• seminar

seminar

SEMINAR

CLT seminar planning meeting for the spring semester

Date: 2010-01-14 10:15 - 12:00

Location: L307

Permalink

SEMINAR

 

 

Linguists have been dreaming of the creation of a meaning-based grammar for generations. The first scholar to express the need for such a linguistic description was apparently Georg von der Gabelentz (1891). He drew a distinction between a synthetic and an analytic system of a language. He connected the synthetic system (das synthetische System) with grammatical synonymy: the speaker who wants to express a thought or a feeling has different possibilities for doing so. More or less the same aspiration was later expressed by dozens of linguists (Jespersen, Bernštein, Ščerba, Bondarko, Zolotova, Mel’čuk, Halliday, Dik van Valin etc.) My own model of functional syntax (FS) is based on the following main principles:1. FS consequently follows the principle of going “from meaning to form”; to be more specific, “from semantic categories to linguistic means”. So, the surface structures John is cold, John has a hangover, John is in a coma, John is freezing are all dealt with in the same chapter entitled Physiological state because they have a similar semantic structure (a Physiogical state with an actant, an Experiencer). 2. In the communicative situation, FS reflects the point of view of the speaker who is searching for a suitable way to express his/her thoughts. However, FS does not aim to describe the mental processes taking place in the speaker’s head. 3. The linguistic description in FS is based on semantic structures, which reflect the state of affairs and the speaker’s comments on it. 4. The states of affairs are situations or fragments of reality (the real, virtual, or inner world) as they are interpreted by the speaker.

... Full abstract in http://tournesol.cs.chalmers.se/~aarne/talks/mustajoki.docMore information:http://www.helsinki.fi/~mustajok/pdf/Functional_Syntax_RLJ.pdf

 

Date: 2009-11-19 10:15

Location: Department of Swedish Language, Lennart Torstenssonsgatan 8, room L307

Permalink

SEMINAR

Mikael Roll (Lund) kommer att tala om ERP-studier av svensk ordföljd, prosodi och morfologi. Han disputerade nyligen på avhandlingen The Neurophysiology of Grammatical Constraints.

Genom att titta på förändringar i hjärnaktivitet i realtid har Mikael Roll undersökt hur lyssnare använder prosodiska och pragmatiska signaler för att förutsäga syntaktisk och morfologisk struktur.
En intonationsstigning i början av att-satser ökar t.ex. förväntningen på huvudsatsordföljd. I avsaknad av en stigning ser man en s.k. P600-effekt i att-satser med huvudsatsordföljd. P600 visar ombearbetning av den oväntade strukturen. Liknande P600-effekter ger ändelsen i ett ord där accent 2 bytts ut mot accent 1, d.v.s. om t.ex. ordet 'fiskar' har den ordaccent som 'fisken' brukar ha. Detta tyder på att ordaccenten används för att för att förutsäga ordstrukturen.
När man hör ett objekt som 'glass/glassen/den' i en sats som 'Han åt glass/glassen/den...', verkar både dess form och innehåll skapa olika typer av förväntning på kommande ord, vilket ger upphov till olika effekter för olika sorters objekt i meningar av typen 'Han åt glass/glassen/den inte'.

http://www.sol.lu.se/lingvistik/person/MikaelRoll

Date: 2009-12-07 15:15 - 17:00

Location: L308

Permalink

SEMINAR

Speaker: Aarne Ranta

The leading idea of GF (Grammatical Framework) is the division between abstract and concrete syntax. The corresponding levels of representation are abstract syntax trees and strings. But other linguistic formalisms often work on levels that are between these two: phrase structure trees, dependency trees, and word alignments. In the talk, we will show how these intermediate levels can be derived from GF and used in the visualization of grammars. We will also discuss some tentative ideas about their use in robust parsing (with Nivre's MaltParser) and statistical machine translation (with tools such as Moses).

Date: 2009-10-29 10:15 - 11:15

Location: Department of Swedish Language, Lennart Torstenssonsgatan 8, L307

Permalink

SEMINAR

Presenter:     Richard Sproat, Oregon Health & Science University

Of all artifacts left behind by ancient cultures, few are as evocative as symbols. When one looks at an inscription on stone or clay, it is natural to ask: What does it mean? Was this a form of writing, or some sort of non-linguistic system? If it was writing, can we hope to decipher it?

In this talk, I examine these and related questions, and the possible role of statistical methods in answering them. I start with a brief review of the history of successful decipherment, and computational approaches to decipherment.  In the bulk of the talk, I discuss a topic that made its rounds of the popular science press earlier this year: Namely, the claimed "entropic evidence" that the 4000-year-old Indus Valley symbol system constituted a script. I present simple counterevidence to the usefulness of the proposed measure to support any such claim; and I review the large amount of archaeological and comparative cultural evidence against the script hypothesis, which must in any case be taken into account in any complete discussion of this subject.

(Portions of this talk are based on joint work with Steve Farmer and Michael Witzel.)

Date: 2009-12-10 10:15 - 12:00

Location: Department of Swedish Language, Lennart Torstenssonsgatan 8, room L307

Permalink

SEMINAR

Kimmo Koskenniemi from Helsinki is visiting us and will give this talk

Date: 2009-12-10 15:06 - 17:00

Location: EDIT Room, 3364, Chalmers,

Permalink

SEMINAR

I will present the intitial results of the experiments that will be 
reported in my thesis somtime next year. The MedEval test collection has
been constructed at Språkdata, GU, during the last few years. It is an
information retrieval test collection with documents from the medical
domain. The test collection allows the user to state user group, doctor
or patient, or none of these. The possibility to state user group is a
unique feature for the MedEval test collection.

In the initial experiments I have studied search terms with different
features to see what features give good results, and which features do
not. The features studied are for example, term broadness, domain and
doctor-patient language. I have also looked at compounds used as search
keys and the compound constituents. Even though splitting compounds and
using the constituents as search keys often is an effective strategy, it
is not alway the case. I will demonstrate examples of when decomposition
works, and when it does not.


Welcome!

Karin Friberg Heppin

Date: 2009-10-15 10:00 - 12:00

Location: L307

Permalink

SEMINAR

Frames in formal semantics - Montague's temperature and Fernando's event strings Robin Cooper

Abstract

More information about this event…

Date: 2009-10-27 13:15 - 15:00

Location: T340, Gamla Hovrätten

Permalink

SEMINAR

The Interplay of Grammatical Theory and Language Engineering: The Evolution of HPSG Ivan A. Sag (Stanford University)

In this talk I'll provide one person's perspective on the evolution of
HPSG from its GPSG origins to the construction-based incarnation
(Sign-Based Construction Grammar) that blends in aspects of Berkeley
Construction Grammar. Since its inception, the development of HPSG
theory has been influenced by on-going attempts to make it
computationally real. Language engineering, where matters of detail
and scale, as well as the demand for analytic precision, are of
paramount concern, has had an overall positive effect on the
development of HPSG; but it has had negative consequences as well.  In
the process of telling this story, I'll also discuss some current work
in SBCG.

home page

Date: 2009-09-23 15:15 - 17:15

Location: L308

Permalink

SEMINAR

A demonstration of a speech-controlled LEGO robot, built by students of the Language Technology Programme. NOTE the unusual location, the computer lab (T225) in the FLOV department!

During the Project Course at the Language Technology Programme, a group of students constructed and implemented a speech-controlled robot, called Ruben the Robot. Ruben is built in LEGO Mindstorms NXT, connected via Bluetooth to a computer with speech synthesis and speech recognition. The dialogue system, synthesis, recognition, and LEGO controller, are all implemented in Python. 

The seminar will be presented by a non-empty subset of the project members {Kim Demaret, Jonatan Uppström, Håkan Edqvist, Martin Hammarstedt}. 

Ruben also has a Swedish webpage: http://www.cling.gu.se/~st7mhamm/ruben/

Date: 2009-10-01 10:15 - 12:00

Location: T225 (The computer lab at Hovrätten)

Permalink

X
Loading