communicative abilities in receptive and expressive vocabulary, syntax/MLU, and pragmatics (i.e., conversational initiations) more than prefabricated toy objects during language stimulation. https://doi.org/10.23641/asha.24093780.

Writing Skills in 1st to 2nd Graders with Language Difficulties, Typical Development, and Dyslexia with/without Language Difficulties

Objectives: The purpose of this study was to investigate characteristics of writing ability and writing related variables such as executive functioning (syntactic ability, sentence construction, planning ability) and working memory. Additionally, this study tried to find out the variables that can discriminate between the four groups.Methods: Fourteen 1st to 2nd graders in each group (dyslexia (DY), dyslexia with language difficulty (LRD), language difficulty without dyslexia (LD), and typical development (TD)) participated in four writing tasks and four writing related tasks.Results: First, there were three performance groups in word writing and sentence writing (TD > LD > DY = LRD), while there were two different groups were observed in short text writing and handwriting (TD = LD > DY = LRD). Secondly, there were three performance groups in syntactic skills and sentence construction (TD> DY> LD= LRD) and in planning skills (TD> DY= LD> LRD), while there were two performance groups in copying (TD= LD> DY= LRD). Thirdly, discriminant analysis showed a 94.6% hit ratio; and word writing, syntactic ability, sentence construction, and sentence writing were good discriminators for the four groups.Conclusion: This study was meaningful in including four groups of children considering language and decoding abilities. For 1st to 2nd graders, sentence writing and word writing assessment and intervention considering executive functions seem to play the key role for each student’s school adaptation with language and/or reading difficulties.

Phonological Mean Length of Utterance in 48-60-Month-old Persian-Speaking Children With Isfahani Accent: Comparison of Story Generation and Conversation Samples

Objective Phonological Mean Length of Utterance (PMLU), a quantitative measure for assessing phonological skills, is a diagnostic and clinical criterion in phonological development. Moreover, it is an indicator showing the efficacy of the intervention. The PMLU is a word level measure that can be calculated on the child’s transcribed speech sample (transcription). To calculate PMLU, all consonants and vowels of the child’s produced words, and target words (standard production of words in native adults) are individually scored. The proportion of Whole-Word Proximity (PWP), another phonological quantitative measure, includes the ratio of the produced PMLU to the targeted word PMLU. PWP indirectly reflects the intelligibility of speech. Since languages are distinctive in syllabic and phonological structures, PMLU should be studied as a language-specific measure. PMLU has specifically been designed to assess phonological skills in spontaneous speech. Spontaneous speech sampling methods are advantageous since they consider the effect of morphological and syntactic skills, length, and complexity of words, and they could show the normal development of word complexity. This study was conducted to determine PMLU and PWP in 48 to 60 months old Persian-speaking children with Isfahani accents and to compare them in story generation and conversation sampling methods. The potential sensitivity of PMLU to growth was also examined. Materials & Methods This is an observational and cross-sectional study that was conducted for one year in 2016 in Isfahan City, Iran. A total of 100 children (51 boys and 49 girls) aged 48-60 months participated in story generation sampling, and 67 children (32 boys and 35 girls) participated in conversation sampling. The participants were selected from 261 kindergartens under the supervision of the Welfare Organization of Isfahan Province using the convenience sampling method. After completing the consent form and considering the inclusion criteria, conversation and story-generation samples were collected. Audio samplings were done in the same room using the same software (Clear Record Litev. 2.1). Raters transcribed the first 50 words of the recorded speech samples. Finally, we used the formula to calculate the target PMLU and child PMLU based on the normal production of an adult who speaks Persian with an Isfahani accent and the child’s production, respectively. To evaluate inter-rater reliability, raters randomly transcribed 20% of all samples to recalculate values. Participants’ story generation and conversation scores were entered into SPSS16 separately. The Kolmogorov-Smirnov test was used to examine the data distribution. Based on the data distribution, paired t-test and Wilcoxon test were used to compare measurements, and the Pearson and Spearman tests were used to investigate the associations. Results Child PMLU, target PMLU, and PWP measures of the story generation method were 8.794, 8.811, 0.998, and those of the conversation method were 9.068, 9.093, and 0.998, respectively. Correlation test results showed significant relationship between age and PWP in story generation (r=0.308) and conversation (r=0.313). Comparing child PMLU in story generation and conversation showed a significant relationship between child PMLU in both sampling methods (P=0.000). The result of the target PMLU comparison in two methods of story generation and conversation (P=0.000) was significant. PWP did not significantly differ between the two sampling methods (P=0.973). The inter-rater reliability was calculated at 0.70. Conclusion This study can be used as a basis for quantitative studies in the field of children’s phonological assessment using Persian whole words. However, longitudinal studies in different age groups with a high level of evidence in this field can convince therapists to use whole-word measures in clinics.

Impact of language-training programs on French educators’ interactions and child syntax development

Studies that measure the effectiveness of language-training program for children’s language development do not clearly identify in adult-child interaction which characteristics of educators’ language are most effective, particularly for children over 4 years of age who need to develop complex syntactic skills. Our study aimed to analyze the language use of 26 educators (with different professions: teachers, animators, and employees) participating in French language-training program for preschoolers. This training was based on linguistic research on language acquisition and focused on syntax. A total of 131 children were recorded in adult-child joint narration in a dyadic interaction across six French cities and 20 preschools. The statistical analysis of the linguistic and pragmatic properties of the educators’ language while interacting with children demonstrated that the educators changed their language practices after training. This study also revealed that syntax complexity of educators’ input has a significant impact on syntax development of children.

A model of Chinese reading comprehension: The role of cognition and motivation

This study examined the contribution of reading motivation to reading comprehension among Chinese elementary children from a self-determination theory perspective. A total of 379 Chinese third to fifth graders were assessed on four types of reading motivation (i.e., external regulation, introjected regulation, identified regulation and intrinsic motivation), cognitive factors important to Chinese reading comprehension (i.e., verbal working memory, reading fluency and syntactic skills), and reading comprehension. Structural equation modelling analysis results showed that introjected regulation negatively predicted reading comprehension and identified regulation positively predicted reading comprehension. The contribution of identified regulation to reading comprehension was mediated by its contribution to cognitive factors. These results underscore the possible positive association of reading motivation with cognitive factors important to reading comprehension development.

Comprehending L2 Comprehension: A study of Arabic-Swedish bilingual preschoolers’ performance on a Swedish proficiency test

About a fifth of all children in Sweden learn the societal language Swedish outside of the home, i.e., they have Swedish as a second language (L2). Many of these children have lower socioeconomic status (SES) backgrounds, which predicts lower language proficiency. The aim of the present study is twofold: to contribute to a greater understanding of L2-Swedish proficiency in preschoolers with lower SES backgrounds, and to find out how proficiency tests should be adapted for bilingual children such that the tests are valid, i.e., unbiased to the language status (L1 or L2). We investigate test performance on a Swedish receptive language proficiency test (the Comprehension scale of The New Reynell Developmental Language Scales, NRDLS) which has a monolingual norming sample. The participants are 51 bilingual children (3-5-years of age) with Arabic as their L1, and who attend preschools in Swedish neighborhoods with lower SES. Results indicate that in contrast to the norming sample, bilingual children’s raw scores for subsections of the test are not progressively more difficult. Thus, we need to be aware that bilingual children’s high proficiency in a particular aspect of the language does not necessarily imply that they are proficient in aspects that would be considered easier from a monolingual perspective. In addition, there are indications that unfamiliarity with L2 lexical items, that are typically acquired early in L1, causes bilingual children to fail on tasks aimed at assessing syntactic skills, even though they appear to understand the syntactic pattern. We conclude with suggestions for special considerations and adaptations to assess individual L2-comprehension in preschoolers more accurately, such that practitioners in turn can support the children’s language development.

Writing motivation in Chinese children with developmental dyslexia

BackgroundRecent models of writing acknowledge the influential role of motivation in writing development. There has been a paucity of research on writing motivation in children. This study investigated writing motivation among Chinese children with developmental dyslexia.MethodsForty‐one Chinese children with developmental dyslexia (RD) in Grades 3 to 5 and their typically developing counterparts matched on age (CA controls) (n = 41) were administered tasks assessing their writing motivation (perceived value of writing and writing apprehension), cognitive‐linguistic skills (word dictation, syntactic skills and working memory), and writing abilities (quality and productivity).ResultsThe results showed that the RD group performed significantly less well than the CA group in all cognitive‐linguistic skills and writing abilities measures. No significant between‐group differences were found in writing motivation measures. Multiple regression analysis results showed that perceived value of writing significantly predicted writing productivity in the RD group and writing quality in the CA group. Word dictation was the only significant predictor of writing quality in both the RD and the CA group.ConclusionsThese findings suggest that writing motivation influences writing performance in Chinese children with developmental dyslexia in elementary grades. Specifically, perceived value of writing predicts writing quality in typically developing Chinese children but predicts only writing productivity, not writing quality, in Chinese children with developmental dyslexia. The impact of writing motivation on writing performance becomes stronger with advances in component skills acquisition.

Morphosyntactic Skills Influence the Written Decoding Accuracy of Italian Children With and Without Developmental Dyslexia.

Some types of developmental dyslexia (DD) are associated with morphology. Deep DD leads to morphological and semantic errors, and possible comorbidity with syntactic deficits; phonological-output-buffer DD causes problems in decoding longer morphologically complex words. In addition, cross-linguistic studies highlight the effects of morphological awareness on reading accuracy and fluency. The role of morphosyntactic abilities on reading is, however, not clear. This study explores the influence of morphosyntactic competence on reading in Italian children with and without DD. A total of 14 children with DD and 28 with Typical Development (TD) attending the Italian primary school were tested on written decoding, syntactic comprehension of different grammatical structures, and syntactic production of direct object clitic pronouns. DD children were significantly less accurate and slower in reading than TD children. Syntactic skills of the two groups did not differ significantly, but some differences in their acquisitional pace emerged. Syntactic comprehension and production of direct-object-clitic pronouns predicted reading accuracy standard scores, thus suggesting that morphosyntactic abilities, beyond clitics’ weak phonological status, affect decoding accuracy. Decoding accuracy was influenced by reading errors related to morphology (morphological, semantic, and phonological-output-buffer errors). Decoding speed was a specific weakness of DD children and was rather affected by multi-letter combinations. Consistent with a dual-route approach to orthographic processing, we argue that accuracy depends on fine-grained decoding strategies maximizing the precise ordering of letters, thus it is more sensitive to morphosyntactic skills. Morphological reading errors were associated with phonologically weak (determiners, clitic pronouns, and prepositions) and salient words (verbs). This suggests that the decoding of function words and morphologically complex words is particularly demanding and related to both phonological and morphosyntactic skills. Age had a negative predictive effect on semantic errors, compatible with the gradual acquisition of lexical decoding strategies, which seemed to be slowed down by DD. We conclude that oral morphosyntactic skills play a role in reading accuracy in the Italian shallow orthography for both DD and TD children. It is then advisable to assess children’s linguistic profile during DD diagnoses to establish whether some reading errors are related to morphosyntactic weakness. In this case, ad hoc morphosyntactic training might support reading accuracy.

Computational Models of Readers' Apperceptive Mass.

Recent progress in machine-learning-based distributed semantic models (DSMs) offers new ways to simulate the apperceptive mass (AM; Kintsch, 1980) of reader groups or individual readers and to predict their performance in reading-related tasks. The AM integrates the mental lexicon with world knowledge, as for example, acquired via reading books. Following pioneering work by Denhière and Lemaire (2004), here, we computed DSMs based on a representative corpus of German children and youth literature (Jacobs et al., 2020) as null models of the part of the AM that represents distributional semantic input, for readers of different reading ages (grades 1–2, 3–4, and 5–6). After a series of DSM quality tests, we evaluated the performance of these models quantitatively in various tasks to simulate the different reader groups' hypothetical semantic and syntactic skills. In a final study, we compared the models' performance with that of human adult and children readers in two rating tasks. Overall, the results show that with increasing reading age performance in practically all tasks becomes better. The approach taken in these studies reveals the limits of DSMs for simulating human AM and their potential for applications in scientific studies of literature, research in education, or developmental science.

Word- and Text-Level Processes Contributing to Fluent Reading of Word Lists and Sentences.

In this study, we investigated how word- and text-level processes contribute to different types of reading fluency measures. We aimed to increase our understanding of the underlying processes necessary for fluent reading. The sample included 73 Dutch Grade 3 children, who were assessed on serial word reading rate (familiar words), word-list reading fluency (increasingly difficult words), and sentence reading fluency. Word-level processes were individual word recognition speed (discrete word reading) and sequential processing efficiency (serial digit naming). Text-level processes were receptive vocabulary and syntactic skills. The results showed that word- and text-level processes combined accounted for a comparable amount of variance in all fluency outcomes. Both word-level processes were moderate predictors of all fluency outcomes. However, vocabulary only moderately predicted sentence reading fluency, and syntactic skills merely contributed to sentence reading fluency indirectly through vocabulary. The findings indicate that sequential processing efficiency has a crucial role in reading fluency across various measures besides individual word recognition speed. Additionally, text-level processes come into play when complexity and context availability of fluency measures increases, but the exact timing requires further study. Findings are discussed in terms of future directions and their possible value for diagnostic assessment and intervention of reading difficulties.

What Can Bilingual Children Tell Us About the Developmental Relationship Between Vocabulary and Grammar?

Strong correlations between vocabulary and grammar are well attested in language development in monolingual and bilingual children. What is less clear is whether there is any directionality in the relationship between the two constructs, whether it is predictive over time, and the extent to which it is affected by language input. In the present study, we analyzed data from 100 bilingual children with English as an additional language who were tested on measures of vocabulary breadth and depth, morphology, and syntax at three time points at 6-month intervals from the age of 5 and 8. We used bivariate growth models to test the directionality of the relationship between vocabulary breadth and depth, and measures of morphology and syntax; testing bilingual children allowed us to use measures of English input as covariates in the analyses. All the models showed a correlation between vocabulary and grammar, but no correlation between their growth slopes, suggesting that vocabulary and grammar grow independently. Three of the four bivariate models showed a significant correlation between the intercept of grammar skills and the slope of vocabulary growth. Length of exposure to English predicted the intercept of vocabulary breadth and grammar, suggesting that children exposed to English earlier had larger vocabularies and better morpho-syntactic skills. Current English input predicted the intercept of both measures of vocabulary as well as the slope for vocabulary depth, the only measure for which there was a significant relationship between intercept and slope, suggesting a Matthew effect for this dimension of vocabulary. All materials, data, and code are available at https://osf.io/vaq56/. Research highlights Vocabulary breadth and morphological and syntactic skills increased linearly for all participants, without any difference between lower and higher achieving children. Vocabulary depth grew more over time for those children with deeper vocabulary knowledge and higher levels of current English input at the start of the study. All of the bivariate growth models showed a correlation between vocabulary and grammar, but failed to show any correlation between their growth. Significant relationships between the intercept of grammar and the growth of vocabulary showed steeper lexical growth in children with better grammar skills. Length of exposure to English had an effect on morphological and syntactic skills, while only current English input had an effect on vocabulary depth.

Controlling hallucinations at word level in data-to-text generation

Data-to-Text Generation (DTG) is a subfield of Natural Language Generation aiming at transcribing structured data in natural language descriptions. The field has been recently boosted by the use of neural-based generators which exhibit on one side great syntactic skills without the need of hand-crafted pipelines; on the other side, the quality of the generated text reflects the quality of the training data, which in realistic settings only offer imperfectly aligned structure-text pairs. Consequently, state-of-art neural models include misleading statements –usually called hallucinations—in their outputs. The control of this phenomenon is today a major challenge for DTG, and is the problem addressed in the paper. Previous work deal with this issue at the instance level: using an alignment score for each table-reference pair. In contrast, we propose a finer-grained approach, arguing that hallucinations should rather be treated at the word level. Specifically, we propose a Multi-Branch Decoder which is able to leverage word-level labels to learn the relevant parts of each training instance. These labels are obtained following a simple and efficient scoring procedure based on co-occurrence analysis and dependency parsing. Extensive evaluations, via automated metrics and human judgment on the standard WikiBio benchmark, show the accuracy of our alignment labels and the effectiveness of the proposed Multi-Branch Decoder. Our model is able to reduce and control hallucinations, while keeping fluency and coherence in generated texts. Further experiments on a degraded version of ToTTo show that our model could be successfully used on very noisy settings.