Linguists have always been at the forefront of the corpus revolution in the humanities. It still proves hard though to bring together the interests of computationally oriented linguists with those of more theoretically oriented ones, though. We argue that progress can be made by applying quantitative corpus methods in the field of semantic micro-typology, in particular by exploiting the possibilities of translation corpora. To do so, we focus on one of the most challenging tense-aspect categories found across languages: the Perfect. Its use at the sentence and discourse level varies across languages, and it competes with past and present tenses. Instead of avoiding this variation, we embrace it to unveil the meaning of the Perfect, using a ‘smart’ integration of quantitative and qualitative methodology in a data intensive approach. Over the next couple of years, we aim to develop a micro-typology of the Have Perfect grounded in a technique we dub Translation Mining (Wälchli & Cysouw 2012), based on translation equivalences between English, Dutch, German, French and Spanish. The analysis has three key ingredients: (i) a semantic map of the sentence-level meanings of the Perfect, (ii) a semantic map of the discourse interaction usages of the Perfect, (iii) an integrated truth-conditional and inquisitive semantics of the Perfect. The project sets a gold standard for the integration of quantitative corpus methods in theoretical linguistics. It is further developed as a basis for new finer-grained analysis of L2 tense/aspect acquisition, to promote inquiry-based learning in the five school languages the project represents and to help translators by means of the development of an online course module and a translation software plugin (MIT license).
The project offers opportunities for internships and thesis research to BA/MA students of linguistics, artificial intelligence, translation, education and any of the language programmes (English, Dutch, German, French, Spanish, possibly others). We hope to extend these opportunities to research on L2 acquisition in the nearby future. Feel free to send an e-mail to one of the project leaders if you are interested in joining our perfect investigations!
Our team is based at Utrecht University and currently consists of:
We cordially invite you to attend our fourth expert meeting on Wednesday 21 March, 9.30-12.00 (Drift 21, room 1.05, 3512BR Utrecht). The focus of the meeting will be on (in)definiteness across languages. Click here for more details.
Our project organizes several small-scale expert meetings. The meetings will consist of two half-hour talks followed by discussion. We cordially invite you to attend the third meeting on Friday 16 March, 9.30-12.00 (Drift 23, room 0.13, 3512BR Utrecht). The focus of the meeting will be on competition between past, present perfect and recent past. Click here for more details.
Martijn presented his work on the perfect of recent past at the Grote Taaldag conference in Utrecht. You can download the slides here.
Martijn presented his work on lexical preferences in the Perfect construction at the CLIN28 conference in Nijmegen. You can download the poster here.
Henriëtte presented our work at the Workshop in honour of Barbara Partee's honorary doctorate conference at the ILLC in Amsterdam. You can download the slides here.
Henriëtte presented our work at the Autour de L'Étranger de Camus et de ses traductions conference at Université Sorbonne Nouvelle in Paris. You can download the slides here.
Our project will organize several small-scale expert meetings. The meetings will consist of two half-hour talks followed by discussion. We cordially invite you to attend the first meeting on Tuesday 7 November, 10.30-12.00 (Drift 23, room 0.12, 3512BR Utrecht). The focus of the meeting will be on aspect from a multilingual perspective. Click here for more details.
Our second expert meeting will be on tense and aspect in second-language acquisition, on Thursday 30 November, 10.30-12.00 (again in Drift 23, room 0.12). Speakers will be Paz González (Leiden University) and Laura Dominguez (University of Southampton). Further details TBA.
We presented our work to the public at the DRONGO language festival. We talked with several people about valorization, and hope to put this into practice later on in the project.
Henriëtte presented our work at the Workshop on Logic and Algorithms in Computational Linguistics 2017 at Stockholm University, Sweden. You can download the slides here.
We updated our overview of student research. Please do contact us if you would want to do your BA/MA thesis/internship with us!
At Friday June 23rd, we had a great kick-off workshop for our project. You can find all abstracts and slides here.
At Friday June 23rd, we will have the kick-off workshop for our project. We're bringing together linguists from different kinds of backgrounds and with different aims and ask them to reflect on corpus methodology in their work: Stephan Th. Gries, Eva Vanmassenhove, Martijn van der Klis, Antonio Toral, Tommaso Caselli, Jet Hoek & Nicholas Asher. Click here to check the full programme and abstracts. Participation is for free but it would be great if you could send a short email to Bert Le Bruyn if you’re intending to join for lunch.
We are grateful to NWO for the funding for the Time in Translation programme we received through the Free competition programme. This funding means that 2 PhD positions will open up in 2017, and a post-doc will be hired in 2018. Martijn van der Klis will start as PhD1 as of June 2017.
Henriëtte presented our work at the UiL OTS Colloquium at Utrecht University, The Netherlands. You can download the slides here.
With Anne Verkleij and Vincent Wimmers we're currently exploring the differences between formal and informal dialogue by comparing our data from the Europarl corpus to those based on movie subtitles. With Mandy Woelk we're exploring the statistical potential of Analyses of Similarities.
We are currently working on improving the attribution of tenses. We are aware that the current (mostly automatic) attribution will not always yield the correct result, and are of course aiming to minimize the error rate.