TABU dag 2018

is supported by

NWO logo

CLCG logo

Cafe Hooghoudt logo

PMB logo

GUF logo

Taal Unie logo

BCN logo

Clariah logo

R-Studio logo

NDI logo

Fryske Akademy logo

Fryske Akademy logo

Optima Grafische Communicatie logo

CCG logo

Meertens Instituut logo

CCG logo

Keynote Speakers

We are proud to have the following keynote speakers presenting at 39th TABU Dag:

Barbara Partee

Distinguished University Professor Emerita of Linguistics and Philosophy at the Department of Linguistics, University of Massachusetts Amherst

Barbara H. Partee is a Distinguished University Professor Emerita of Linguistics and Philosophy at UMass-Amherst. She is one of the founders of contemporary formal semantics. It was through her interaction with the philosopher and logician Richard Montague in the 1970s that the research traditions of generative linguistics and formal analytic philosophy first converged. She was largely responsible for popularizing this approach to the semantics of natural languages among linguists, especially so at a time when much of the field was gripped by controversy surrounding Generative Semantics.

Talk: Lexical Semantics in Formal Semantics: History and Challenges

The early achievements of formal semantics that made it of interest to linguists were in compositionality, i.e. ‘the semantics of syntax’, and in the semantics of a wide range of function words or morphemes and the constructions in which they occur -- determiners, tense and aspect markers, plurality, negation, and more. It was common to specify semantic types for open-class lexical items, but to leave their meanings as unanalyzed primitives.

Over the years, there has been great progress on the study of some semantic properties of open-class words, including aspectual properties of verbs, context-sensitive properties of words like local and enemy, polarity items, relational nouns, and more. These lexical studies have all been driven by the goals of compositional semantics. What has still remained unanalyzed is the “remainder” of the content of open-class words.

The question of whether and how to add a fuller treatment of lexical content to formal semantics has a complex history that I will sketch, but was a side issue until recently among formal semanticists, though not in the context of cognitive and computational perspectives.

In the talk I’ll discuss foundational issues concerning the nature of semantic competence. The question of whether there is a difference between describing a person’s language and describing their competence in that language arises especially sharply for lexical semantics, I believe. I will also discuss lexical semantics before and in the beginnings of formal semantics, concentrating on key issues and disputes. I’ll note some achievements of formal semantics in the realm of lexical semantics, empirical achievements that haven’t required resolving foundational questions.

Bridging past and future, I’ll discuss meaning postulates – some of their appeal, how they connect to some other theories, and how they may help us avoid some theoretical pitfalls in the realm of lexical semantics.

Felix Hill

Reserch Scientist at DeepMind

Felix is a Research Scientist at Deepmind. He did his PhD at the University of Cambridge with Anna Korhonen, working on unsupervised language and representation learning with neural nets. As well as Anna, he has collaborated with (and learned a lot from) Yoshua Bengio, Kyunghyun Cho, Jason Weston and Jay McClelland. As well as developing computational models that can understand language, he is interested in using models to better understand how people understand language, and is currently doing both at Deepmind.

Talk: End-to-End Grounded Language Learning in Simulated Worlds

Developing systems that can understand and execute natural language instructions in the physical world is a long-standing challenge for Artificial Intelligence. Previous attempts to replicate human-like grounded language understanding involved hard-coding linguistic and physical principles, which is notoriously laborious and difficult to scale. Here we explore approaches to the problem based on deep learning and reinforcement learning agents in simulated environments. Without any hard-coded knowledge, these agents exploit their general-purpose learning algorithms to infer the meaning of sequential symbolic instructions, pertaining to objects, locations and relations, and approximating a range of linguistic phenomena. These agents naturally generalise beyond their training experience, for instance, extending predicates to unfamiliar objects, and interpreting novel word combinations. Moreover, while the initial learning is slow, the speed at which such agents acquires new words accelerates as a function of how much they already know. These observations suggest that the approach may ultimately scale to a wider range of natural language, which may bring us towards machines capable of learning language via interaction with human users in the real world.

Sidney Segalowitz

Professor (Psychology and Neuroscience) at Brock University, Canada

Sid Segalowitz is Professor (Psychology and Neuroscience) and founding Director of the multidisciplinary Centre for Lifespan Development Research at Brock University. He has served as editor-in-chief of the journal Brain and Cognition from 2002 to 2014, and has published extensively on topics in cognitive and developmental neuroscience. His research has focused on psychophysiological measures (especially ERPs) to reflect attentional control processes in normative developmental and some special populations. Recent ERP work has focused on functions of the anterior cingulate cortex and medial prefrontal cortex reflecting the cognitive-emotional interactions that occur when a person is in the midst of needing increased attention allocation, e.g., while making an unwanted erroneous response or having to inhibit a habitual response. His work has shown dramatic age-related effects, associations with personality traits related to reward sensitivity and to empathy and, more generally, relations to self-regulation. Other recent work focuses on early brain responses in affective information processing, especially to emotional faces.

Talk: The Meaning of “Meaning” as a Cortical Process

All meaningful stimuli carry cognitive information as well as sensory information, the distinction often being mapped onto the timing of ERPs. This timing can be used to address the question of when the meaning of a stimulus becomes available to the viewer. I will review work in our lab addressing this question, which has convinced us that our processing of stimulus meaning is temporally complex to the point of questioning what we mean by “meaning”. I will illustrate with our studies using meaningful stimuli portraying objects, faces and words, and sometimes interacting with the viewer’s personality traits.

Paul Warren

Professor of Linguistics at Victoria University of Wellington, New Zealand

Paul Warren is Professor of Linguistics at Victoria University of Wellington, New Zealand. He teaches and researches in psycholinguistics (Introducing Psycholinguistics, Cambridge University Press, 2012) and in phonetics, especially the description of New Zealand English and of intonation (Uptalk, Cambridge University Press, 2016). Paul is on the editorial boards of Language and Speech, the Journal of the International Phonetic Association, Laboratory Phonology, and Te Reo (the journal of the Linguistics Society of New Zealand). He is a founding member of the Association for Laboratory Phonology.

Talk: Psycho-socio-phonetics: dynamism in language processing

In this talk I will consider recent trends in psycholinguistics that remind us that language and language users are fundamentally evolving and adaptive systems. In particular, I focus on research that demonstrates that both life-long learning and recent experience shape how we process language. Since social indexicality is an important aspect of this experience, it has been increasingly recognised that our processing systems have to be sensitive to socially stratified variation. As a consequence we have witnessed greater collaboration between psycho- and sociolinguists in the area of variation and language processing. As examples of research in this area I present recent findings in the production and perception of some socially-stratified segmental and intonational features of New Zealand English. This research demonstrates listeners’ sensitivity in their interpretation of both types of feature to social characteristics associated with the speaker.

Amie Fairs

Max Planck Institute for Psycholinguistics, The Netherlands

Amie is a PhD candidate at the Max Planck Institute for Psycholinguistics in Nijmegen, using behavioural and neuroimaging methods to investigate how people multitask with language tasks. At the same time, she is an Open Science Advocate at the Max Planck Institute. She was initially trained up as an Open Access Ambassador for the Max Planck Society as a whole, but learnt more about open science and now promotes and advocates for open science practices to become more widely accepted in research. As an early career researcher, she is trying to practice what she preaches. She is additionally working with some other open science advocates to investigate what the barriers are that prevent uptake of open science practices in different places around the world.

Talk: How to make your work accessible, why it’s important (& why you should do it)

There is a big push in scientific publishing for researchers to make any papers they publish openly accessible to the public, partially driven by the fact that lots of research is paid by public funds. The traditional publishing system, with subscription-based journals, typically doesn't provide papers free for readers. However, this system is changing, with the introduction of open access journals, and authors self-archiving their work.

In this talk, I aim to give an overview of what open access is and give practical information about how you can make your papers open. I'll also discuss why making your papers open is important, both for you as an individual researcher, and for the world in general. There are pros and cons to publishing openly, and we'll talk about these.

Aside from publishing manuscripts openly, there is also another type of paper: a preprint. A preprint is a final, non-peer-reviewed version of your manuscript. There has been a recent push for more disciplines to make preprints freely accessible, by uploading them to different preprint repositories. However, there is more concern about the pros, and especially cons, of sharing a preprint. I'll talk more about what a preprint is, how to share it, and discuss why you would - and why you might not - want to share a preprint.

Publishing papers - both final versions and preprints - is just the tip of the open science iceberg. At the end of the talk, I'll introduce other areas of open science, and let people know how they can get involved. Bring your open science hats - we want a lot of discussion!