Eurac Research CLARIN Centre Repository
Not a member yet
43 research outputs found
Sort by
AThEME Verona-Trento Corpus
No full textThe AThEME Verona-Trento Corpus is a spoken corpus composed of data collected during the AThEME project in Work Package 2 ‘Regional Languages’ by the units of Verona and Trento for minority languages and dialects spoken in the area between Innsbruck and the Po Valley (Tyrolean, Trentino, Fodom Ladin, Fassan Ladin, Mòcheno, Cimbrian, and Venetan). The corpus also contains data on the Germanic minority languages Timavese (PRIN 2017) and Saurano. The corpus contains audio recordings and partial transcriptions of the responses to a phonological questionnaire (topics: obstruents, final devoicing, s-retraction, realization of /r/) and a morpho-syntactical questionnaire (topics: adjectives, pronouns, auxiliary selection, pro-drop, complementizers). The data collection was done via linguistic fieldwork interviews and took place between 2014 and 2019
ACTER (Annotated Corpora for Term Extraction Research) v1.5
No full textACTER (Annotated Corpora for Term Extraction Research) is a manually annotated dataset for term extraction, covering 3 languages (English, French, and Dutch), and 4 domains (corruption, dressage, heart failure, and wind energy)
MT@BZ annotation guidelines v1.0
No full textThe MT@BZ annotation guidelines are guidelines for legal Italian-German machine translation quality assessment. Particularly, they cover the South Tyrolean German variety. They are based on version 1.3.3 of the Annotation Guidelines for English-Dutch Machine Translation Quality Assessment (https://www.lt3.ugent.be/publications/annotation-guidelines-for-english-dutch-machine-tr/). The guidelines also include specific instructions on how to annotate errors in WebAnno/INCEpTION and which sources to consult when assessing the correctness of a translation
Kolipsi-2 Corpus v1.0
No full textThe Kolipsi-2 Corpus is a written learner corpus of German and Italian L2 speakers originating from South Tyrol (Italy). It has been developed as a by-product of the KOLIPSI II project, a replication study of the KOLIPSI project on “South-Tyrolean pupils and the second language: a linguistic and socio-psychological investigation” that was conducted 7 years after the original study.
The data collection for this second edition took place in spring 2014 and is based on two standardized tests for written productions, that were aligned with the original tasks for the KOLIPSI study. However, while the first task remained the same for both editions, the second task was slightly adapted. The two tasks consisted of (1) writing an e-mail to a friend retelling a given event at the supermarket based on a picture story (narrative text genre) and (2) writing an e-mail about negative aspects of social-media chats prompted by a letter to the editor in a youth magazine (argumentative text genre). For both tasks a time limit of 25 minutes was fixed and no additional reference material was allowed.
CEFR levels have been assigned to all L2 learner texts, providing a holistic score as well as evaluations of coherence, sociolinguistic appropriateness, lexical accuracy, lexical diversity, grammar and orthography.
Person-related metadata provides information about:
- the writer's language background, including L1(s), the L1(s) of mother and father, and a self-declared language group affiliation as well as the pre-dominant language spoken in the area the writer is residing in
- the writer's results from an additional language test in the L2 (dialang test)
- the writer's competence in the local German dialect (for students with L1 Italian only)
- the writer's age, gender and socio-economic status
- whether the writer lives in an urban or rural environment
- the language, location and type of school the writer attended
- an anonymous identifier for the writer's school class to account for class effects
All texts have been transcribed manually adding transcription annotations that reflect surface features of the text, such as the graphical arrangement, and include error annotation on the orthographic level.
In addition to that, all texts were automatically annotated, adding tokenisation, sentence splitting, POS-tagging and lemmatization using an orthographically corrected target version of the corpus.
Kolipsi-1 L2 belongs to the Kolipsi Corpus Family, a series of related learner corpora collected in South Tyrolean upper secondary schools. The corpora of the Kolipsi Corpus Family contain Italian and German learner texts that were collected in the course of the KOLIPSI project in 2007/2008 (Kolipsi-1) and a follow-up study in 2014/2015 (Kolipsi-2). The aim of both corpus studies was to analyse the second language competences of South-Tyrolean pupils from upper secondary schools (between 16-18 years old), and to contextualize the results of such investigation by commenting on crucial sociolinguistic and psychosocial aspects that influence it. The results of the follow-up study should be compared to the results of the original KOLIPSI project
VinKo (Varieties in Contact) Corpus v1.0
No full textVINKO is a spoken corpus based on crowdsourced audio recordings that has been designed to provide relevant linguistic information about the minority languages and dialects spoken in the area between Innsbruck and the Po Valley. The corpus contains audio recordings from local languages and varieties spoken in the regions Trentino-Alto Adige/Südtirol, Veneto, and Friuli-Venezia Giulia, with particular focus on the so-called 'language contact' between Germanic (Cimbrian, Mòcheno, Tyrolean, Saurano) and Romance (Ladin, Trentino, and Veneto dialects). The data collection took place from June 2017 to May 2021
Kolipsi-1 Corpus v1.0
No full textThe Kolipsi-1 L2 is a written learner corpus of German and Italian L2 speakers originating from South Tyrol (Italy). It has been developed as a by-product of the KOLIPSI project “South-Tyrolean pupils and the second language: a linguistic and socio-psychological investigation”. In addition, data from L1 pupils were collected exclusively for the creation of a native speaker reference corpus.
The data collection took place in autumn 2007 and is based on two standardized tests for written productions. The two tasks consisted of (1) writing an e-mail to a friend retelling a given event at the supermarket based on a picture story (narrative text genre) and (2) in writing a letter to a friend discussing holiday plans (argumentative text genre). For both tasks a time limit of 30 minutes was fixed and no additional reference material was allowed.
CEFR levesl have been assigned to all L2 learner texts, providing a holistic score as well as evaluations of coherence, lexis, grammar and sociolinguistic appropriateness.
Person-related metadata provides information about:
- the writer's language background, including L1(s), the L1(s) of mother and father, and a self-declared language group affiliation
- the writer's age, gender and socio-economic status
- the writer's district of residence and whether he lives in an urban or rural environment
- the language, location and type of school the writer attended
- whether the writer passed the local bilinguality exam or not
- an anonymous identifier for the writer's school class and L2 teacher to account for class effects
All texts have been transcribed manually adding transcription annotations that reflect surface features of the text, such as the graphical arrangement, and include error annotation on the orthographic level.
In addition to that, all texts were automatically annotated, adding tokenisation, sentence splitting, POS-tagging and lemmatization using an orthographically corrected target version of the corpus.
Kolipsi-1 L2 belongs to the Kolipsi Corpus Family, a series of related learner corpora collected in South Tyrolean upper secondary schools. The corpora of the Kolipsi Corpus Family contain Italian and German learner texts that were collected in the course of the KOLIPSI project in 2007/2008 (Kolipsi-1) and a follow-up study in 2014/2015 (Kolipsi-2). The aim of both corpus studies was to analyse the second language competences of South-Tyrolean pupils from upper secondary schools (between 16-18 years old), and to contextualize the results of such investigation by commenting on crucial sociolinguistic and psychosocial aspects that influence it. The results of the follow-up study should be compared to the results of the original KOLIPSI project
LEKO v1.0
No full textThe LEKO corpora LEKO_Kolipsi and LEKO_Merlin provide lexical annotations for phraseological elements in Italian L2 writing on the basis of a subset of the texts of the Kolipsi-1 corpus and the Merlin corpus respectively. The annotations were jointly created by the University of Innsbruck (Austria) and Eurac Research Bolzano (Italy) within the project LEKO, whose aim was to describe the use of phrasemes in these texts. There are manual annotations for phraseme category, lexical errors, morpho-syntactic features and error explanations.
LEKO_Kolipsi contains about 55 000 tokens in 282 texts from 141 pupils of the final year of upper secondary school, representing two different text types (email and letter, narrative and argumentative genre) as described in the Kolipsi-1 documentation.
LEKO_Merlin contains about 9 000 tokens in 50 texts from 50 examinees, who took part in an official language test (TELC) for Italian.
The documents have been transcribed according to the Kolipsi-1 and Merlin Transcription guidelines. Annotation guidelines for the lexical annotations can be found here.
Note: The LEKO corpora do not contain manual annotations for non-lexical errors, foreign word insertions, target language transcriptions, ambiguous writings or other annotations available in the base corpora Kolipsi-1 and Merlin. In order to retrieve any of those annotations and/or full target versions of the student writings please consult the base corpora directly
Kolipsi-2 Corpus v1.1
No full textThe Kolipsi-2 Corpus is a written learner corpus of German and Italian L2 speakers originating from South Tyrol (Italy). It has been developed as a by-product of the KOLIPSI II project, a replication study of the KOLIPSI project on “South-Tyrolean pupils and the second language: a linguistic and socio-psychological investigation” that was conducted 7 years after the original study.
The data collection for this second edition took place in spring 2014 and is based on two standardized tests for written productions, that were aligned with the original tasks for the KOLIPSI study. However, while the first task remained the same for both editions, the second task was slightly adapted. The two tasks consisted of (1) writing an e-mail to a friend retelling a given event at the supermarket based on a picture story (narrative text genre) and (2) writing an e-mail about negative aspects of social-media chats prompted by a letter to the editor in a youth magazine (argumentative text genre). For both tasks a time limit of 25 minutes was fixed and no additional reference material was allowed.
CEFR levels have been assigned to all L2 learner texts, providing a holistic score as well as evaluations of coherence, sociolinguistic appropriateness, lexical accuracy, lexical diversity, grammar and orthography.
Person-related metadata provides information about:
- the writer's language background, including L1(s), the L1(s) of mother and father, and a self-declared language group affiliation as well as the pre-dominant language spoken in the area the writer is residing in
- the writer's results from an additional language test in the L2 (dialang test)
- the writer's competence in the local German dialect (for students with L1 Italian only)
- the writer's age, gender and socio-economic status
- whether the writer lives in an urban or rural environment
- the language, location and type of school the writer attended
- an anonymous identifier for the writer's school class to account for class effects
All texts have been transcribed manually adding transcription annotations that reflect surface features of the text, such as the graphical arrangement, and include error annotation on the orthographic level.
In addition to that, all texts were automatically annotated, adding tokenisation, sentence splitting, POS-tagging and lemmatization using an orthographically corrected target version of the corpus.
Kolipsi-1 L2 belongs to the Kolipsi Corpus Family, a series of related learner corpora collected in South Tyrolean upper secondary schools. The corpora of the Kolipsi Corpus Family contain Italian and German learner texts that were collected in the course of the KOLIPSI project in 2007/2008 (Kolipsi-1) and a follow-up study in 2014/2015 (Kolipsi-2). The aim of both corpus studies was to analyse the second language competences of South-Tyrolean pupils from upper secondary schools (between 16-18 years old), and to contextualize the results of such investigation by commenting on crucial sociolinguistic and psychosocial aspects that influence it. The results of the follow-up study should be compared to the results of the original KOLIPSI project
Beldeko Summary Corpus v1.0.0
No full textBeldeko Summary Corpus v1.0.0
The Beldeko (Belgisches Deutschkorpus) Summary Corpus is a learner corpus that consists of summaries written by advanced L2 German learners (CEF level B2-C1) with L1 Dutch. It has been created with the aim of investigating the academic writing skills in L2 German of third-year students of two bachelor programmes in Applied Linguistics and Linguistics and Literature, respectively.
The corpus consists of 301 summaries (70774 tokens) written by 115 students of three intact classes (convenience sampling). The texts were collected at Ghent University (in 2013 and in 2014) and University College of Ghent (in 2013) as pre- and posttests of an intervention study on collaborative writing carried out by Carola Strobl in the context of her PhD research (Strobl, C. (2015). Affordances of online technologies for academic writing instruction in a foreign language. Ghent University, unpublished doctoral dissertation). 82 students produced three summaries each (pretest, posttest immediately after the three-weeks-intervention, delayed posttest six weeks after the intervention; missing data are indicated as n.a. in the metadata file) and 33 students produced two summaries each (pretest and posttest, missing data are indicated as n.a. in the metadata file).
The metadata file (Beldeko_Summary_1.0.0_metadata.xlsx) provides information about:
• Institution of data collection (HG= University College of Ghent, UG= Ghent University)
• Year of data collection (2013, 2014)
• Participants´ gender (f, m)
• Number of texts written and number of tokens in each text (T1, T2, T3)
The individual file names of the corpus reveal institution, year, unique ID of participant (per institution per year), text number, in the given order.
The summaries contain between 37-330 words each, with a mean of 230 words (the targeted word count was between 220-250 words). Outliers regarding text length were unfinished texts produced by students who struggled with the time restriction of 60 minutes. The texts were written in class, on computers. Students were allowed to use online auxiliary means such as dictionaries. The task consisted in summarizing two texts (fragments of newspaper articles or interviews or websites) about a topic related to language variation in German each time (Kiezdeutsch, Mundartdebatte in der Schweiz, Viadrinisch, Varianten-Wörterbuch des Deutschen; see also word files provided in metadata). More specifically, the topics were distributed as follows:
Kiezdeutsch: HG_2013_T1, UG_2013_T1, UG_2014_T1
Mundartdebatte in der Schweiz: HG_2013_T2, UG_2013_T2, UG_2014_T2
Viadrinisch: HG_2013_T3,
Varianten-Wörterbuch des Deutschen: UG_2014_T
LEONIDE - Longitudinal Learner Corpus in Italiano, Deutsch and English 1.1
No full textLEONIDE is a longitudinal corpus of student essays documenting the language competences and writing development of lower secondary school students in three different languages.
The corpus contains 2.512 texts from 163 pupils, who participated in the project “One school, many languages” conducted in eight schools in the officially multilingual Italian province of South Tyrol / Alto Adige (Zanasi & Stopfner, 2018). The aim of the project was to document the development of the pupils' plurilingual linguistic and communicative skills by collecting oral and written language samples in Italian, German and English, in order to obtain a global view of their individual linguistic repertoire.
LEONIDE contains all the texts written by the participating students during the course of the project, the overall size of the corpus amounts to ca. 240.000 tokens. The texts were collected over the span of 3 consecutive years (2015-2018) in public middle schools (i.e. lower secondary school, grade 6 to grade 8). The pupils were 11 years old at the beginning of the data collection and 13 years old at the end. In each grade, two written texts were collected that differ with respect to genre: the first text was elicited using a picture story re-telling task; the second text is an opinion text on different aspects related to the pupils’ life and public discourse. For each genre and each grade, the corpus provides texts in the three languages German, Italian and English. In order to reflect the school system of the Province of South Tyrol / Alto Adige, about half of the texts was collected in four schools in which German is the main language of teaching and Italian is taught as L2. The other half of the texts was collected in four schools in which Italian is the main language of teaching and German is taught as L2. In all schools, English is taught as L3 (i.e. as a foreign language at school). Subdivided by language, the corpus contains 844 Italian, 833 German and 835 English texts.
Manual annotation:
The corpus is fully anonymised and annotated with target hypotheses correcting orthography errors in the text as well as annotations on structural elements (paragraphs, line breaks, bullet points, symbols or emoticons etc.), foreign word insertions and transcript surface features (e.g. deletions, corrections or insertions of the student, unreadable or ambiguous items).
Automatic annotation:
Automatic linguistic annotation included sentence splitting, tokenisation, lemmatisation and part-of-speech-tagging.
Text metadata:
The corpus provides a series of relevant person-related metadata (e.g. age, gender, first language(s), school and possible special needs of the students) as well as task-related metadata (e.g. task year, text genre, etc.)
Usage:
As the corpus documents the development of plurilingual competences of individual learners over a period of three years, it will allow both quantitative research on the characteristics of young learners’ language over a relatively long period, as well as investigations of the development of individuals taking into account a wide range of person related metadata. In addition, it allows contrastive analyses of the young learners’ progress in their L1, L2 and L3.
Availability:
The corpus will be available for corpus queries via an ANNIS search interface and as download for academic purposes (ACA-BY-NC-NORED 1.0) on the Eurac Research Clarin Centre by the end of 2020.
References:
Zanasi, L. & Stopfner, M. (2018). Rilevare, osservare, consultare. Metodi e strumenti per l’analisi del plurilinguismo nella scuola secondaria di primo grado. In C. M. Coonan, A. Bier Ada & E. Ballarin (Ed.), La didattica delle lingue nel nuovo millennio. Le sfide dell’internazionalizzazione (pp. 135-148). Edizioni Ca’Foscari. http://doi.org/10.30687/978-88-6969-227-7/009
Glaznieks, A., Frey, J.-C., Stopfner, M., Zanasi, L. & Nicolas, L. (accepted): LEONIDE: A longitudinal trilingual corpus of young learners of Italian, German and English. In: International Journal of Learner Corpus Linguistics