Language Sciences Day 2021
The Language Science Day organized annually by the Utrecht institute of Linguistics OTS (UiL OTS) will be co-hosted this year by the Centre for Digital Humanities (CDH), both at Utrecht University. The theme of this year’s edition is “Language Science and Digital Humanities”.
This Language Science and Digital Humanities Day will be held on Friday, 16 April 2021, from 13:30 to 17:00h. Due to corona restrictions, the meeting will be held online, using MS Teams and (probably) Zoom.
Programme 13:30 to 17:00h, exact times to be announced
by René Kager (UiL OTS, UU) and Hugo Quené (CDH, UU)
by Tejaswini Deoskar (UiL OTS, UU)
Lexico-syntactic patterns for inference in natural language processing systems
Why does the sentence “Alice doesn’t believe that Bob ate” give rise to the inference “Alice believes that Bob didn’t eat”, while the structurally identical “Alice doesn’t know that Bob ate” triggers the inference “Bob ate”, but not “Alice knows that Bob didn’t eat”?
The formal linguistics and semantics literature has long been concerned with the complex array of inferences that are triggered by different open class lexical items — the broad consensus is that lexically-triggered inferences are conditioned by various aspects of the syntactic contexts that the lexical item occurs in. Such lexico-syntactic interactions — e.g. between clausal verbs, embedded clause type and negation — are necessary to capture in large scale natural language understanding systems that aim to do general language inference, but are rarely explicitly modelled. This talk will discuss the specific challenges faced in modelling and learning lexicon-syntactic patterns from data in language processing systems, the state-of-the-art and role of linguistic theory, and my on-going work in this area. I will further discuss implications for the problem of “generalisation” and domain-adaptation of language processing systems, i.e., their applicability to genres or domains of text outside the domain of their training data.
poster and demo session
brief plenary pitches of posters and demonstrations
poster presentations and demonstrations
|Jan Odijk and Jelte van Boheemen||Semi-Automatic Analysis of Spontaneous Language|
|Martijn van der Klis||Project Time in Translation|
|Konstantinos (Kokos) Kogkalidis||From text to meaning: a type-driven Dutch treebank and its applications|
|Anika van der Klis||Using open-source speech recognition for the annotation of infant-directed speech|
|Anne-France Pinget and Hugo Quené||Modeling effects of obstruent voicing on vowel pitch frequency in Dutch|
|Karin Wanrooij||Can the mismatch response be used to predict or diagnose dyslexia? (a systematic review)|
by Kees van Deemter (Information and Computing Sciences, UU)
Explanation and Rationality in Models of Language Use
When theories of human behaviour aim to offer explanations, they often use rationality as their linchpin: to the extent that a theory helps us to see behaviour as optimising some form of rationality/utility, we feel that our theory explains this behaviour. This approach is not uncontroversial, however. For example, four decades of research in Behavioural Economics have shown that people behave in ways that are not easily explained by rationality alone.
Rationality has long had its adherents in the explanation of language use as well, for example via the Gricean Maxims. Recently, a Bayesian approach known as Rational Speech Act (RSA) theory has made inroads into the computational modelling of language use. In a nutshell, the idea is to build tightly coupled models of language comprehension and production in which speakers and hearers assume each other to behave rationally.
In this talk I will sketch a series of experiments focussing on the way in which speakers refer. These experiments paint a less “rational” picture of human language use, and they offer confirmation of a model, known as Probabilistic Referential Overspecification (PRO), that balances rationality with other considerations. I hope to engage in a discussion of the dilemma of having to choose between two these very different models, one of which is elegant and explanatory yet empirically inadequate, while the other is messy yet empirically very adequate.
Conceptualization in Reference Production: Probabilistic Modeling and Experimental Testing. R.P.G. van Gompel, K. van Deemter, A. Gatt, R. Snoeren, E.J. Krahmer (2019). Psychological Review 126 (3), 345-373.
closure formal programme online — informal meetings