Search

Highlights • A Holographic mobile-based application is used to practice the pronunciation of basic English vocabulary words. • A funny 3D holographic robot interacts as virtual teacher in charge of practicing. • An experiment with 70 Spanish pre-school children has been carried out. • Results shown a significant impact on the children's motivation and performance. Abstract This paper describes a holographic mobile-based application designed to help Spanish-speaking children to practice the pronunciation of basic English vocabulary words. The mastery of vocabulary is a fundamental step when learning a language but is often perceived as boring. Producing the correct pronunciation is frequently regarded as the most difficult and complex skill for new learners of English. In order to address these problems this research takes advantage of the power of multi-channel stimuli (sound, image and interaction) in a mobile-based hologram application in order to motivate students and improve their experience of practicing. We adapted the prize-winning HolograFX game and developed a new mobile application to help practice English pronunciation. A 3D holographic robot that acts as a virtual teacher interacts via voice with the children. To test the tool we carried out an experiment with 70 Spanish pre-school children divided into three classes, the control group using traditional methods such as images in books and on the blackboard, and two experimental groups using our drills and practice software. One experimental group used the mobile application without the holographic game and the other experimental group used the application with the holographic game. We performed pre-test and post-test performance assessments, a satisfaction survey and emotion analysis. The results are very promising. They show that the use of the holographic mobile-based application had a significant impact on the children's motivation. It also improved their performance compared to traditional methods used in the classroom. [ABSTRACT FROM AUTHOR]

In this paper we study the effect of different lexical resources for selecting synonyms and strategies for word sense disambiguation in a lexical simplification system for the Spanish language. The resources used for the experiments are the Spanish EuroWordNet, the Spanish Open Thesaurus and a combination of both. As for the synonym selection strategies, we have used both local and global contexts for word sense disambiguation. We present a novel evaluation framework in lexical simplification that takes into account the level of ambiguity of the word to be simplified. The evaluation compares various instances of the lexical simplification system, a gold standard, and a baseline. The paper presents an in-depth qualitative error analysis of the results. [ABSTRACT FROM AUTHOR]

This paper explores the use of linguistic information for the selection of data to train language models. We depart from the state-of-the-art method in perplexity-based data selection and extend it in order to use word-level linguistic units (i.e. lemmas, named entity categories and part-of-speech tags) instead of surface forms. We then present two methods that combine the different types of linguistic knowledge as well as the surface forms (1, naïve selection of the top ranked sentences selected by each method; 2, linear interpolation of the datasets selected by the different methods). The paper presents detailed results and analysis for four languages with different levels of morphologic complexity (English, Spanish, Czech and Chinese). The interpolation-based combination outperforms the purely statistical baseline in all the scenarios, resulting in language models with lower perplexity. In relative terms the improvements are similar regardless of the language, with perplexity reductions achieved in the range 7.72–13.02%. In absolute terms the reduction is higher for languages with high type-token ratio (Chinese, 202.16) or rich morphology (Czech, 81.53) and lower for the remaining languages, Spanish (55.2) and English (34.43 on the English side of the same parallel dataset as for Czech and 61.90 on the same parallel dataset as for Spanish). [ABSTRACT FROM AUTHOR]

It is estimated that between 50% and 75% of all cases of dementia are due to Alzheimer's disease (AD), the most common neurodegenerative disease among World population. However, a long preclinical period of AD makes it difficult to differentiate between people with Mild Cognitive Impairment (MCI) that would progress to dementia from people with MCI that would not. One of the most promising solutions to detect MCI which will evolve to dementia (preAD) comes from the field of automatic speech analysis. Speech is a complex physiological and neurocognitive language-mediated process, which can be significantly altered in pathological aging and exhibit high levels of sensitivity for the diagnosis of neurological diseases. The purpose of this research is to offer a detailed perspective on the speech changes in MCI and mild AD when compared to healthy aging (HA), that would allow to detect pathological processes prior to the clinical expression of AD. Based on our previous research record on speech in HA, MCI and AD, we provide a global review of dementia-related speech traits and propose a reading-based protocol for assessing ongoing neurodegenerative processes in the elderly. We report the results of speech analysis in elderly people with different cognitive profiles, who performed a standardized reading task and were further analyzed for correlations between neurocognitive assessment indicative of cognitive impairment stage (HA, MCI or AD) and acoustic, temporal and prosodic traits in speech. We show that evolution from HA to AD exhibits a steady pattern of speech changes in parallel to the cognitive decline, which consists in significant increase in duration and phonation time, extension of pauses and voice breaks, intensification of variation in syllabic production, and decrease in speech energy and intensity leading to dysphony. In doing so, we prove that a standardized reading task is a very useful type of stimuli for detecting dementia-related speech traits and, in view of this, we discuss the relevance of reading for preclinical automated diagnosis of AD. The main contribution of this paper is a corpus of recordings of the standardized reading task performed by healthy elderly people and people with MCI and AD in Spanish language, and which can be used for further research purposes. In this respect, our work fills an important gap existing in corpora-based studies of speech and language impairments related to progression to dementia. [ABSTRACT FROM AUTHOR]

Abstract: This paper describes a preprocessing module for improving the performance of a Spanish into Spanish Sign Language (Lengua de Signos Española: LSE) translation system when dealing with sparse training data. This preprocessing module replaces Spanish words with associated tags. The list with Spanish words (vocabulary) and associated tags used by this module is computed automatically considering those signs that show the highest probability of being the translation of every Spanish word. This automatic tag extraction has been compared to a manual strategy achieving almost the same improvement. In this analysis, several alternatives for dealing with non-relevant words have been studied. Non-relevant words are Spanish words not assigned to any sign. The preprocessing module has been incorporated into two well-known statistical translation architectures: a phrase-based system and a Statistical Finite State Transducer (SFST). This system has been developed for a specific application domain: the renewal of Identity Documents and Driver''s License. In order to evaluate the system a parallel corpus made up of 4080 Spanish sentences and their LSE translation has been used. The evaluation results revealed a significant performance improvement when including this preprocessing module. In the phrase-based system, the proposed module has given rise to an increase in BLEU (Bilingual Evaluation Understudy) from 73.8% to 81.0% and an increase in the human evaluation score from 0.64 to 0.83. In the case of SFST, BLEU increased from 70.6% to 78.4% and the human evaluation score from 0.65 to 0.82. [Copyright &y& Elsevier]

Abstract: Web directories organize voluminous information into hierarchical structures, helping users to quickly locate relevant information and to support decision-making. The development of existing ontologies and Web directories either relies on expert participation that may not be available or uses automatic approaches that lack precision. As more users access the Web in their native languages, better approaches to organizing and developing non-English Web directories are needed. In this paper, we have proposed a semi-automatic framework, which consists of anchor directory boosting, meta-searching, and heuristic filtering, to construct domain-specific Web directories. Using the framework, we have built a Web directory in the Spanish business (SBiz) domain. Experimental results show that the SBiz Web directory achieved significantly better recall, F-value, efficiency, and satisfaction rating than the benchmark directory. Subjects provided favorable comments on the SBiz Web directory. This research thus contributes to developing a useful framework for organizing domain-specific information on the Web and to providing empirical findings and useful insights for end-users, system developers, and researchers of Web information seeking and knowledge management. [Copyright &y& Elsevier]

• A novel database of Spanish esophageal voices designed for developing new software for speakers with this pathology. • Includes recordings of 100 sentences from 30 esophageal speakers. The size and variety of the database makes it unique. • Includes the phonetic labels of the recordings. Automatic phonetic labelling procedure is described. • The main acoustic characteristics of the voices are described and compared with a reduced set of healthy voices. • The database has successfully been used for reducing Automatic Speech Recognition errors for esophageal speakers. A laryngectomee is a person whose larynx has been removed by surgery, usually due to laryngeal cancer. After surgery, most laryngectomees are able to speak again, using techniques that are learned with the help of a speech therapist. This is termed as alaryngeal speech, and esophageal speech (ES) is one of the several alaryngeal speech production modes. A considerable amount of research has been dedicated to the study of alaryngeal speech, with a wide range of aims such as helping speech therapists with evaluation and diagnosis, and improving its quality and intelligibility using digital signal processing techniques. We present to you a database of Spanish ES voices, named AhoSLABI, which is designed to allow the development of new support technologies for this speech impairment. The database primarily consists of recordings of 31 laryngectomees (27 males and 4 females) pronouncing phonetically balanced sentences. Additionally, it includes parallel recordings of the sentences by 9 healthy speakers (6 males and 3 females) to facilitate speech processing tasks that require small parallel corpora, such as voice conversion or synthetic speech adaptation. Apart from the sentences, the database includes sustained vowels and a small set of isolated words, which can be valuable for research on ES analysis, diagnosis and evaluation. The paper describes the main contents of the database, the recording protocols and procedure, as well as the labeling process. The main acoustic characteristics of the voices, such as speaking rate, durations of the recordings, phones and silences, and other such characteristics are compared with those of a reduced set of healthy voices. In addition, we describe an experiment using the database to improve the performance of an ASR system for ES speakers. This new resource will be made available to the scientific community with the hope that it will be used to improve the quality of life of the laryngectomees. [ABSTRACT FROM AUTHOR]