Lexical Simplification Research Articles

Deep learning approaches to lexical simplification: A survey

AbstractLexical Simplification (LS) is the task of substituting complex words within a sentence for simpler alternatives while maintaining the sentence’s original meaning. LS is the lexical component of Text Simplification (TS) systems with the aim of improving accessibility to various target populations such as individuals with low literacy or reading disabilities. Prior surveys have been published several years before the introduction of transformers, transformer-based large language models (LLMs), and prompt learning that have drastically changed the field of NLP. The high performance of these models has sparked renewed interest in LS. To reflect these recent advances, we present a comprehensive survey of papers published since 2017 on LS and its sub-tasks focusing on deep learning. Finally, we describe available benchmark datasets for the future development of LS systems.

Over-lexicalization and Under-lexicalization of Physical Violence Expression in Laut Bercerita and Its Translation by Leila S. Chudori

Although research on lexicalization has been extensively documented, little is known about over- and under-lexicalization in translation. "Laut Bercerita" and its translation contain mostly violent physical expressions, leading to either excessive or deficient lexical representation compared to the source text. This research aims to analyze over- and under-lexicalization in expressions of physical violence in the novel "Laut Bercerita" and its translation "The Sea Speaks His Name". The research method used is comparative descriptive, which combines comparing the meanings of the source text and target text. Data were collected using note-taking techniques. In the analysis stage, the data were first segmented or classified based on categories of physical violence. Subsequently, the data were analyzed by adopting a lexical variation of Fowler’s theory. The results of the study showed that over-lexicalization occurred in the form of physical violence such as (1) kicking, (2) torturing, (3) electric shocking, (4) slapping, (5) punching, and (6) stepping on, accounting for 47%. Meanwhile, under-lexicalization occurred in the form of physical violence such as (1) hitting, (2) kicking, (3) torturing, (4) electric shocking, (5) handcuffing, (6) punching, and (7) stepping on, accounting for 57%. Under-lexicalization occurs when there is a reduction of lexical items and simplification of meaning concepts, while over-lexicalization occurs with the addition of lexical items and complex meanings. Over-lexicalization and under-lexicalization imply that the translation may not meet standard translation norms through lexicon choices but still maintains equivalence.

Comparative analysis of lexical simplification in Hungarian-English translated and interpreted texts

Comparative analysis of lexical simplification in Hungarian-English translated and interpreted texts

Analyzing Lexical Simplification in Interviews of Donald Trump and Joe Biden: Tracking the Trend of Simplification in Presidential Discourse Using Lexical Sophistication Indices

Analyzing Lexical Simplification in Interviews of Donald Trump and Joe Biden: Tracking the Trend of Simplification in Presidential Discourse Using Lexical Sophistication Indices

Lexical simplification via single-word generation

Lexical simplification via single-word generation

EASIER corpus: A lexical simplification resource for people with cognitive impairments.

Thanks to technologies such as the Internet and devices now available to people, we have increasingly greater access to larger quantities of information. However, people with ageing disabilities or intellectual disabilities, non-native speakers, and others have difficulties reading and understanding information. For this reason, it is essential to provide text simplification mechanisms when accessing information. Natural Language Processing methods can be applied to simplify textual content and improve understanding. These methods often use machine learning algorithms and models which require resources, such as corpora, to be trained and tested. This article presents the EASIER corpus, a resource that can be used to build lexical simplification methods to process Spanish domain-independent texts. The EASIER corpus is composed of 260 annotated documents with 8,155 words labelled as complex and 5,130 words with at least one proposed context-aware synonym associated. Expert linguists in easy-to-read and plain language guidelines have annotated the corpus based on their experience adapting texts for people with intellectual disabilities. Sixteen annotation guidelines that discriminate between complex and simple words have been defined to help other groups of experts to generate new annotations. Additionally, an inter-annotator agreement test was performed to validate the corpus, obtaining a Fleiss Kappa coefficient of 0.641. Furthermore, a qualitative evaluation was conducted with 45 users (including people with intellectual disabilities, elderly people, and a control audience). Complex word identification tasks achieved moderate results, but the synonyms proposed to replace complex words achieved almost perfect ratings. This resource has been integrated into the EASIER platform, a tool that helps people with cognitive impairments and intellectual disabilities to read and understand texts more easily.

Audiovisual mediation through English intralingual and interlingual subtitling

Abstract This paper reports on a case study implemented at the University of Salento and, partially, at the Autonomous University of Barcelona, concerning the use of subtitles as a didactic tool to develop intercultural communication skills. This study examines the intralingual and interlingual translations of the reportage Fortress Italia: Capsized in Lampedusa, about migrant arrivals in Italy and Europe. Although the video is of interest to international viewers, the lack of proper subtitling may undermine its accessibility to non-native English speakers. On the one hand, subtitles do not appear when Standard English is used, so the comprehension of those utterances depends on the receivers’ listening skills; on the other hand, the official retextualizations are characterized by formal register and editorial additions that may affect their readability. For these reasons, an alternative rendering was commissioned to a number of undergraduate and postgraduate students of Translation and Interpreting, in order to enquire into new areas of adoption of English as an International Language and as a Lingua Franca. The analysis of the selected corpus of extracts will pinpoint the strategies of lexical and structural simplification and condensation, along with the selection of specific verb tenses and aspects, which are expected to enhance the envisaged recipients’ understanding of the video’s message. Since these features of English are actively selected, by the subjects that were involved in this research, so as to foster cross-cultural communication between the authors and viewers of the news report, this study contends that specific lingua-franca uses can be activated when subtitling multimodal texts. Hence, the notion of ‘audiovisual mediation’ will be introduced in order to label an approach to audiovisual translation aiming to: (i) make the illocutionary force accessible and acceptable to the envisaged, international audience; and (ii) overcome the conventional associations between dubbing and domestication, and subtitling and foreignization.

Lexical simplification in learner translation: A corpus-based approach

The advance of corpus-based methodology in translation studies has greatly enhanced our understanding of the nature of translational language. While most research efforts have focused on identifying the unique features of translations carried out by professionals, comparatively fewer studies have investigated the linguistic features of student translations. In this corpus-based study, we examine if learner translations carried out by Hong Kong students exhibit lexical simplification features vis-à-vis comparable written texts. The study is based on two comparable corpora: the International Corpus of English in Hong Kong (ICE-HK) and the Parallel Learner Translation Corpus (PLTC) compiled at The Hong Kong Polytechnic University. Following Laviosa (1998), we compare four main lexical features (lexical density, type-token ratio, core vocabulary coverage, and list head coverage) to investigate if student translations show a simplification trend. The results demonstrate that Chinese-to-English translation is not lexically simpler than English as a Second Language (ESL) writing. Furthermore, it is lexically denser than ESL writing. Our study aims to provide new insights into learner translation as a form of constrained communication.

Do easy-to-read adaptations really facilitate sentence processing for adults with a lower level of education? An experimental eye-tracking study

The Easy-to-Read guidelines recommend visual support and lexical simplification to facilitate text processing, but few studies have empirically verified the efficacy of these guidelines. This study examined the influence of these recommendations on sentence processing by examining eye movements at the text- and word-level in adult readers. We tested 30 non-university adults (low education level) and 30 university adults (high education level). The experimental task consisted of 60 sentences. Half were accompanied by an image and half were not, and half contained a low-frequency word and half a high-frequency word. Results showed that visual support and lexical simplification facilitated processing in both groups of adults, and non-university adults were significantly slower than university adults at sentence processing. However, lexical simplification resulted in faster processing in the non-university adults’ group. Conclusions focus on the mechanisms in which both adaptations benefit readers, and practical implications for reading comprehension.

SimpLex: a lexical text simplification architecture

SimpLex: a lexical text simplification architecture

EASIER System. Evaluating a Spanish Lexical Simplification Proposal with People with Cognitive Impairments

Cognitive impairments in text reading and understanding not only affect individuals with intellectual disabilities but also deaf and deaf-blind ones, the illiterate, and immigrants whose native language is different, among others. Technology to support accessibility to texts is required to help people to understand information. This article addresses an evaluation of the EASIER tool conducted with intellectual disabilities and older people for the identification of complex and unusual words in Spanish texts and proposes them simpler synonyms for the complex word detected. Fifty participants were recruited in collaboration with an organization that works with people with intellectual disabilities. The participants were divided into two groups: Group 1 represented 25 older people, and Group 2 represented 25 people with intellectual disabilities. The evaluation showed that, for complex word identification, the selected texts were more complex in content for people with intellectual disabilities than for older people, which consequently suggests that the proposal had a greater benefit for the intellectually disabled group. Related to the proposal of substitutes for complex words, there is a high number of substitutions accepted in participants with a high school level of education for Group 1 and primary level of education for Group 2. Additional multivariate analysis shows that the system is adequate for people of advanced age.

Lexical simplification benchmarks for English, Portuguese, and Spanish.

Even in highly-developed countries, as many as 15–30% of the population can only understand texts written using a basic vocabulary. Their understanding of everyday texts is limited, which prevents them from taking an active role in society and making informed decisions regarding healthcare, legal representation, or democratic choice. Lexical simplification is a natural language processing task that aims to make text understandable to everyone by replacing complex vocabulary and expressions with simpler ones, while preserving the original meaning. It has attracted considerable attention in the last 20 years, and fully automatic lexical simplification systems have been proposed for various languages. The main obstacle for the progress of the field is the absence of high-quality datasets for building and evaluating lexical simplification systems. In this study, we present a new benchmark dataset for lexical simplification in English, Spanish, and (Brazilian) Portuguese, and provide details about data selection and annotation procedures, to enable compilation of comparable datasets in other languages and domains. As the first multilingual lexical simplification dataset, where instances in all three languages were selected and annotated using comparable procedures, this is the first dataset that offers a direct comparison of lexical simplification systems for three languages. To showcase the usability of the dataset, we adapt two state-of-the-art lexical simplification systems with differing architectures (neural vs. non-neural) to all three languages (English, Spanish, and Brazilian Portuguese) and evaluate their performances on our new dataset. For a fairer comparison, we use several evaluation measures which capture varied aspects of the systems' efficacy, and discuss their strengths and weaknesses. We find that a state-of-the-art neural lexical simplification system outperforms a state-of-the-art non-neural lexical simplification system in all three languages, according to all evaluation measures. More importantly, we find that the state-of-the-art neural lexical simplification systems perform significantly better for English than for Spanish and Portuguese, thus posing a question if such an architecture can be used for successful lexical simplification in other languages, especially the low-resourced ones.

Creating a list of word alignments from parallel Russian simplification data.

This work describes the development of a list of monolingual word alignments taken from parallel Russian simplification data. This word lists can be used in such lexical simplification tasks as rule-based simplification applications and lexically constrained decoding for neural machine translation models. Moreover, they constitute a valuable source of information for developing educational materials for teaching Russian as a second/foreign language. In this work, a word list was compiled automatically and post-edited by human experts. The resulting list contains 1409 word pairs in which each “complex” word has an equivalent “simpler” (shorter, more frequent, modern, international) synonym. We studied the contents of the word list by comparing the frequencies of the words in the pairs and their levels in the special CEFR-graded vocabulary lists for learners of Russian as a foreign language. The evaluation demonstrated that lexical simplification by means of single-word synonym replacement does not occur often in the adapted texts. The resulting list also illustrates the peculiarities of the lexical simplification task for L2 learners, such as the choice of a less frequent but international word.

Physicians’ use of plain language during discussions of prostate cancer clinical trials with patients

ObjectiveThis study described physicians’ use of plain language during patient-physician cancer clinical trial discussions. MethodsVideo-recorded clinical interactions and accompanying transcripts were taken from a larger study of communication and clinical trials (PACCT). Interactions (n = 25) were selected if they included invitations to participate in a clinical trial. We used descriptive, qualitative discourse analysis, a method that identifies language patterns at or above the sentence level. We first excerpted discussions of clinical trials, then identified instances of plain language within those discussions. Finally, we inductively coded those instances to describe physicians’ plain language practices. ResultsThe analysis identified four plain language practices. Lexical simplification replaced medical terminology with simpler words. Patient-centered definition named, categorized, and explained complex medical terminology. Metaphor explained medical terminology by comparing it with known concepts. Finally, experience-focused description replaced medical terminology with descriptions of patients’ potential physical experiences. ConclusionThese plain language practices hold promise as part of effective information exchange in discussions of cancer clinical trials. Testing is needed to identify patient preferences and the extent to which these practices address patient health literacy needs. Practice ImplicationsPending further testing, these plain language practices may be integrated into physician clinical trial and other communication training.

Artificial fine-tuning tasks for yes/no question answering

AbstractCurrent research in yes/no question answering (QA) focuses on transfer learning techniques and transformer-based models. Models trained on large corpora are fine-tuned on tasks similar to yes/no QA, and then the captured knowledge is transferred for solving the yes/no QA task. Most previous studies use existing similar tasks, such as natural language inference or extractive QA, for the fine-tuning step. This paper follows a different perspective, hypothesizing that an artificial yes/no task can transfer useful knowledge for improving the performance of yes/no QA. We introduce three such tasks for this purpose, by adapting three corresponding existing tasks: candidate answer validation, sentiment classification, and lexical simplification. Furthermore, we experimented with three different variations of the BERT model (BERT base, RoBERTa, and ALBERT). The results show that our hypothesis holds true for all artificial tasks, despite the small size of the corresponding datasets that are used for the fine-tuning process, the differences between these tasks, the decisions that we made to adapt the original ones, and the tasks’ simplicity. This gives an alternative perspective on how to deal with the yes/no QA problem, that is more creative, and at the same time more flexible, as it can exploit multiple other existing tasks and corresponding datasets to improve yes/no QA models.

Collection and evaluation of lexical complexity data for Russian language using crowdsourcing

Estimating word complexity with binary or continuous scores is a challenging task that has been studied for several domains and natural languages. Commonly this task is referred to as Complex Word Identification (CWI) or Lexical Complexity Prediction (LCP). Correct evaluation of word complexity can be an important step in many Lexical Simplification pipelines. Earlier works have usually presented methodologies of lexical complexity estimation with several restrictions: hand-crafted features correlated with word complexity, performed feature engineering to describe target words with features such as number of hypernyms, count of consonants, Named Entity tag, and evaluations with carefully selected target audiences. Modern works investigated the use of transforner-based models that afford extracting features from surrounding context as well. However, the majority of papers have been devoted to pipelines for the English language and few translated them to other languages such as German, French, and Spanish. In this paper we present a dataset of lexical complexity in context based on the Russian Synodal Bible collected using a crowdsourcing platform. We describe a methodology for collecting the data using a 5-point Likert scale for annotation, present descriptive statistics and compare results with analogous work for the English language. We evaluate a linear regression model as a baseline for predicting word complexity on handcrafted features, fastText and ELMo embeddings of target words. The result is a corpus consisting of 931 distinct words that used in 3,364 different contexts.

Predicting lexical complexity in English texts: the Complex 2.0 dataset

Identifying words which may cause difficulty for a reader is an essential step in most lexical text simplification systems prior to lexical substitution and can also be used for assessing the readability of a text. This task is commonly referred to as complex word identification (CWI) and is often modelled as a supervised classification problem. For training such systems, annotated datasets in which words and sometimes multi-word expressions are labelled regarding complexity are required. In this paper we analyze previous work carried out in this task and investigate the properties of CWI datasets for English. We develop a protocol for the annotation of lexical complexity and use this to annotate a new dataset, CompLex 2.0. We present experiments using both new and old datasets to investigate the nature of lexical complexity. We found that a Likert-scale annotation protocol provides an objective setting that is superior for identifying the complexity of words compared to a binary annotation protocol. We release a new dataset using our new protocol to promote the task of Lexical Complexity Prediction.

The Impact of Lexical and Syntactic Simplification of Materials on Listening Comprehension of Intermediate EFL Learners

The Impact of Lexical and Syntactic Simplification of Materials on Listening Comprehension of Intermediate EFL Learners

Pattern-Based Syntactic Simplification of Compound and Complex Sentences

With the advent of new technologies, simplifying the text automatically has been very popular and of high importance among natural language researchers during the last decade. The predominant research done in the area of Automatic Sentence Simplification(ASS) is inclined to either lexical or syntactical simplification of sentences. From the literature survey, it is observed that existing research in lexical simplification makes use of word substitution technique. This causes word sense ambiguity in cases where the word synonyms are not appropriate for a sentence in the given context. In contrast syntactical simplification though is accurate and applicable to Natural Language Processing (NLP) tasks requires tremendous efforts to construct rules for a given domain. The research proposes a framework called, Pattern-based Automatic Syntactic Simplification(PASS) which identifies sentences and applies rules based on grammatical patterns to simplify the sentences thereby making it more generic for NLP tasks. PASS is evaluated by human experts to rate the usefulness of the framework based on fluency, adequacy and simplicity of the sentences. Furthermore, the framework is automatically evaluated with the available online corpus using automatic metrics of SARI, BLEU, and FKGL. The proposed approach generates promising results in the field of ASS and could be used as a preliminary module for NLP tasks as well as other natural language-related applications like summarization, anaphora resolution, question-answering, and many more.

Converting the Words of God: An experimental evaluation of stylistic choices in the new Dutch Bible translation

In an experiment stylistic choices in the new Dutch Bible translation were evaluated. 185 participants evaluated 4 fragments which differed in type of variation (lexical or syntactic), source of the fragment (Bible - classical text) and the way the variations were applied (single - mixed). Also the religious¬ness of the evaluator was taken into account. Examined is whether lexical modernization and syntactic simplification are evaluated the same, whether single variations show a better insight into the evaluations and if the evaluation is influenced by the source of the text or the religiousness of the evaluator. The results showed that there is a difference in evaluation between syntactic and lexical variation, that single variation gives better insight into the evaluations and that source of the text and religiousness do not influence the evaluations. This study demonstrates that translators must involve readers at an early stage in their discussions on a required or appropriate register.