Michael Walsh

Michael Walsh

Telefon 0049 711 685-81382
Universität Stuttgart
Institut für Maschinelle Sprachverarbeitung
Pfaffenwaldring 5b
70569 Stuttgart


My principal area of research is concerned with exemplar-theoretic accounts of language acquisition. The key idea behind Exemplar Theory is the notion that the acquisition of language is significantly facilitated by repeated exposure to concrete language input. As we are developing we store language input in memory, in rich detail, and our emergent linguistic ability is heavily influenced by how similar new language input is to language exemplars we have stored in mind, how frequently we are exposed to such input, and how recently too. Much of my work focuses on computational modelling, in particular the modelling of local grammatical knowledge. I am also interested in how Exemplar Theory can be applied to language technology, with a particular focus on statistical language modelling, and how intonation and the mental lexicon interact.


Seminar: Probabilistic models of language and cognition

This Masters level course includes introductory lectures on probability, Bayesian inference, and aspects of machine learning, which serve as a basis for seminar discussions concerning research papers focussing on the following areas: probabilistic phonology, exemplar theory, word acquisition, syntax acquisition, and connectionism.



Peer reviewed journals:


Book chapters:

  • Julie Carson-Berndsen, Gina Joue and Michael Walsh (2001) Phonotactic constraint ranking for speech recognition. In: Computational Linguistics in the Netherlands 1999. Selected Papers from the Eleventh CLIN Meeting.


Peer reviewed conferences, workshops, and abstracts:

  • Antje Schweitzer and Michael Walsh (2016). Exemplar Dynamics in Phonetic Convergence of Speech Rate. Proceedings of Interspeech 2016, San Francisco, USA, pp. 2100-2104.
  • Michael Walsh, Katrin Schweitzer, Hinrich Schütze and Dermot Lynott (2014): Examining the influence of pitch accents on word learning in German. Proceedings of the 15th Australasian International Speech Science and Technology Conference (SST), Christchurch, New Zealand.
  • Michael Walsh, Katrin Schweitzer and Nadja Schauffler (2013). Exemplar-based pitch accent categorisation using the Generalized Context Model. Proceedings of Interspeech 2013, Lyon, France, pp. 258-262
  • Junfei Guo, Juan Liu, Michael Walsh and Helmut Schmid (2013). Class-Based Language Models for Chinese-English Parallel Corpus. In Alexander F. Gelbukh (Ed.): Computational Linguistics and Intelligent Text Processing - 14th International Conference, CICLing 2013, Samos, Greece, Springer 2013 Lecture Notes in Computer Science ISBN 978-3-642-37255-1
  • Daniel Duran, Jagoda Bruni, Michael Walsh, Hinrich Schütze, and Grzegorz Dogil (2012). Phonological constraints verified by a rich memory exemplar model: extrametricality and articulatory binding in Polish obstruent-sonorant rhymes. Book of abstracts of the 13th Conference of Laboratory Phonology, Stuttgart.
  • Michael Walsh, Daniel Duran, and Jagoda Bruni (2013). Exemplar-based categorisation of vowels using  acoustic and articulatory data. Book of abstracts of Phonetik & Phonologie 9, Zurich
  • Michael Walsh, Gregory M.P. O'Hare and Julie Carson-Berndsen (2005): An Agent-based Framework for Speech Investigation. In Proceedings of Eurospeech 9th European Conference on Speech Communication and Technology, Interspeech 2005, September 4th-8th Lisbon, Portugal, 2005. 
  • Robert Kelly, Moritz Neugebauer, Michael Walsh and Stephen Wilson (2004) Annotating Syllable Corpora with Linguistic Data Categories in XML. In Proceedings of the INTERA-ISO workshop on "A Registry of Linguistic Data Categories within an Integrated Language Resource Repository Area" held in conjunction with the Fourth International Conference on Language Resources and Evaluation (LREC 2004), Lisbon, Portugal.
  • Michael Walsh (2003) Recasting the Time Map Model as a Multi-Agent System. In Proceedings of the 15th International Congress of Phonetic Sciences 2003, Barcelona, Spain.
  • Michael Walsh, Robert Kelly, Julie Carson-Berndsen, Gregory M.P. O'Hare and Tarek Abu-Amer (2003) A Multi-Agent Computational Linguistic Approach to Speech Recognition. In Proceedings of 18th International Joint Conference on Artificial Intelligence (IJCAI-03), Acalapulo, Mexico, AAAI Press, 2003.
  • Stephen Wilson, Julie Carson-Berndsen, and Michael Walsh (2003) Enhancing Phonological Representations for Multilingual Speech Technology. In Proceedings of the 15th International Congress of Phonetic Sciences 2003, Barcelona, Spain.
  • Robert Ross, Brian McEleney, Robert Kelly, Tarek Abu-Amer, Michael Walsh, Julie Carson-Berndsen and Gregory M.P. O'Hare (2002) Speaking Autonomous Intelligent Devices. In Proceedings of the 13th Irish Conference on Artificial Intelligence and Cognitive Science (AICS '02),Limerick, Ireland.
  • Michael Walsh, Stephen Wilson and Julie Carson-Berndsen (2002) XiSTS - XML in Speech Technology Systems. In Proceedings of the COLING 2002 Workshop on XML & NLP, Taipei, Taiwan.
  • Julie Carson-Berndsen and Michael Walsh (2001) Defining constraints for multilinear speech processing. In Proceedings of Eurospeech 2001, Aalborg, Denmark.
  • Julie Carson-Berndsen and Michael Walsh (2000) Generic Techniques for Multilingual Speech Technology Applications. In Proceedings of TALN 2000, Lausanne, Switzerland.
  • Julie Carson-Berndsen and Michael Walsh (2000) Interpreting Multilinear Representations in Speech. In Proceedings of the 8th Australian Conference on Speech Science and Technology, Canberra, Australia.