Rare Words Research Articles

Philippians, 3, 8 in Romanian language: a case of primacy of norm

In the present paper, the study of the variants given by the translators of the bible text into the Romanian language during the old age and contemporary to the verse under Philippians, 3, 8, which contains in the Greek version a rare word in the Bible (σκῠ́βᾰλον), highlights the fact that the act of translation places the text on the path of a language, subjecting it to the state of the language at a given moment and to its evolution. At any moment, there is a formal adjustment, which follows what is considered to be correct or at least linguistically possible: first, it is that of the foreign text to the new language—Romanian; then, it is that of the Romanian text to the renewals of the Romanian language—adjustment that may ignore the original requirements, from a given moment in the history of producing and receiving that text in the Romanian language and culture; thus, the correctness of the text is controlled by the norm of the language and by the appearance of its semantic accuracy. In the last sequence of Phil, 3. 8, the history of rendering the element of comparison in the Romanian language constitutes a very good case of primacy of the norm of the language over the originary content (in relation to the direct source text)—recoverable, but exclusively by metalinguistic means.

Words ranking and Hirsch index for identifying the core of the hapaxes in political texts

This paper deals with a quantitative analysis of the content of official political speeches. We study a set of about one thousand talks pronounced by the US Presidents, ranging from Washington to Trump. In particular, we search for the relevance of the rare words, i.e. those said only once in each speech -- the so-called hapaxes. We implement a rank-size procedure of Zipf-Mandelbrot type for discussing the hapaxes' frequencies regularity over the overall set of speeches. Starting from the obtained rank-size law, we define and detect the core of the hapaxes set by means of a procedure based on an Hirsch index variant. We discuss the resulting list of words in the light of the overall US Presidents' speeches. We further show that this core of hapaxes itself can be well fitted through a Zipf-Mandelbrot law and that contains elements producing deviations at the low ranks between scatter plots and fitted curve -- the so-called king and vice-roy effect. Some socio-political insights are derived from the obtained findings about the US Presidents messages.

Bilingual novel word learning in sentence contexts: Effects of semantic and language variation

Abstract Research on word learning demonstrates that meanings of novel words are better recalled when those words are encountered in a variety of semantic contexts than when encountered in the same repeated context. The present study examines how semantic and non-semantic contextual variations in word learning encounters impact learning the meanings of novel words. Across two experiments, highly proficient Spanish-English bilinguals studied rare words in repeated or varied semantic contexts, presented in only one language or in both languages. Participants recalled more word meanings when semantic context was varied rather than repeated at study, and this effect did not differ for the more and less proficient languages. However, varying the language context provided no benefit relative to studying words in only one language. Performance patterns were consistent for visual and auditory study modalities, and in both recall and recognition tasks, performance was less accurate overall in the less proficient language than in the more proficient language. These findings replicate the advantage of semantic variation observed in monolingual studies and is the first to directly compare learning from varied and repeated semantic contexts for more and less proficient languages in bilinguals.

A Hierarchical Clustering Approach to Fuzzy Semantic Representation of Rare Words in Neural Machine Translation

Rare words are usually replaced with a single $ $unk$ $>$ token in the current encoder–decoder style of neural machine translation, challenging the translation modeling by an obscured context. In this article, we propose to build a fuzzy semantic representation (FSR) method for rare words through a hierarchical clustering method to group rare words together, and integrate it into the encoder–decoder framework. This hierarchical structure can compensate for the semantic information in both source and target sides, and providing fuzzy context information to capture the semantic of rare words. The introduced FSR can also alleviate the data sparseness, which is the bottleneck in dealing with rare words in neural machine translation. In particular, our method is easily extended to the transformer-based neural machine translation model and learns the FSRs of all in-vocabulary words to enhance the sentence representations in addition to rare words. Our experiments on Chinese-to-English translation tasks confirm a significant improvement in the translation quality brought by the proposed method.

Mothers’ Sophisticated Vocabulary Input and Children’s Story Comprehension

ABSTRACT Research Findings: The purpose of this study was to examine mothers’ sophisticated vocabulary while reminiscing with their preschool-aged children, and its relation to children’s story comprehension. The study used a cross-lagged panel design in which all assessments were collected twice, 6 months apart. We also compared two methods of examining sophisticated vocabulary – rare words (i.e., words not on the Dale-Chall list) and word tiers (i.e., Tier 2 and Tier 3 words). We found that mothers’ rare words predicted children’s later story comprehension controlling for children’s age, mothers’ education, and children’s receptive vocabulary. Mothers’ Tier 3 words were related concurrently to children’s story comprehension at time 2. Children’s story comprehension did not predict mothers’ later sophisticated vocabulary suggesting that this is not a bidirectional relation. Reasons for why mothers’ sophisticated vocabulary input in the context of reminiscing might matter to children’s story comprehension are discussed. Practice or Policy: These results suggest that researchers and educators may wish to examine the impact of vocabulary instruction on story comprehension in pre-readers in addition to vocabulary outcomes given the importance of early narrative comprehension to later reading comprehension.

Rare Words: A Major Problem for Contextualized Embeddings and How to Fix it by Attentive Mimicking

Pretraining deep neural network architectures with a language modeling objective has brought large improvements for many natural language processing tasks. Exemplified by BERT, a recently proposed such architecture, we demonstrate that despite being trained on huge amounts of data, deep language models still struggle to understand rare words. To fix this problem, we adapt Attentive Mimicking, a method that was designed to explicitly learn embeddings for rare words, to deep language models. In order to make this possible, we introduce one-token approximation, a procedure that enables us to use Attentive Mimicking even when the underlying language model uses subword-based tokenization, i.e., it does not assign embeddings to all words. To evaluate our method, we create a novel dataset that tests the ability of language models to capture semantic properties of words without any task-specific fine-tuning. Using this dataset, we show that adding our adapted version of Attentive Mimicking to BERT does substantially improve its understanding of rare words.

A hybrid representation-based simile component extraction

Simile, a special type of metaphor, can help people to express their ideas more clearly. Simile component extraction is to extract tenors and vehicles from sentences. This task has a realistic significance since it is useful for building cognitive knowledge base. With the development of deep neural networks, researchers begin to apply neural models to component extraction. Simile components should be in cross-domain. According to our observations, words in cross-domain always have different concepts. Thus, concept is important when identifying whether two words are simile components or not. However, existing models do not integrate concept into their models. It is difficult for these models to identify the concept of a word. What’s more, corpus about simile component extraction is limited. There are a number of rare words or unseen words, and the representations of these words are always not proper enough. Exiting models can hardly extract simile components accurately when there are low-frequency words in sentences. To solve these problems, we propose a hybrid representation-based component extraction (HRCE) model. Each word in HRCE is represented in three different levels: word level, concept level and character level. Concept representations (representations in concept level) can help HRCE to identify the words in cross-domain more accurately. Moreover, with the help of character representations (representations in character levels), HRCE can represent the meaning of a word more properly since words are consisted of characters and these characters can partly represent the meaning of words. We conduct experiments to compare the performance between HRCE and existing models. The experiment results show that HRCE significantly outperforms current models.

The Core Vocabulary: The Foundation of Proficient Comprehension

AbstractA group of words, labeled the core vocabulary, can be expected to be prominent across all texts. Scholarship made possible by digital databases of words and new analytic systems has shown that approximately 2,500 morphological families account for most of the words in texts—an average of 91.5% of all words in the Common Core State Standards exemplars from kindergarten through College and Career Ready. The rare vocabulary of texts is so extensive and varied that, without a foundation in the core vocabulary, students are likely to struggle with the meanings of rare words. Words are in the core vocabulary because they represent critical concepts, are part of morphological or word families that share a root word, and are often versatile in their meaning and function. The author describes the semantic, morphological, and multiple‐meaning knowledge represented by these 2,500 word families and ways in which teachers can foster these proficiencies.

Optimizing segmentation granularity for neural machine translation

In neural machine translation (NMT), it has become standard to translate using subword units to allow for an open vocabulary and improve accuracy on infrequent words. Byte-pair encoding (BPE) and its variants are the predominant approach to generating these subwords, as they are unsupervised, resource-free, and empirically effective. However, the granularity of these subword units is a hyperparameter to be tuned for each language and task, using methods such as grid search. Tuning may be done inexhaustively or skipped entirely due to resource constraints, leading to sub-optimal performance. In this paper, we propose a method to automatically tune this parameter using only one training pass. We incrementally introduce new BPE vocabulary online based on the held-out validation loss, beginning with smaller, general subwords and adding larger, more specific units over the course of training. Our method matches the results found with grid search, optimizing segmentation granularity while significantly reducing overall training time. We also show benefits in training efficiency and performance improvements for rare words due to the way embeddings for larger units are incrementally constructed by combining those from smaller units.

Explanations of Three Rare Words in the Tale Known as Úath Beinne Étair and a Re-Assessment of this Title

Explanations of Three Rare Words in the Tale Known as Úath Beinne Étair and a Re-Assessment of this Title

Explanations of Three Rare Words in the Tale Known as <em>Úath Beinne Étair</em> and a Re-Assessment of this Title

Explanations of the three rare words sennin, sincreth and nemceissi in a tale to which the title Úath Beinne Étair has been assigned are proposed. It is also argued that there is little justification for assigning this title to the tale.

«Словарь редкой лексики по произведениям школьной программы»: методика составления словника и основные принципы лексикографического описания

Given the current multi-faceted crisis of the reading culture, a crisis to which the secondary education system is especially sensitive, it is crucial to foster a culture of fiction reading. The fact that most literary works in the school curriculum were written more than 100 years ago impedes their understanding by contemporary young readers. This is a situation that can be improved by the very mobile technologies which have become an integral part of our lives. This article describes the basic principles applied during the compilation of the Dictionary of Rare Words from the Literary Works Covered by the School Curriculum (hereinafter the Dictionary). The Dictionary is designed as a part of a cross-platform mobile application which should increase motivation for learning and the efficacy of secondary education. The Dictionary is easy to use due to its integration into the electronic library of works covered by the school literature curriculum according to the basic component of the Federal State Educational Standard. The word list for the Dictionary was compiled using the tools of corpus linguistics that allowed us to contextualise the work and to include all the necessary semantic components of the lexeme into the dictionary entry. After comparing the concordance based on the corpus of Russian literature for school education with the word list of The New Colloquial Dictionary of Russian Vocabulary by O. N. Lyashevskaya and S. A. Sharov, we identified more than 20,000 lexical units not included into the colloquial dictionary that would therefore benefit from lexicographic description. Nouns, which numbered about 9,500 entries in this list, were chosen to be the first part-of-speech described in the Dictionary. During the word list compilation, certain principles were employed to reduce the number of nouns. More than 3,100 words belonging to various semantic groups were selected for lexicographic description. The word list includes not only obsolescent words, but also low-frequency lexemes, including proper names and occasional words. This approach contributes to students learning a larger number of terms, develops a culture of reading literary texts among students and broadens their horizons. The Dictionary is based on the principles of educational lexicography. The purpose of a dictionary entry is not just to clarify the meaning of a word and provide information about its correct use but, above all, to create an image of the described realia in the reader’s imagination. To do this, the authors, if necessary, provide amplifications and elements of encyclopaedic information. Numerous images illustrate the content of the dictionary entries.

Father, Womb, Blood: Apollo’s embryological theory, the ethics of revenge, and the supposed exclusion of women in Aischylos’ Eumenides

In Aischylos’ Eumenides, Apollo intimates a theory according to which the father is the sole genetic parent of the child. The status of this conception, and whether it is depicted as an outlandish idea, has been much and inconclusively discussed. This paper considers a neglected piece of evidence: Apollo’s use of the very unusual word αὐτάδελφον when addressing Hermes. In light of the Greeks’ awareness of this etymology as well as the other instances of this rare word in tragedy, the author argues that Aischylos’ text highlights the etymological connection to δελφύς, the womb, thus evoking the role of the mother. This suggests that Aischylos subtly lets his, and Apollo’s, language rebel against the notion of merely paternal kinship, and the concomitant ideas about revenge, retaliation and children’s obligations to their parent.

Forming the image of Other: Religion in the Geopolitical Maps Around Jungar and Kalmyk Khanates

The first third of XVIII century was very important for Kalmyks and Jungars. The political affairs, where Kalmyks, Russians, Jungars and Manchus as well as the external issues have been communicated, had been closely related with the religious matters, led to intensification of regional political processes. In Kalmyk khanate the Buddhist affairs were under the supervision of the famous Shakur Lama, and it was during his rule when the Islamic issue had had some importance, because it had been closely related with the plan to establish the Kalmyk-Crimean khanate (perhaps, initially the Ottomans’ idea). In Jungaria the situation was even more complex due to the policy of Qing, when the leaders of this khanate were accused in conversion to Islam. The rare word “Thomkar” from Russian archival materials illustrates very special skill of Qing officials - it was “created” for Jungars to define them as being “converted” to Islam. The meaning of all these events allow to suppose, that religion at the time under consideration became the state tool for implementation of the virtual geopolitical maps to change the “usual” image of someone into “necessary” “enemy face”.

Lost in Machine Translation: Contextual Linguistic Uncertainty

The article considers the issues related to the semantic, grammatical, stylistic and technical difficulties currently present in machine translation and compares its four main approaches: Rule-based (RBMT), Corpora-based (CBMT), Neural (NMT), and Hybrid (HMT). It also examines some "open systems", which allow the correction or augmentation of content by the users themselves ("crowdsourced translation"). The authors of the article, native speakers presenting different countries (Russia, Greece, Malaysia, Japan and Serbia), tested the translation quality of the most representative phrases from the English, Russian, Greek, Malay and Japanese languages by using different machine translation systems: PROMT (RBMT), Yandex. Translate (HMT) and Google Translate (NMT). The test results presented by the authors show low "comprehension level" of semantic, linguistic and pragmatic contexts of translated texts, mistranslations of rare and culture-specific words,unnecessary translation of proper names, as well as a low rate of idiomatic phrase and metaphor recognition. It is argued that the development of machine translation requires incorporation of literal, conceptual, and content-and-contextual forms of meaning processing into text translation expansion of metaphor corpora and contextological dictionaries, and implementation of different types and styles of translation, which take into account gender peculiarities, specific dialects and idiolects of users. The problem of untranslatability ('linguistic relativity') of the concepts, unique to a particular culture, has been reviewed from the perspective of machine translation. It has also been shown, that the translation of booming Internet slang, where national languages merge with English, is almost impossible without human correction.

Zu ptolemäischen Urkunden

Abstract Notes on the text and interpretation of three documentary papyri: In P.NYU II 45, an official is asked to release a prisoner around harvest time in order that the crown may not loose revenues. The restoration of a line in BGU VIII 1876 adds another instance of the rare word παραχωρήσιμος. The subscription of P.Enteux. 14 is now entirely restored and some suggestions are made for a name in the σῶμα of the petition.

Neural Machine Translation for Low-resource English-Bangla

Neural machine translation has recently been able to gain state-of-the-art translation quality for many language pairs. However, neural machine translation has been less tested for English-Bangla language pair, two linguistically distant and widely spoken languages. In this paper, we apply neural machine translation to the task of English-Bangla translation in both directions and compare it against a standard phrase-based statistical machine translation system. We obtain up to +0.30 and +4.95 BLEU improvement over phrase-based statistical machine translation for English-to-Bangla and Bangla-to-English respectively. Due to low-resource and morphological richness of Bangla, English-Bangla translation task produces a large number of rare words. We apply subword segmentation with byte pair encoding to handle this rare words issue. We obtain up to +0.69 and +0.30 BLEU improvement over baseline neural machine translation for English-to-Bangla and Bangla-to-English respectively. We further investigate our system output for several challenging linguistic properties like subject-verb agreement, noun inflection, long distance reordering and rare words translation. We observe that neural machine translation with and without subword segmentation significantly outperform the phrase-based statistical machine translation system, thus establishing itself as the state-of-the-art technology for low-resource English-Bangla machine translation.

Effective Subword Segmentation for Text Comprehension

Representation learning is the foundation of machine reading comprehension and inference. In state-of-the-art models, character-level representations have been broadly adopted to alleviate the problem of effectively representing rare or complex words. However, character itself is not a natural minimal linguistic unit for representation or word embedding composing due to ignoring the linguistic coherence of consecutive characters inside word. This paper presents a general subword-augmented embedding framework for learning and composing computationally-derived subword-level representations. We survey a series of unsupervised segmentation methods for subword acquisition and different subword-augmented strategies for text understanding, showing that subword-augmented embedding significantly improves our baselines in various types of text understanding tasks on both English and Chinese benchmarks.

Neural Machine Translation: A Review of the Approaches

Neural Machine Translation (NMT) has presented promising results in Machine translation, convincingly replacing the traditional Statistical Machine Translation (SMT). This success of NMT in machine translation tasks therefore projects to more translation tasks using NMT. This paper systematically reviews the hitherto proposed NMT systems since 2014. 86 NMT papers have been selected and reviewed. The peak of NMT systems were proposed in 2016 and the same was the case for many machine translation workshops who provided datasets for NMT tasks. Most of the proposed systems covered English, German, French and Chinese translation tasks. BLEU score accompanied by significance tests has been seen to be the best metric for NMT systems evaluation. Human judgement for fluency and adequacy is also important to support the metrics. There is still room for further improvement in translations regarding rich source translations and rare words. There is also need for extensive NMT works in other languages to maximize the apparent capabilities of NMT systems. RNN Search and Moses are basically used to develop SMT baselines for model comparisons. Results provide futuristic and directional insights into further translation tasks.

Unseen Word Representation by Aligning Heterogeneous Lexical Semantic Spaces

Word embedding techniques heavily rely on the abundance of training data for individual words. Given the Zipfian distribution of words in natural language texts, a large number of words do not usually appear frequently or at all in the training data. In this paper we put forward a technique that exploits the knowledge encoded in lexical resources, such as WordNet, to induce embeddings for unseen words. Our approach adapts graph embedding and cross-lingual vector space transformation techniques in order to merge lexical knowledge encoded in ontologies with that derived from corpus statistics. We show that the approach can provide consistent performance improvements across multiple evaluation benchmarks: in-vitro, on multiple rare word similarity datasets, and invivo, in two downstream text classification tasks.