The computational linguistics group at Uppsala University is delighted to invite you to an afternoon of public lectures on language technology and cognition by leading experts in the field. The lectures will take place in Room 7-0042, English Park Campus, Uppsala University, on the 9th of March according to the schedule below. Attendance is free for anyone interested.
| 13.15-14.00 | John Hale Department of Linguistics Cornell University |
Entropy Reduction and Asian Languages This talk presents a particular conceptualization of human language understanding as information processing. From this viewpoint, understanding a sentence word-by-word is a kind of incomplete perception problem in which comprehenders over time become more certain about the linguistic structure of the utterance they are trying to understand. The Entropy Reduction hypothesis holds that the scale of these certainty-increases reflects psychological effort. This claim revives the application of information theory to psycholinguistics, which languished since the 1950s. But in contrast to that earlier work, modern applications of information theory to language-understanding now use generative grammars to specify the relevant structures and their probabilities. This representation makes it possible to apply standard techniques from computational linguistics to work out "expectations" in Vasishth's sense. The talk exemplifies the general theory using examples from some subset of Korean, Chinese and Japanese. These languages diverge from Romance and Germanic in ways that help stress-test a general cognitive theory of sentence processing. |
| 14.00-14.45 | Shravan Vasishth Department of Linguistics University of Potsdam |
Expectation Versus Retrieval Cost: An Empirical Investigation In human sentence processing, expectations about upcoming material can help or hinder parsing, and completing dependencies is often costly due to factors like increased interference. Both factors probably play a role in sentence processing, but how exactly does this relationship play out? Using data from Chinese relative clauses (self-paced reading), German (eyetracking, self-paced reading, event-related potentials, and eye-tracking and ERP co-registration), and English (eyetracking), I argue that expectation may play a relatively early role in determining difficulty, and that retrieval-based costs might become evident in later processing steps. That is, when the effect due to expectations vs retrieval occurs may depend on how far we are into processing a word in a sentence. This has interesting implications for the way the eye-parser connection should be implemented in models of eye-movement control in reading, and suggests one way that a two-factor model incorporating expectation+retrieval cost could be spelt out. |
| 14.45-15.15 | Break | |
| 15.15-16.00 | Walter Daelemans CLiPS Research Center University of Antwerp |
Computational Stylometry Some of the language variation in texts can be linked to psychological and sociological properties of their authors (personality, mental health, age, gender, education level, region, ...). Assuming that this link is robust and invariant, techniques can be developed to determine these psychological and sociological variables on the basis of text. This "computational stylometry" (or text profiling) gives rises to interesting applications such as gender and age detection from text, prediction of onset of Alzheimer's disease from writing, forensic applications, demographic marketing etc. After providing an overview of achievements and issues in text profiling, I will focus on a number of problematic cases: computational stylometry in applications with small amounts of training data, difficult text types (e.g. social media language), and finally on the question whether author (group) characteristics can be distinguished from other factors leading to linguistic variation: topic, register, and genre. |