Low-frequency Items Research Articles

Zipfian distributions facilitate children's learning of novel word-referent mappings

The word-frequency distributions children hear during language learning are highly skewed (Zipfian). Previous studies suggest that such skewed environments confer a learnability advantage in tasks that require the learner to discover the units that have to be learned, as in word-segmentation or cross-situational learning. This facilitative effect has been attributed to contextual facilitation from high frequency items in learning lower frequency items, and to better learning under the increased predictability (lower entropy) of skewed distributions. Here, we ask whether Zipfian distributions facilitate learning beyond the discovery of units, as expected under the predictability account. We tested children's learning of novel word-referent mappings in a learning task where each mapping was presented in isolation during training, and did not need to be dicovered. We compared learning in a uniform environment to two skewed environments with different entropy levels. Children's learning was overall better in the two skewed environments, even for low frequency items. These results extend the facilitative effect of Zipfian distributions to additional learning tasks and show they can facilitate language learning beyond the discovery of units.

CoreRec: A Counterfactual Correlation Inference for Next Set Recommendation

Next set recommendation aims to predict the items that are likely to be bought in the next purchase. Central to this endeavor is the task of capturing intra-set and cross-set correlations among items. However, the modeling of cross-set correlations poses challenges due to specific issues. Primarily, these correlations are often implicit, and the prevailing approach of establishing an indiscriminate link across the entire set of objects neglects factors like purchase frequency and correlations between purchased items. Such hastily formed connections across sets introduce substantial noise. Additionally, the preeminence of high-frequency items in numerous sets could potentially overshadow and distort correlation modeling with respect to low-frequency items. Thus, we devoted to mitigating misleading inter-set correlations. With a fresh perspective rooted in causality, we delve into the question of whether correlations between a particular item and items from other sets should be relied upon for item representation learning and set prediction. Technically, we introduce the Counterfactual Correlation Inference framework for next set recommendation, denoted as CoreRec. This framework establishes a counterfactual scenario in which the recommendation model impedes cross-set correlations to generate intervened predictions. By contrasting these intervened predictions with the original ones, we gauge the causal impact of inter-set neighbors on set prediction—essentially assessing whether they contribute to spurious correlations. During testing, we introduce a post-trained switch module that selects between set-aware item representations derived from either the original or the counterfactual scenarios. To validate our approach, we extensively experiment using three real-world datasets, affirming both the effectiveness of CoreRec and the cogency of our analytical approach.

Genotyping of SNPs in bread wheat at reduced cost from pooled experiments and imputation

Key messagePooling and imputation are computational methods that can be combined for achieving cost-effective and accurate high-density genotyping of both common and rare variants, as demonstrated in a MAGIC wheat population.The plant breeding industry has shown growing interest in using the genotype data of relevant markers for performing selection of new competitive varieties. The selection usually benefits from large amounts of marker data, and it is therefore crucial to dispose of data collection methods that are both cost-effective and reliable. Computational methods such as genotype imputation have been proposed earlier in several plant science studies for addressing the cost challenge. Genotype imputation methods have though been used more frequently and investigated more extensively in human genetics research. The various algorithms that exist have shown lower accuracy at inferring the genotype of genetic variants occurring at low frequency, while these rare variants can have great significance and impact in the genetic studies that underlie selection. In contrast, pooling is a technique that can efficiently identify low-frequency items in a population, and it has been successfully used for detecting the samples that carry rare variants in a population. In this study, we propose to combine pooling and imputation and demonstrate this by simulating a hypothetical microarray for genotyping a population of recombinant inbred lines in a cost-effective and accurate manner, even for rare variants. We show that with an adequate imputation model, it is feasible to accurately predict the individual genotypes at lower cost than sample-wise genotyping and time-effectively. Moreover, we provide code resources for reproducing the results presented in this study in the form of a containerized workflow.

Validation of two measures for assessing English vocabulary knowledge on web-based testing platforms: long-form assessments.

The goal of the current work was to develop and validate web-based measures for assessing English vocabulary knowledge. Two existing paper-and-pencil assessments, the Vocabulary Size Test (VST) and the Word Familiarity Test (WordFAM), were modified for web-based administration. In Experiment 1, participants (n=100) completed the web-based VST. In Experiment 2, participants (n=100) completed the web-based WordFAM. Results from these experiments confirmed that both tasks (1) could be completed online, (2) showed expected sensitivity to English frequency patterns, (3) exhibited high internal consistency, and (4) showed an expected range of item discrimination scores, with low frequency items exhibiting higher item discrimination scores compared to high frequency items. This work provides open-source English vocabulary knowledge assessments with normative data that researchers can use to foster high quality data collection in web-based environments.

A benchmark for evaluating Arabic word embedding models

AbstractModelling the distributional semantics of such a morphologically rich language as Arabic needs to take into account its introflexive, fusional, and inflectional nature attributes that make up its combinatorial sequences and substitutional paradigms. To evaluate such word distributional models, the benchmarks that have been used thus far in Arabic have mimicked those in English. This paper reports on a benchmark that we designed to reflect linguistic patterns in both Contemporary Arabic and Classical Arabic, the first being a cover term for written and spoken Modern Standard Arabic, while the second for pre-modern Arabic. The analogy items we included in this benchmark are chosen in a transparent manner such that they would capture the major features of nouns and verbs; derivational and inflectional morphology; high-, middle-, and low-frequency patterns and lexical items; and morphosemantic, morphosyntactic, and semantic dimensions of the language. All categories included in this benchmark are carefully selected to ensure proper representation of the language. The benchmark consists of 45 roots of the trilateral, all-consonantal, and semivowel-inclusive types; six morphosemantic patterns (’af‘ala; ifta‘ala; infa‘ala; istaf‘ala; tafa‘‘ala; and tafā‘ala); five derivations (the verbal noun, active participle, and the contrasts in Masculine-Feminine; Feminine-Singular-Plural; Masculine-Singular-Plural); and morphosyntactic transformations (perfect and imperfect verbs conjugated for all pronouns); and lexical semantics (synonyms, antonyms, and hyponyms of nouns, verbs, and adjectives), as well as capital cities and currencies. All categories include an equal proportion of high-, medium-, and low-frequency items. For the purpose of validating the proposed benchmark, we developed a set of embedding models from different textual sources. Then, we tested them intrinsically using the proposed benchmark and extrinsically using two natural language processing tasks: Arabic Named Entity Recognition and Text Classification. The evaluation leads to the conclusion that the proposed benchmark is truly reflective of this morphologically rich language and discriminatory of word embeddings.

A multi-scale evolutionary deep learning model based on CEEMDAN, improved whale optimization algorithm, regularized extreme learning machine and LSTM for AQI prediction

With the rapid development of economy, air pollution occurs frequently, which has a huge negative impact on human health and urban ecosystem. Air quality index (AQI) can directly reflect the degree of air pollution. Accurate AQI trend prediction can provide reliable information for the prevention and control of air pollution, but traditional forecasting methods have limited performance. To this end, a dual-scale ensemble learning framework is proposed for the complex AQI time series prediction. First, complete ensemble empirical mode decomposition adaptive noise (CEEMDAN) and sample entropy (SE) are used to decompose and reconstruct AQI series to reduce the difficulty of direct modeling. Then, according to the characteristics of high and low frequencies, the high-frequency components are predicted by the long short-term memory neural network (LSTM), and the low-frequency items are predicted by the regularized extreme learning machine (RELM). At the same time, the improved whale optimization algorithm (WOA) is used to optimize the hyper-parameters of RELM and LSTM models. Finally, the hybrid prediction model proposed in this paper predicts the AQI of four cities in China. This work effectively improves the prediction accuracy of AQI, which is of great significance to the sustainable development of the cities.

Efficient top/bottom-k fraction estimation in spatial databases using bounded main memory

Spatial databases store objects with their locations and certain types of attached items. A variety of modern applications have been developed by leveraging the utilization of locations and items in spatial objects, such as searching points of interest, hot topics, or users' attitude in specified spatial regions. In many scenarios, the high and low-frequency items in a spatial region are worth noticing, considering they represent the majority's interest or eccentric users' opinion. However, existing works have yet to identify such items in an interactive manner, despite the significance of the endeavor in decision-making systems. This study recognizes a novel type of analytical query, called top/bottom-k fraction query, to discover such items in spatial databases. To achieve fast query response, we propose a multilayered data summary that is spread out across the main memory and external memory. A memory-based estimation method for top/bottom-k fraction queries is proposed. To maximize the use of the main memory space, we design a data summary tuning method to dynamically allocate memory space among different spatial partitions. The proposed approach is evaluated with real-life datasets and synthetic datasets in terms of estimation accuracy. Evaluation results demonstrate the effectiveness of the proposed data summary and corresponding estimation and tuning algorithms.

A probabilistic model for co-occurrence analysis in bibliometrics.

The co-occurrence analysis of Medical Subject Heading (MeSH) terms extracted from the PubMed database is popularly used in bibliometrics. Practically for making the result interpretable, it is necessary to apply a certain filter procedure of co-occurrence matrix for removing the low-frequency items due to their low representativeness. Unfortunately, there is rare research referring to determine a critical threshold to remove the noise of co-occurrence matrix. Here, we proposed a probabilistic model for co-occurrence analysis that can provide statistical inferences about whether the paired items co-occur randomly. With help of this model, the dimensionality of co-occurrence matrix could be reduced according to the selected threshold. The conceptual model framework, simulation and practical applications are illustrated in the manuscript. Further details (including all reproducible codes) can be downloaded from the project website: https://github.com/xizhou/co-occurrence-analysis.git.

The signed mental lexicon: Effects of phonological neighborhood density, iconicity, and childhood language experience

Most of what is known about the mental lexicon comes from studies of spoken language and their written forms. Signs differ from spoken/written words in two important ways that may affect lexical recognition: their phonological composition is unique (e.g., more simultaneous than serial structure; few minimal pairs) and many signs are iconic. Using an unprimed lexical decision task in American Sign Language (ASL) and the first available estimates of phonological neighborhood density for any sign language, we found that phonological neighborhood density had an inhibitory effect on latency among low frequency signs. This is the first clear evidence that phonological neighbors spontaneously compete during sign recognition. Iconicity negatively affected accuracy but not reaction times, suggesting that iconicity plays a role in task-related decision processes but not lexical retrieval. Many deaf signers have delayed first language acquisition, and we found that language deprivation had lasting, negative effects on phonological processing and sign recognition speed and accuracy. This work indicates that the lexicons of both spoken and signed languages are organized by form, that lexical recognition occurs through form-based competition (most evident for low frequency items), and that form-meaning mappings do not drive lexical access even when iconicity is pervasive in the lexicon.

Research on the Live Broadcast Operation Strategy of Agricultural Products E-commerce

With the national rural revitalization programme and the development of e-commerce in rural areas, live broadcast of agricultural products has bccome hotspots recently. Farmers hope to catch the express train of live broadcastm, sell agricultural products and broaden the road towards wealth. However, there are constraints on internalconditions and external environment on live streaming. That is, in terms of internal conditions, live broadcast of agricultural products is restricted by product attributes, live streaming talent and online netwok technology and operating experience. In terms of external environment, with the competition and reconstruction of the online live broadcast platforms in 2020, the threshold for live brodcast will be raised, and agricultural products are still unpopular in the live broadcast e-commerce industry. Live streaming of agricultural products is in those low-frequency items. The article proposes an operatoin strategy in the live broadcast for agricultural products of average farmer households based on SWOT analysis, and then proposes its implementation principle of minimizing constraint conditions + maximizing product value, which can help average farmers use e-commerce channels to operate a practicable livebroadcast.

Learned sketches for frequency estimation

The Count-Min sketch and its variations are widely used to solve the frequency estimation problem due to its sub-linear space cost. However, the collisions between high-frequency and low-frequency items introduce a significant estimation error. In this paper, we propose two learned sketches called the Learned Count-Min sketch and Learned Augmented sketch. We combine the machine learning methods with the traditional Count-Min sketch and Augmented sketch to improve the performance. We used a regression model trained from historical data to predict the frequencies and separate the high-frequency items and low-frequency items. The experimental results indicated that our learned sketches outperform the traditional Count-Min sketch and Augmented sketch. The learned sketches can provide a more accurate estimation with a more compact synopsis size.

Frequency effects in palatalization: Articulatory evidence from English

Lexical frequency and its influence on speech production have been widely acknowledged across languages and varieties, and frequency effects are so prevalent in various phonological and coarticulatory phenomena. Previous experimental studies have reported that high lexical frequency leads a greater chance of overt coarticulation (for example, more frequent overt palatalization of /d/ in would you (high-frequency) than bad you (low-frequency)). This study tests for articulatory evidence of frequency effects in palatalization by examining palatalized consonants resulted by lexical and post-lexical palatalization in English using ultrasound imaging. Comparisons of tongue contours of palatalized consonants produced by 12 native speakers of American English show that differences in lexical frequency are not directly linked to articulatory gestures. That is, speakers do not necessarily produce a greater degree of palatalization (i.e., tongue contour closer to the palate) in high-frequency words or phrases. Moreover, while most speakers maintain some degree of articulatory contrast in high- vs. low-frequency contexts, the way speakers differentiate palatalized consonants from high- vs. low-frequency items is highly individualized. The findings from this study are in line with individual variation in coarticulation, and merit further exploration in the role of lexical frequency in speech production.Lexical frequency and its influence on speech production have been widely acknowledged across languages and varieties, and frequency effects are so prevalent in various phonological and coarticulatory phenomena. Previous experimental studies have reported that high lexical frequency leads a greater chance of overt coarticulation (for example, more frequent overt palatalization of /d/ in would you (high-frequency) than bad you (low-frequency)). This study tests for articulatory evidence of frequency effects in palatalization by examining palatalized consonants resulted by lexical and post-lexical palatalization in English using ultrasound imaging. Comparisons of tongue contours of palatalized consonants produced by 12 native speakers of American English show that differences in lexical frequency are not directly linked to articulatory gestures. That is, speakers do not necessarily produce a greater degree of palatalization (i.e., tongue contour closer to the palate) in high-frequency words or phrases. More...

Evidence for degraded low frequency verbal concepts in left resected temporal lobe epilepsy patients

According to a large neuropsychological and neuroimaging literature, the bilateral anterior temporal lobe (ATL) is a core region for semantic processing. It seems therefore surprising that semantic memory appears to be preserved in temporal lobe epilepsy (TLE) patients with unilateral ATL resection. However, recent work suggests that the bilateral semantic system is relatively robust against unilateral damage and semantic impairments under these circumstances only become apparent with low frequency specific concepts. In addition, neuroimaging studies have shown that the function of the left and right ATLs differ and therefore left or right ATL resection should lead to a different pattern of impairment. The current study investigated hemispheric differences in the bilateral semantic system by comparing left and right resected TLE patients during verbal semantic processing of low frequency concepts. Picture naming and semantic comprehension tasks with varying word frequencies were included to investigate the pattern of impairment. Left but not right TLE patients showed impaired semantic processing, which was particularly apparent on low frequency items. This indicates that, for verbal information, the bilateral semantic system is more sensitive to damage in the left compared to the right ATL, which is in line with theories that attribute a more prominent role to the left ATL due to connections with pre-semantic verbal regions.

Centrality ratings, forensic relevance, and production frequency: Which one best identifies central and peripheral items?

SummaryProduction frequency has often been used to identify central and peripheral information, under the assumption that high frequency implies that the item is central. However, no research to date has tested the relationship between centrality and frequency. Participants watched a video of a bank robbery and completed a free recall test, from which frequency for recalled items was computed. Two groups then watched the same video and rated centrality and forensic relevance for each item. Results showed that most, but not all, items with high frequency were rated as central and forensically relevant but that low frequency items were not diagnostic of either item centrality or forensic relevance. Forensic relevance was a better indicator of item centrality than frequency. We concluded that frequency measures should be avoided to determine centrality. Also, if centrality ratings cannot be collected, forensic relevance ratings may be more appropriate for this purpose.

The Effect of Lexical Frequency in Modulating Translation Asymmetric on L2 Learners

Mao Cui, Eunhae Oh. 2018. The effect of lexical frequency in modulating translation asymmetries on L2 learners. Language and Information, 22.1, 57-71. The study examined response time in Chinese-English two-way translation in order to explore the role of word frequency in modulating translation asymmetries between forward and backward translation. 23 Chinese learners of English at two different proficiency levels performed a word translation task of 90 concrete nouns, one-syllable in English words and two characters in Chinese words. The results showed that there were asymmetries between the two translation directions, with backward translation taking significantly longer time than forward translation. Word frequency played a role in modulating the asymmetries in that translating high-frequency items, regardless of direction, took significantly shorter time than the mid- and low-frequency items. But backward translation displayed larger magnitude of sensitivity to the variation of frequency than forward translation. It was suggested that both forward and backward translation were conceptually-mediated with differing strength in the linking between the lexical items and their conceptual representations, being stronger with high-frequency items than mid- and low-frequency items. Conceptual activation is easier for high frequency items.

Distribution and duration of signs and parts of speech in Swedish Sign Language

In this paper, we investigate frequency and duration of signs and parts of speech in Swedish Sign Language (SSL) using the SSL Corpus. The duration of signs is correlated with frequency, with high-frequency items having shorter duration than low-frequency items. Similarly, function words (e.g. pronouns) have shorter duration than content words (e.g. nouns). In compounds, forms annotated as reduced display shorter duration. Fingerspelling duration correlates with word length of corresponding Swedish words, and frequency and word length play a role in the lexicalization of fingerspellings. The sign distribution in the SSL Corpus shows a great deal of cross-linguistic similarity with other sign languages in terms of which signs appear as high-frequency items, and which categories of signs are distributed across text types (e.g. conversation vs. narrative). We find a correlation between an increase in age and longer mean sign duration, but see no significant difference in sign duration between genders.

Early Detection of English Attrition in Korean-English Bilingual Children

This longitudinal case study utilized a body-part picture-naming task (Hawaii Assessment of Language Access project) to assess the changing proficiencies and dominance of Korean and English from two Korean sisters who returned back to Korea after living in Honolulu, Hawaii for two years. The task consisted of two strata based on the frequency of body-part vocabularies: high-frequency and low-frequency items. Children's accuracy and response time were gauged to demonstrate the participants' relative proficiency in English and Korean over three sessions starting from 18 months after their return until the 25th month after return. Since infrequent language use limits accessibility to that language, which in turn eventually causes further language loss(O'Grady et al., 2009), less frequently used words would be more vulnerable to proficiency change than high-frequency words. The result of this study suggests that lexical accessibility is one of the earliest linguistic aspects influenced by attrition, which can show up as early as around 21 months after their return to Korea.

Frequency effects and structural change - the Afrikaans preterite

According to emergent grammar and exemplar theory in cognitive linguistics, the frequency of an item affects its behaviour in terms of structural change. In this article, I illustrate how high frequency items, such as preterital modal auxiliaries and copulas in Afrikaans, resist regularising with the rest of the Afrikaans verbal system. Items with a moderately high frequency can resist change for a time, but succumb to it eventually, such as mog (“might”) and wis (“knew”). While the course of change can also be affected by other factors, such as het (“have”) and had (“had”), and dink (“think”) and gedink/dag/dog (“thought”) show, the data in diachronic Afrikaans corpora from 1911 to 2010 confirm that high frequency items resist structural change to a large extent, while low frequency items do not. This links with the cognitive representation of language and language processing, and illustrates how the use of language shapes the structure of language.

Effects of first language processes and representations on second language perception: The case of vowel epenthesis by Japanese speakers

Aims and objectives/purpose/research questions: Japanese speakers are known to perceive “illusory vowels” within consonant clusters illicit in their language. The present study examines how this perceptual vowel epenthesis is affected by first language (L1) processes (restoration of vowels devoiced through Japanese high vowel devoicing), L1 representations (loanword representations in Japanese speakers’ lexicons), and proficiency in English. Design/methodology/approach: The participants judged the presence or absence of a mora (e.g., ム /mu/) in an auditorily presented English word (e.g., homesick). The 40 test items contained a heterosyllabic consonant cluster with four different voicing patterns to examine whether the vowel restoration process is related to vowel epenthesis. Twenty of the test items are frequently used as loanwords in Japanese, meaning that they are stored in the L1 lexicon with a vowel inserted inside the consonant cluster (e.g., /hoomusikku/). The other 20 are low-frequency items that are virtually nonwords for the non-native participants. Data and analysis: The vowel epenthesis rates and reaction times (RTs) were obtained from 14 introductory learners, 15 intermediate learners, and 19 native speakers. Findings/conclusions: The results show the main effects of Voice, Loanword Representation, and Proficiency, as well as the interaction among the three factors. Negative correlations between vowel epenthesis rates and RTs were also observed for the learners. The results indicate differential effects of vowel restoration and loanwords on perceptual epenthesis by learners of different proficiency levels. Originality: The present study was one of the first attempts to test the relation between proficiency and perceptual vowel epenthesis using real English words. Significance/implications: The findings demonstrate the robustness of L1 processes and representations in second language perception while substantiating the existing argument for early vowel epenthesis. They also raise questions regarding the effects of training and the role of native speaker input.

Segregating attention from response control when performing a motor inhibition task: Segregating attention from response control

Considerable work has demonstrated that inferior frontal gyrus (IFG), anterior insula cortex (AIC) and the supplementary motor area (SMA) are responsive during inhibitory control tasks. However, there is disagreement as to whether this relates to response selection/ inhibition or attentional processing. The current study investigates this by using a Go/No-go task with a factorial design. We observed that both left IFG and dorsal pre-SMA were responsive to no-go cues irrespective of cue frequency. This suggests a role for both in the inhibition of motor responses. Generalized psychophysiological interaction (gPPI) analyses suggest that inferior frontal gyrus may implement this function through interaction with basal ganglia and by suppressing the visual representation of cues associated with no-go responses. Anterior insula cortex and a more ventral portion of pre-SMA showed greater responsiveness to low frequency relative to higher frequency stimuli, irrespective of response type. This may reflect the hypothesized role of anterior insula cortex in marking low frequency items for additional processing (cf. Menon and Uddin, 2010). Consistent with this, the gPPI analysis revealed significantly greater anterior insula cortex connectivity with visual cortex in response to low relative to high frequency cues.