Topics for internship, lab rotation, and thesis research

Gezocht – student-assistenten:
1 voor de ‘ISLA language game’ (zie helemaal onderaan); en
1 vnl. voor ‘Mijn Spraaktrainer’ []

- In our research (PI) group, research projects there are several possibilities for internship and thesis research. You can assist in our experiments, or set up your own experiments. In general, we try to draw up research plans together with the students involved, taking into account their background, interests and future possibilities. We are flexible, and believe that this is advantageous for you and us.
If you are interested, have questions, me, e.g. send an e-mail to Helmer – ‘w.strik -at-’, and then we can provide more information about current possibilities.
- Our research (at the university) ranges from fundamental to applied, and at the spin-off company NovoLanguange we develop products.
For more information on our research see e.g. research projects, some [demo-]videos, and for publications see e.g. my researchgate and my publications page.
In our research we develop applications, and in turn these applications are employed to carry out innovative research. Topics in our research are eLearning, ‘computer-assisted language learning’ (CALL), second language acquisition, eHealth, speech therapy, language and speech technology (L&ST), cognition (e.g. measuring brain activity during speech production and perception), etc.
- Therefore, topics for internships and theses can have different types of focus. Technology can be an important aspect (e.g. you can contribute to developing language and speech technology), a minor aspect (e.g. you can use the applications, with language and speech technology), or not be involved at all.

Some examples (brief):

In the project ‘Idiomatic Second Language Acquisition’ (ISLA) we study the acquisition of formulaic expressions (FEs), such as idioms (opaque or not), speech formula (‘dat meen je niet!’, ‘hoe kom je erbij?’, ‘bekijk het maar!’), multiword expressions (‘op een gegeven moment’, ‘in ieder geval’), etc. German (L1) learners of Dutch (L2) are offered intensive training in FEs through an available Computer Assisted Language Learning (CALL) system.
Possible research topics/questions are:

  • How can an ASR-based CALL system provide optimal training and feedback in FE production that is comparable to teacher tutoring?
  • How can the ASR component be optimized to handle the peculiarities of German accented Dutch?
  • Which aspects of learner behaviour should be monitored to gain insight into the processes underlying FE learning?
  • How can the CALL system be improved, adapted so that it can handle more challenging tasks related to FE acquisition, and to provide appropriate feedback.
  • How can the ASR component be adapted and optimized to deal the Dutch spoken by German learners?
  • Interactions of the users with the system are stored in log-files. How can these log-files be used, and possibly improved, to obtain insights in acquisition of FEs?
  • Later we will carry out experiments in which behavioural and online (cognitive) measurements will be made. You can be involved in (preparing, running, analysing) these experiments, or carry out some (pilot-)experiments.
  • We also carry out ´smaller scale web-experiments´ in which we test to what extent natives and non-natives are familiair with these FEs.

Dysartrische spraak – hiervan hebben we al veel opnames (spraaksignalen), waarmee o.a. onderstaande onderwerpen (of combinaties daarvan) onderzocht kunnen worden :

  • Waarom zijn mensen met dysartrie moeilijk te verstaan, en wat kan gedaan worden om hun verstaanbaarheid te verbeteren?
  • Vergelijking van dysartrische en (neurologisch) gezonde spraak, en het verzamelen van geschikt materiaal hiervoor: criteria, akoestische analyse, oordelen van luisteraars (o.a. m.b.t. verstaanbaarheid), etc.
  • Kenmerkende, objectieve akoestische eigenschappen van dysartrische spraak, zoals toonhoogte (gemiddelde en variatie), luidheid, grootte v.d. klinkerruimte, etc.
  • Subjectieve oordelen van luisteraars (met verschillende procedures, schalen): verstaanbaarheid, begrijpelijkheid, spraakkwaliteit, etc.
  • Verbeteren van ‘automatische spraakherkennig’ (ASH), bijv. door het trainen van verschillende akoestische en/of taalmodellen met verschillende ASR-technieken.
  • Automatische detectie van uitspraakfouten, voor onderzoek m.b.v. spraakcorpora, en voor het inzetten van die algoritmes bij spraaktraining.
  • Ontwikkelen en testen van apps voor spraaktraining: feedback (wat, wanneer, hoe, …), wat is mogelijk, wat is effectief, etc.?

´De Digitale Luistervink´, een e-Health project samen met het RadboudUMC en Apps4Air B.V. Voor Astma en COPD patienten ontwikkelen we digitale spraak- en ademanalyse zodat via de smartphone gecontroleerd kan worden hoe het met de patiënt gaat. We zijn al begonnen met de registratie van signalen, maar er moeten nog meer opnamen gemaakt worden. Studenten kunnen assisteren bij het registreren en analyseren van de signalen, en bedenken van methodes hoe de spraak- en ademanalyse thuis het beste ingezet kan worden.
Zie ‘Spraakverandering als vroege marker longaanval’, en ‘Voice and sound recognition for exacerbation management’.

CLS Valorisation Grant (VG) for an ISLA language game
In dec. 2016, we have been awarded the CLS Valorisation Grant (VG) to develop and test an ISLA language game.
Through this Valorisation Grant, we intend to further develop an existing test into an auditory, simple language game that can be played by everybody, even people with a visual handicap, which will be enabled through speech synthesis. In this way, people will have the opportunity to evaluate and improve their knowledge of idiomatic expressions. Simultaneously, we can collect more data, also from the new group of visually impaired language speakers. This will not only open up new opportunities for valorisation, but also adds considerably to the scientific impact of our research. One important issue in research on figurative language is the contribution of the dimension ‘imageability’, the extent to which an expression can be visualized. Imageability has been found to play an important role in learning idiomatic expressions. The data obtained from visually impaired participants allow us to investigate how these speakers represent idiomatic expressions, and whether imageability plays a role for visually impaired speakers. To increase exposure, the auditory game will be exhibited in the new section of the MuZIEum in Nijmegen, ‘Krijg je het inZICHT?’, which illustrates the possibilities language and speech technologies offer to the visually impaired.