Spanish Corpus Research Articles

Explainable ICD multi-label classification of EHRs in Spanish with convolutional attention

BackgroundThis work deals with Natural Language Processing applied to Electronic Health Records (EHRs). EHRs are coded following the International Classification of Diseases (ICD) leading to a multi-label classification problem. Previously proposed approaches act as black-boxes without giving further insights. Explainable Artificial Intelligence (XAI) helps to clarify what brought the model to make the predictions. GoalThis work aims to obtain explainable predictions of the diseases and procedures contained in EHRs. As an application, we show visualizations of the attention stored and propose a prototype of a Decision Support System (DSS) that highlights the text that motivated the choice of each of the proposed ICD codes. MethodsConvolutional Neural Networks (CNNs) with attention mechanisms were used. Attention mechanisms allow to detect which part of the input (EHRs) motivate the output (medical codes), producing explainable predictions. ResultsWe successfully applied methods in a Spanish corpus getting challenging results. Finally, we presented the idea of extracting the chronological order of the ICDs in a given EHR by anchoring the codes to different stages of the clinical admission. ConclusionsWe found that explainable deep learning models applied to predict medical codes store helpful information that could be used to assist medical experts while reaching a solid performance. In particular, we show that the information stored in the attention mechanisms enables DSS and a shallow chronology of diagnoses.

Über-Reden durch An-Reden: Die phatische Funktion als Mittel zum Appell in englischen, spanischen und deutschen Werbetexten

Advertisements are intended to persuade the addressees to buy a particular product or service. In order to achieve this communicative purpose, text-producers make use of various strategies. Apart from directly appellative speech acts like recommendations, requests or invitations, they use referential, expressive, and phatic utterances as an indirect means of persuasion. The phatic function, which, according to Jakobson (1960), provides the appro-priate channel for communication, is of particular importance in this respect. If the channel does not work properly, the persuasive function will never reach its aim. In the following paper, a small corpus of English, Spanish and German advertising texts is analyzed in order to find the indicators of the phatic function which are instrumental in making persuasion work.

Herencia medieval y neologismo en el l\xe9xico crom\xe1tico de textos historiogr\xe1ficos en el siglo de Oro seg\xfan el Corpus diacr\xf3nico del espa\xf1ol

AbstractThe main objective of this investigation is to examine with textual support the use of the vocabulary of color during the Golden Period through the investigation that it is a medieval heritage or a neologism. This chromatic terminology was little used during the thirteenth century and in the following centuries its use is multiplied by the communication needs of a society that demands descriptive information on many social, cultural and economic issues, as happens with historiographic documents. To this end, a quantitative investigation has been designed in three phases: (a) search for chromatic voices that designate six selected colors; (b) determination of the first documentation of this group of voices; and (c) quantitative analysis to find a trend regarding the silver duality. For the search of these empirical references, the Diachronic Corpus of Spanish has been used as a data bank to collect textual examples on the chromatic vocabulary extracted from different historiographic sources. The result is the presentation of 100 terms with their respective empirical testimonies that reflect the continuity of the medieval heritage in the colors white, black and red, while formal neologisms are very frequent for the rest.

Corpora and L2 acquisition: the L1 Portuguese – L2 Spanish subcorpus of CEDEL2

This paper presents the L1 Portuguese – L2 Spanish subcorpus of Corpus Escrito del Español L2 (CEDEL2), a new methodological resource for second language acquisition (SLA) research, which is freely searchable and downloadable (http://cedel2.learnercorpora.com). CEDEL2 is a large-scale, multi-L1 learner corpus of L2 Spanish which contains written productions from learners at all proficiency levels as well as 6 native control subcorpora (total size: over 1,100,000 words from over 4,000 participants). CEDEL2 follows strict corpus design criteria (Sinclair, 2005) and learner corpus design recommendations (Tracy-Ventura & Paquot, 2021a). In its current version (CEDEL2 v. 2), its Portuguese component includes an L1 Portuguese – L2 Spanish subcorpus, with 21,662 words written by 164 participants, and an L1 Portuguese native subcorpus, with 3,500 words from 16 L1 speakers of European Portuguese. Thanks to their design features (e.g., same design across subcorpora, inclusion of metadata about SLA-relevant variables, dual native control subcorpora) and freely available web interface, CEDEL2 and its Portuguese subcorpora allow researchers to investigate a wide range of topics in SLA.

Perspectival factors and pro-drop: A corpus study of speaker/addressee pronouns with creer ‘think/ believe’ and saber ‘know’ in spoken Spanish

This paper examines overt and covert speaker/addressee pronouns with the cognitive verbs creer ‘think/believe’ and saber ‘know’ in a corpus of spoken peninsular Spanish – the Madrid and Alcalá samples of PRESEEA (2014– ) – with a focus on 1st person singular (yo) creo que ‘(I) think that’. Departing from the observation made in the literature that overt pronouns are highly frequent with creer and that topic shift cannot account for all of them, it will be argued that perspectival factors related to evidentiality/epistemicity and subjectivity influence overt pronoun realization. A corpus study was conducted to investigate whether (i) [person] and [polarity] and (ii) the type of complement affect overt pronoun realization with the cognitive verbs creer and saber. The results indicate that the type of belief expressed in the embedded clause should be taken into account, as well as person and polarity. The ultimate trigger for phonetic realization of speaker/addressee pronouns will be argued to be the notion of contrast: cognitive verbs whose embedded complement encodes evaluations and non-visual, abstract information have high frequencies of overt pronoun realization because these contexts favor the evoking of alternative perspective holders. Overt pronouns will be analyzed as the result of a [+contrast] feature which is assigned to the specifier of a functional category encoding perspective in the split IP.

Constructing Corpus and Word Embedding for Spanish Covid-19 Data

Severe acute respiratory syndrome coronavirus 2 (COVID 19), colloquially referred to as coronavirus, escalated into a global pandemic with severe transmission and mortality rates in 2019. Despite the escalation of the virus’ worldwide impact in 2020, numerous studies on Natural Language Processing in Spanish have neglected corpus construction or word embedding, especially conspicuous in its absence being the corpora involving coronavirus or infectious diseases. Additionally, corpus construction or word embedding conducted in the medical field do not display efficacy in production pertaining to coronavirus or infectious diseases. To supplement this potentially detrimental insufficiency, this study collects Spanish Language data to build a relevant coronavirus corpus through appropriate preprocessing and then obtains a word embedding. Performance of the corpus and word embedding are then tested through word similarity evaluations, a cosine similarity evaluation, and a visualization evaluation with the existing Spanish corpus. After comparison, corpus and word embedding suitable for coronavirus will be suggested.

Word embeddings and semantic shifts in historical Spanish: Methodological considerations

Abstract Word embeddings have recently been applied to detect and explore changes in word meaning on large historical corpora. While word embeddings are useful in many Natural Language Processing tasks, there are a number of questions that need to be addressed concerning accuracy and applicability of these methods for historical data. There is a scarce literature on the stability and replicability of these embeddings, especially on small corpora, which are common in historical work. It also remains unclear whether methods used to evaluate embeddings in contemporary data can also be used for historical data sets. Our overarching goal is to use word embeddings for investigating semantic shifts in the history of Spanish. In the work presented here, we focus on methodological questions that arise. We first examine the stability and applicability of three commonly used word embeddings models on a small corpus of medieval and classical Spanish. Comparing our results with a study on the word algo as a test case, we show that a rank-averaging method can produce more stable results from the embeddings. We corroborate previous theoretical work while demonstrating the applicability of our method when training word embeddings on small corpora for the analysis of semantic change. Second, we investigate how best to evaluate different embeddings models. We show that an existing analogy test cannot be used without modification. Our new analogy test, consisting of roughly 10,000 questions for medieval and classical Spanish, will be released with the article.

La traducción de la COVID-19 a la lengua de signos española

The year 2020 will be remembered for being the year when a new coronavirus made the world wobble. Health systems in all countries are faced with COVID-19, a disease for which we do not have sufficient resources. The successful implementation of preventive measures by society is essential to prevent its spread. However, the information provided by governments is not accessible to all citizens and deaf people find it difficult not to be in sign language, which has serious consequences for their health and that of everyone else. The present paper analyses the translation techniques applied to translate terms about COVID-19 into Spanish sign language (LSE). Through the compilation of a parallel video graphic corpus in LSE and Spanish with subtitles, and taking as a reference the proposal of translation techniques by Hurtado (2001), we have identified the procedures used by a deaf translator to achieve these equivalences. The results reveal that borrowing, either from oral language (through fingerspelling) or from other sign languages (with the widespread sign for coronavirus), is the main resource to cope with the neologisms that have arisen. Furthermore, the use of different techniques for the same word has been found, even when there is an equivalent in LSE dictionaries. Among all of them, generalization stands out as a procedure to make texts more understandable in the target culture, which indicates the importance of cultural and social knowledge in the translator. All of this demonstrates the functionality of translation techniques and the adaptation that is made of them taking into account the (con)text, as well as the dynamic nature of the equivalents. This exploratory research allows us to go deeper into these translation procedures in order to make this information more accessible to the deaf community and to fight together against COVID-19.

Localización y contabilización de sufijos nominales en corpus de aprendientes de español como segunda lengua

This study aims to analyze and to record the number of words with noun suffixes in a corpus produced by learners of Spanish as second language. Obtaining the suffix index is of utmost importance to draw conclusions about both, the number of nominalizations and the percentage of occurrences coincident with Spanish suffixes as well as idiosyncratic forms. The study is based on the theoretical framework of the interlanguage hypothesis within the theories of second language acquisition and the latest research on automatic handling of errors in Spanish corpus, particularly lexicons. The text corpus is divided in groups according to the proficiency level of the students, who have different native languages. Regarding the methodology, the Nooj system will be used so as to identify the terms that have noun suffixes, and then two different statistical techniques will be implemented to measure the data: the non-parametric test of Wicoxon for independent samples and the non-parametric test of Kruskal-Wallis. The obtained results will show substantial differences in the use and distribution of Spanish noun suffixes among the corpus that do not share the same level of proficiency. Indeed, a significant increase in matching nominal suffixes will be found in the intermediate level sample as compared to the initial level. However, the examined corpus will not show considerable differences in two other cases: existing suffixes in Spanish not combinable with bases and non-existent suffixes.

Gender and race in Langston Hughes’ poetry of the Spanish Civil War

ABSTRACT African American poet Langston Hughes worked as a press correspondent during the Spanish civil war. This experience left an imprint in his production and particularly in his poetry, giving light to significant advances in his entwinement of race, gender, and identity. The acclaimed diversity of Hughes’ feminine models, and his inclusive stance find difficult accommodation in the poetic corpus about Spain. Using a critical appraisal of race, class and gender divisions, the poetic representation of female characters is discussed attending to their different forms of inclusion and exclusion. Their relationship with later developments in Hughes’ poetic construction of African American female agency is assessed, and the singularity of certain characters in the Spanish corpus is explored. Finally, conclusions are drawn to demonstrate the relevant and understudied contributions of this corpus for the better understanding of Hughes’ feminine universe as a whole.

Morphological complexity and rated writing proficiency

Abstract Linguistic complexity measures are used to describe second language (L2) performance and assess levels of proficiency and development. Although morphology is considered crucial in L2 acquisition, morphological complexity has been relatively neglected, hindering comprehensive views of grammatical complexity in L2. This article presents an application of a recently proposed metric of morphological diversity, the Morphological Complexity Index (MCI), in an L2 Spanish corpus of 113 essays classified into four proficiency levels by expert evaluators. The aim of the study is to investigate the relationships of MCI with subjectively rated proficiency and with other four quantitative measures of L2 complexity. Results indicate that morphological complexity, as measured by MCI, does not vary significantly across proficiency levels in this corpus. The MCI shows significant correlations with lexical but not with syntactic complexity measures. Findings are interpreted in the light of the characteristics of the corpus and the acquisition of the Spanish verbal system.

On nonce echo constructions expressing disapproval and annoyance

Abstract The aim of this paper is to explore and contrast the morphosyntactic and semantic features of two equivalent nonce echo constructions (NECs) expressing disapproval and annoyance (i.e. don’t (you) X me and ni X ni Y) in colloquial English and Spanish. A NEC is defined as a reactive clause containing duplicated words of the initiative, showing the property of nonceness, and having a communicative goal. Two types of NECs are found in both languages: an attitudinal echo construction and a referential one, the former being more idiomatic than the latter. Based on the premise that texts are necessarily dialogic, two sets of examples are obtained from English corpora (The Movie Corpus, The TV Corpus, and the Corpus of American Soap Operas); and from two Spanish corpora and a dataset (CREA, CORPES XXI, and the Dataset of Spanish Dialogic Texts Online). Findings suggest that both languages show negative and restricted syntactic templates. Variables are coinages that originate from the processes of conversion in English (e.g. don’t you John me) and gender polarity in Spanish (e.g. ni luna ni *luno). Hence, although both types of variables are morphologically novel and contextually meaningful, Spanish variables are generally ungrammatical and unlikely to exist outside the discursive frame under study.

Lexically specific accumulation in memory of word and segment speech rates

AbstractVariability abounds in speech. According to usage-based accounts, lexical representations reflect phonetic variants of words resulting from contextual conditioning. Because faster speech contexts promote durational shortening of words and segments, words that occur more often in fast speech may be more reduced than words commonly used in slow speech, independent of the target’s contextual speech rate. To test this, linear mixed-effects models including a word form’s ratio of conditioning by fast speech contexts (FRCRATE) are used to predict the duration of Spanish /s/ and words containing /s/ in a corpus of spoken Spanish. Results show that words’ cumulative exposure to relatively fast speech affects phonetic realizations independent of factors operative in the production contexts. Thus, word and segment rates reflect cumulative (lexicalized) effects of words’ experience in fast speech contexts. The results suggest that lexically specific cumulative measures should be incorporated into models of linguistic variation and change.

La expresión «para nada» y su evolución en el español oral. La información del uso en los diccionarios

Este trabajo analiza la evolución del uso de para nada desde su aparición con valor final hasta el momento actual en el que aumenta el uso como actitud de rechazo. Además de los tres diccionarios que sirven como punto de partida, el análisis se realiza sobre 388 casos correspondientes a tres períodos (1975-1985, 1992-2002 y 2006-2016) del Corpus de Referencia del Español Actual y del Corpus del Español del siglo XXI. En los diccionarios analizados se han revisado las marcas de uso, cuya información gramatical y lexicográfica nos parece insuficiente. El uso adecuado o no de para nada como rechazo absoluto depende, en buena medida, de factores pragmáticos que informan sobre el contexto, el nivel de lengua y la interacción entre los participantes. This paper analyses the evolution of the use of the expression para nada ever since its appearance to convey purpose, up to the present time, to convey an attitude of rejection. Besides the three dictionaries, which served as the starting point for the research, the analysis covers 388 cases over three time periods (1975-1985, 1992-2002, and 2006-2016) taken from the Corpus de Referencia del Español Actual (Current Spanish Reference Corpus) and theCorpus del Español del siglo XXI (21st Century Spanish Corpus). The analysis of the usage labels in the dictionaries revealed insufficient grammatical and lexicographical information. The proper or improper use of para nada as total rejection largely depends on pragmatic factors that inform about the context, language level and interaction between the participants.

Os operadores discursivos ahora bien / ahora, (que) e as suas correspondências em traduções literárias para português

Following previous work by the authors (Ponce de León & Duarte 2013; Duarte & Ponce de León 2015; Duarte & Ponce de León 2017; Duarte & Ponce de León 2018; Ponce de León & Duarte 2020), the paper aims to analyse the discourse values of the Spanish forms ahora (que) / ahora bien - studied in detail by specialists (Martín Zorraquino & Portolés 1999; Santos Río 2003; Fuentes 2009; Loureda & Acín 2010) - and their correspondences with Portuguese discourse markers, by analysing Portuguese translations of Spanish literary texts. The paper is divided in three parts: a) In the first part, following our previous work (Ponce de León & Duarte 2020) and other studies - some of them cited above - that analyse discourse markers, the different discourse values assumed by these forms are briefly presented and analysed, from a synchronic and diachronic perspective, based on grammatical and lexicographical resources, as well as on Spanish corpora. b) In the second part, the correspondences of the analysed values of ahora (que) / ahorabien with Portuguese expressions, ora and agora are determined. Studies on these Portuguese markers (Duarte 1989; Gonçalves 2004; Marques & Rei 2020) are used to establish these correspondences. c) In the third part, the correspondences of ahora (que) / ahora bien in a corpus of translations into Portuguese of Spanish-language literary texts are studied, trying to clarify the acceptability of different translation solutions for each value of the forms ahora (que) / ahora bien.

Priming as a Diagnostic of Grammatical Constructions: Second-Person Singular in Chilean Spanish

Structural priming has been described as a measure of association between constructions. Here, we apply priming as a diagnostic to assess the status of the Chilean second-person singular (2sg) voseo, which exists in variation with the more standard tuteo. Despite being the majority variant in informal interactions, Chileans are reported to have little metalinguistic awareness of voseo and they avoid the vos pronoun, in some cases using the tú pronoun with voseo verb forms, leading to proposals that tuteo and voseo are conflated into a single mixed form. The patterning for priming, however, indicates otherwise. Analyses of some 2000 2sg familiar tokens from a corpus of conversational Chilean Spanish reveal that a previous tuteo or voseo favors the repetition of that same form, indicating that speakers do treat these forms as distinct. We also observe that invariable forms with historically tuteo morphology are associated with neither voseo nor tuteo, while the invariable voseo discourse marker cachái ‘you know’ retains a weak association with voseo. Furthermore, while tuteo is favored with a tú subject pronoun, this effect does not override the priming effect, evidence that, even with a tú pronoun, voseo and tuteo are distinct constructions in speakers’ representations.

Exploración del cuantificador “todo” como recurso de GRADACIÓN de valoraciones en un corpus de español oral conformado por un grupo de madres de Santiago de Chile

Este artículo describe el uso del cuantificador todo como recurso de gradación (Martin y White 2005; Hood 2010, 2019) en un corpus de español oral, compuesto por los relatos de un grupo de madres de Santiago de Chile. Los resultados muestran que el cuantificador todo realiza significados congruentes con su función cuantificadora y, adicionalmente, despliega realizaciones menos congruentes, donde cumple una función intensificadora, como también la de graduar una categoría semántica en términos de prototicipidad y de completitud. El análisis demuestra que estas realizaciones de todo como recursos de gradación tienen un rol significativo en la evocación de valoraciones involucradas en la construcción de identidades maternas particulares.

Simple metathesis in loanword phonology: the Arabic-Romance language contact

AbstractSimple metathesis can be found in numerous Ibero-Romance arabisms compared with their Andalusi Arabic etyma. The analysis of a corpus of Spanish, Portuguese and Catalan arabisms illustrates its effects on syllable structure and syllable weight. It can be shown that Arabic-Romance simple metathesis constitutes a motivated structural change that provides for typologically unmarked syllable weight relations within the word. After the resyllabification it entails the involved unstressed syllables no longer excede the stressed syllable in weight. However, it is not an obligatory, systematic process, but merely an optional tendency, which corresponds to the universal tendency expressed by the Weight Law.

Apéndices interrogativos de control de contacto en el habla de inmigrantes colombianos en Madrid

Since the early 20th century reference has been made to the existence of a phatic or contact function in human communication, although only recently has it begun to be studied in depth. This function is realised through the use of various verbal and non-verbal resources with different ends: to ensure that the communication channel is open and working; to initiate, prolong or end communication; and to attract the interlocutor’s attention. With a view to understanding better how phatic resources work, what functions they perform in interaction, and how they are influenced by certain social (sex, age, education) and geolectal factors, we are currently engaged in a socio-pragmatic study which will enable us to document general and variable behaviour patterns. Focusing on the first of those subfunctions, this paper analyses the appearance of contact control question tags in the Colombian immigrant sub-corpus of the Dynamic Corpus of Immigrant Spanish (Corpus Dinámico del Español de la Inmigración –CORDIESIN). The results provide us with a first approximation to which verbal resources (question tags) are most commonly used by Colombians resident in Madrid and how they function. They also permit us to make some comparison with the corresponding state of affairs in the speech of Spanish Madrilenians, the upshot being that there is convergence with the host community.

LiSSS: A New Corpus of Literary Spanish Sentences for Emotions Detection

In this work we present a new corpus in the area of Computational Creativity (CC), the Literary Sentiment Sentence Spanish Corpus (LiSSS). We address this corpus of literary sentences in order to evaluate algorithms of sentiment classication. We have constitute it by manually classifying its sentences in ve emotions: Love, Fear, Happiness, Anger and Sadness/Pain. We also present some baseline classication algorithms applied on our corpus. The LiSSS corpus will be available to the community as a free resource to evaluate or create CC algorithms.