Search

Este artigo apresenta o método de construção duma ferramenta para a anotação de entidades geográficas mencionadas em textos medievais. A nova ferramenta foi desenvolvida a partir dos módulos de língua contemporânea do LinguaKit, pacote multilingue de ferramentas de PLN. Uma coleção de corpora anotados manualmente serviu de recurso para elaborar uma lista de topónimos medievais (gazetteers) e observar padrões para a melhora e implementação de novas regras de reconhecimento dos nomes de lugar. Depois da lista de entidades geográficas, os ativadores contextuais (triggers) foram o recurso determinante na melhora da abrangência. Para o produto final, fizeram-se também ajustes menores na procura de recolher os elementos mais comuns do léxico e os contextos gramaticais das entidades geográficas mencionadas. Ainda que muito trabalho fica por fazer na elaboração de listas para entidades não geográficas, na construção dum modelo de língua medieval e um lexicon específico, o novo módulo pode ser utilizado para anotar textos e mostra uma melhora significativa a respeito dos módulos previamente existentes This paper describes a method to build a tool aimed at recognizing geographical named entities in medieval texts. The new tool has been developed using the corresponding modules for contemporary languages contained in LinguaKit, a suite of NLP tools. A collection of manually annotated corpora served as a resource to build a gazetteer of medieval toponyms and find patterns to improve and implement new rules for the recognition of place names. In addition to the gazetteer, a list of triggers was the most determinant factor to improve recall. Final adjustments considered the most frequent terms of the lexicon and grammatical contexts for geographical named entities. In the process of building a model of medieval language and a specific lexicon, the available tool can already be used to annotate texts and shows a significant improvement when compared with previous modules. However, most work remains to be done in terms of adding specific gazetteers for entities other than geographical Este trabalho foi desenvolvido no marco da rede galega de investigacao TECANDALI, ED341DR2016/011, financiada pela Consellaría de Educación e Ordenación Universitaria da Xunta de Galicia, e do European Regional Development Fund (ERDF) SI

This study examines the sibilant fricatives produced by seventeen Galician and twenty-two Portuguese speakers. Galician and Portuguese are closely related languages that present important continuities, although it is in their phonological and phonetic systems that they diverge most obviously. By means of the analysis of spectra and spectral moments, chiefly the spectral mean, postalveolar sibilants and front (alveolar or alveolo-dental) sibilants are differentiated in both Galician and Portuguese. Much variation has been found in front sibilant realizations among speakers and even between different realizations by the same speaker. This variation is especially striking among male Galician speakers, where it was possible to distinguish three different articulations, identified here as an apico-aveolar [s̺], a lamino-alveolar [s], and a lamino-dental [s̪] sibilant. The research results point to a loss of phonetic diversity on the Portuguese side of the political border, while in Galicia it is better preserved, although it is losing ground This paper was written within the project “Cambio lingüístico no galego actual”, funded by the Ministry of Economy and Competitivity of the Spanish Government (FFI2012-33845), and within the Tecnoloxías e Análise dos Datos Lingüísticos network, funded by the Xunta de Galicia (R2014/007)

Comunicación presentada no Congreso Internacional EUROPHRAS 2019. Modelos productivos en Fraseología (Santiago de Compostela, 24-25 de xaneiro de 2019).

Presentación na Universidade de Antuerpe, 18 de maio de 2018 Xunta de Galicia, Unión Europea (Fondo Social Europeo)

Comunicación presentada no XII Congreso da Asociación Internacional de Estudos Galegos. Novos horizontes para unha Galicia global: redes, territorio e memoria. Madrid, 10-15 de setembro 2018 Os atlas lingüísticos constitúen unha excelente fonte para a realización de estudos que permitan coñecer a variación lingüística nos distintos niveis de análise, unha vez que rexistran nas súas cartas as variantes recorrentes en determinada área asociadas a unha mesma variable. Para o territorio galego contamos cos datos do Atlas Lingüístico de la Península Ibérica (ALPI) (1934) e do Atlas Lingüístico Galego (ALGa) (1974). Estes dous atlas ofrecen numerosa información organizada onomasioloxicamente, entre a que distan corenta anos, o que posibilita a realización de estudos sobre variación e cambio lingüístico. O noso obxectivo é analizar as denominacións existentes para nocello ‘parte saínte dos ósos que forma a articulación da perna co pé’ no territorio galego a partir dos datos do ALPI e do ALGa, que complementamos con datos extraídos doutros atlas rexionais, así como de corpus lexicográficos. A análise das distintas variantes realizarase tendo en conta, por un lado, a distribución xeográfica en ambos os atlas, o que nos permitirá observar a existencia de procesos de cambio lingüístico e, polo outro, as motivacións semánticas, principalmente metafóricas, que dan lugar á variedade de formas.

Dentro del reconocimiento automático del habla, los modelos de lenguaje estadísticos basados en la probabilidad de secuencia de palabras (n-gramas) suponen uno de los dos pilares sobre los que se basa su correcto funcionamiento. En este trabajo se expone el impacto que tienen sobre las prestaciones de reconocimiento a medida que estos modelos se mejoran con más texto de mejor calidad, cuando estos se ajustan a la aplicación final del sistema, y por lo tanto, cuando se reducen el número de palabras fuera de vocabulario (Out Of Vocabulary - OOV). El reconocedor con los distintos modelos de lenguaje ha sido aplicado sobre cortes de audio correspondientes a tres marcos experimentales: oralidad formal, habla en noticiarios, y TED talks en gallego. Los resultados obtenidos muestran claramente una mejora sobre los marcos experimentales propuestos. Within the automatic speech recognition, statistical language models based on the probability of word sequences (n-grams) represent one of the two pillars on which its correct functioning is based. In this paper, the impact they have on the recognition result is exposed as these models are improved with more text of better quality, when these are adjusted to the final application of the system, and therefore, when the number out of vocabulary (OOV) words is reduced. The recognizer with the different language models has been applied to audio cuts corresponding to three experimental frames: formal orality, talk on newscasts, and TED talks in Galician. The results obtained clearly show an improvement over the experimental frameworks proposed. El trabajo realizado está enmarcado en el proyecto del Plan Nacional TraceThem TEC2015-65345-P y en la red gallega TecAnDaLi ED431D 2016/011 financiada por la Xunta de Galicia. Asimismo se beneficia de las ayudas de la Xunta de Galicia de Grupos de Referencia Competitiva GRC2014/024 y Agrupación Estratéxica Consolidada de Galicia acreditación 2016-2019 y a la Unión Europa a través de los fondos FEDER.

Conferencia pronunciada no Romaniches Seminar - Universität Zürich (Zürich, 20-21 de abril de 2018).

Póster presentado no XIII Congreso Internacional de Lingüística Xeral (CILX 2018), celebrado en Vigo do 13 ao 15 de xuño de 2018.

Comunicación presentada no I Encontro Internacional de Língua Portuguesa e Relações Lusófonas (LUSOCONF2018). Bragança, 19-20 de outubro de 2018.

Ministerio de Economía y Competitividad del Gobierno de España

The Corpus Oral Informatizado da Lingua Galega (CORILGA) project aims at building a corpus of oral language for Galician, primarily designed to study the linguistic variation and change. This project is currently under development and it is periodically enriched with new contributions. The long-term goal is that all the speech recordings will be enriched with phonetic, syllabic, morphosyntactic, lexical and sentence ELAN-complaint annotations. A way to speed up the process of annotation is to use automatic speech-recognition-based tools tailored to the application. Therefore, CORILGA repository has been enhanced with an automatic alignment tool, available to the administrator of the repository, that aligns speech with an orthographic transcription. In the event that no transcription, or just a partial one, were available, a speech recognizer for Galician is used to generate word and phonetic segmentations. These recognized outputs may contain errors that will have to be manually corrected by the administrator. For assisting this task, the tool also provides an ELAN tier with the confidence measure of each recognized word. In this paper, after the description of the main facts of the CORILGA corpus, the speech alignment and recognition tools are described. Both have been developed using the Kaldi toolkit

This study describes the introduction of verbal idioms in the Galician language version (Galnet) of the semantic network WordNet; a network that does not traditionally include many phraseological units. To enhance Galnet, a list of 803 Galician verbal idioms was developed to then review each of them individually and assess whether they could be introduced in an existing WordNet synset (a group of synonyms expressing the same concept) or not. Of those 803 idioms, 490 (61%) could be included in this network. Besides, Galnet was enlarged with 750 extra verbal idioms, most of them synonyms or variants of the former. In this study, we present the working methodology for the experiment and an analysis of the results, to help understand the most important problems found when trying to introduce idioms in Galnet. We also discuss the reasons preventing the inclusion of some expressions, and the criteria used to introduce the idioms that finally made it into the network This research has been supported by the Xunta de Galicia and the European Union (under grant GRC2013/40) and by the Ministry of Economy and Competitiveness of the Spanish Government (Project TUNER: Automatic domain adaptation for semantic processing, TIN2015-65308-C5-1-R) SI

Geoprosodic data are useful for studying language change and developing hypotheses about the diachrony of intonation. In the case of Galician, it is particularly interesting to study varieties of Galician and Portuguese which share a common origin but are separated by a long-standing political border. Work to date has concluded that some of these intonation patterns present a prosodic continuum, but has also identified a large part of the Galician linguistic area where a widespread pattern is found that is unrelated to Portuguese. An approach to the study of dialectology and linguistic change will be proposed which supplements traditional geoprosodic studies with sociolinguistic concepts such as contact between languages and language varieties. This article will address questions concerning interaction between geoprosodic variation and contact among languages and language varieties for the purpose of detecting ongoing prosodic change and describing prosodic convergence processes that affect coexisting language varieties in Galicia., Los datos geoprosódicos son muy valiosos para el estudio del cambio lingüístico y para la elaboración de hipótesis sobre la diacronía de la entonación. En el caso del gallego, es especialmente interesante el estudio de las variedades gallegas y portuguesas, de origen común pero separadas secularmente por una frontera política. Los trabajos realizados hasta el momento concluyen que existe una continuidad prosódica de algunos de estos patrones de entonación, si bien también detectan amplias áreas en el territorio lingüístico gallego que presentan un patrón muy extendido geográficamente que no se puede relacionar con el portugués. Nuestra propuesta, complementaria de los estudios geoprosódicos tradicionales, pretende abordar la cuestión de cómo interaccionan la variación geoprosódica y el contacto de lenguas y de variedades con el fin de detectar cambios prosódicos en marcha y formalizar los procesos de convergencia prosódica que afectan a las variedades lingüísticas que conviven en Galicia.

The essential objective of dialectology, and especially geolinguistics, is the study of spatial linguistic variation, special relevance being given to the presentation of results through cartographic representations. The methods of geolinguistic data analysis focused for a long period on the description and evaluation of phenomena conventionally regarded as particularly relevant, either in isolated or in group cases, and the latter only when the discovery of the existence of limits between varieties was straightforward. The application of quantitative methods to geolinguistic studies began at the start of the 1970s and experienced a spectacular impulse over the last two decades. The development of new computer-based tools and the recognition that dialectal variation could not be reduced to simple characterizations encouraged researchers to employ new research methods, with which it was possible to understand in greater detail the patterns which function in geolinguistic variation. This present study seeks to provide an overview of the principal analytical techniques and representation of geolinguistic information developed during in recent years. Moreover, it intends to demonstrate, on the basis of some examples of application, the interest of these new methodologies and their usefulness for a better understanding of the spatial variety of languages. A dialectoloxía, e de xeito especial a xeolingüística, ten como obxectivo esencial o estudo da variación lingüística espacial, dándolle unha especial relevancia á presentación de resultados por medio de representacións cartográficas. Os métodos de análise de datos xeolingüísticos centráronse durante moito tempo na descrición e análise de fenómenos considerados por convención especialmente relevantes, ben illadamente ou ben de xeito agrupado, e isto último só cando resultaba doado descubrir a existencia de límites entre variedades. A aplicación de métodos cuantitativos ó estudo xeolingüístico iniciouse a comezos dos anos setenta e coñeceu un espectacular impulso nas dúas últimas décadas. O desenvolvemento de novas ferramentas informáticas e o recoñecemento de que a variación dialectal non podía ser reducida a caracterizacións simples animou os investigadores á utilización de novos métodos de investigación, cos que foi posible coñecer máis en profundidade os padróns que funcionan na variación xeolingüística. Esta contribución pretende facer un repaso das principais técnicas de análise e representación de información xeolingüística desenvolvidas nos últimos anos. Ademais, búscase demostrar, a partir dalgúns exemplos de aplicación, o interese destas novas metodoloxías e a súa utilidade para un mellor coñecemento da variación espacial das linguas. SI

Comunicación presentada no IVème Congrès International de Dialectologie et Sociolinguistique. Variations, phraséologie et ressorces (IV CIDS), Université Paris Sorbonne (París, 7-9 de setembro de 2016)

This paper introduces a tool that performs segmentation and labelling of sound chains in phono units, syllables and/or words departing from a sound signal and its corresponding orthographic transcription. In addition, it also integrates acoustic analysis scripts applied to the Praat programme with the aim of reducing the time spent on tasks related to analysis, correction, smoothing and generation of graphics of the melodic curve. The tool is implemented for Galician, Spanish and Brazilian Portuguese. Our goal is to contribute, by means of this application, to automatize some of the tasks of segmentation, labelling and prosodic analysis, since these tasks require a large investment of time and human resources., En este artículo se presenta una herramienta que realiza la segmentación y el etiquetado de cadenas sonoras en unidades de fono, sílaba y/o palabra partiendo de una señal sonora y de su correspondiente transcripción ortográfica. Además, integra scripts de análisis acústico que se ejecutan sobre el programa Praat con el fin de reducir el tiempo invertido en las tareas de análisis, corrección, suavizado y generación de gráficos de la curva melódica. La herramienta está implementada para gallego, español y portugués de Brasil. Nuestro objetivo es contribuir con esta aplicación a automatizar algunas de las labores de segmentación, etiquetado y análisis prosódico, pues constituyen tareas que requieren una gran inversión de tiempo y de recursos humanos.

Comunicación presentada no 40º Aniversário do Centro de Linguística da Universidade do Porto. Colóquio Comemorativo. Faculdade de Letras. Porto, 24-25 de novembro de 2016

Neste traballo preséntase unha ferramenta informática, desenvolvida co software estatístico R, que lle permite ao usuario realizar unha análise dialectométrica dos datos prosódicos do corpus fixo recollido no proxecto AMPER. Non se precisa coñecementos previos no uso do software R, polo que a ferramenta pode ser empregada por calquera usuario interesado neste tipo estudos. A análise dialectométrica realizada pola ferramenta inclúe o cálculo de correlacións entre curvas de F0 e a obtención das distancias prosódicas entre as distintas localizacións (falantes ou outra variable de interese) existentes no corpus. Unha vez construída a táboa de distancias prosódicas, o usuario pode aplicar mediante a ferramenta técnicas da estatística multivariante, como o escalado multidimensional (MDS) e a análise de conglomerados. Con esta última metodoloxía, o usuario pode detectar agrupamentos nas localizacións (falantes ou outra variable de interese) segundo a súa proximidade en termos de distancia prosódica., This paper presents a computing tool, developed with the statistical software R,which allows a user to perform a dialectometric analysis of the prosodic data in thefixed corpus collected in the AMPER project. Since no previous knowledge aboutusing R is assumed, this tool can be used by any user interested in such studies.The dialectometric analysis that is performed includes calculation of correlationsbetween F0 curves and of prosodic distances between different locations coveredby the corpus (for speakers or any other variable). Once the table of prosodicdistances has been generated, the user can then apply multivariate statisticaltechniques to it, such as multidimensional scaling (MDS) and cluster analysis. Thelatter allows the user to detect the existence of clusters of locations (by speaker or another variables) according to their closeness in terms of prosodic distance.

Baseándonos sobre todo en argumentos de índole lingüística, mais tamén en indicios biográficos e ideolóxicos, atribuímos esta “Carta de Galicia” a Manuel Freire Castrillón. Argumentos do mesmo tipo móvennos a postular que Freire foi tamén o autor doutros textos en castrapo e en galego, entre eles Un labrador que foy sarxento á os soldados do nobo alistamento (ca. 1808). Finalmente, ofrecemos unha transcrición anotada e minimamente intervida da carta. Basándonos sobre todo en argumentos de índole lingüística, pero también en indicios biográficos e ideológicos, atribuimos esta “Carta de Galicia” a Manuel Freire Castrillón. Argumentos del mismo tipo nos mueven a postular que Freire fue también el autor de otros textos en gallego y en un interlecto castellano-gallego (castrapo); entre estos textos se encontraría el romance Un labrador que foy sarxento á os soldados do nobo alistamento (ca. 1808). Finalmente, ofrecemos una transcripción anotada y mínimamente intervenida de la carta. In this paper we defend that Manuel Freire Castrillón was the author of this “Carta de Galicia” and of some other texts that were written in Galician or in a Spanish-Galician interlect (castrapo); we think that one of them could be the poem Un labrador que foy sarxento á os soldados do nobo alistamento (ca. 1808). The arguments we adduce to defend these hypotheses are of biographical, ideological and, most of all, linguistic nature. Finally, we offer a transcription of this letter and some notes of historical and linguistic nature. SI