Schedule: seminar on word meaning in context
This schedule is subject to change.
Week 1
Jan 19 Introduction and course overview
Part 1: Phenomena
Jan 21 Tests for sense distinctions, and their problems.
Please choose one of two readings:
Cruse, Polysemy and related phenomena from a cognitive linguistic viewpoint -- available on Canvas
Week 2
Jan 26 Two kinds of sense extensions: regular polysemy, and radial categories.
Please choose one of two readings:
Regular polysemy: Johannes Dölling, Systematic polysemy, sections 1-4 (through page 10)
Radial networks in cognitive semantics: Geeraerts, Theories of lexical semantics, Section 5.1.2, pages 192-199
Jan 28 Polysemy -- an overview, including: sense distinction tests; regular polysemy; the core meaning hypothesis and the underspecification hypothesis; thin versus rich semantics
Falcum and Vicente: Polysemy: current perspectives and approaches
Week 3
Feb 2 The view from annotation projects: problems with sense annotation, and alternative meaning characterizations
Reading: Erk et al, Investigations on word senses and word usages]
I will also give a short overview of relevance theory and what it says about word meaning
Feb 4 Background: Concepts as prototypes or exemplars, and their connection to word meaning.
Reading: Greg Murphy, Big Book of Concepts, chapter 3 through page 60, available on Canvas.
If you are curious to hear why the "classical approach" was rejected, part of chapter 2 of the Big Book of Concepts is also on Canvas
Week 4
Feb 9 Distinct senses in the brain?
We are reading: Klepousniotou, Titone and Romero, Making Sense of Word Senses: The Comprehension of Polysemy Dependson Sense Overlap
Feb 11 Concept combination
We are reading James Hampton, Compositionality and Concepts. This is chapter 4 in Hampton/Winter, Compositionality and concepts in linguistics and psychology
Week 5
Feb 16 Class canceled due to inclement weather
Feb 18 Class canceled due to inclement weather
Week 6
Part 2: Representations
Feb 23 Class canceled due to inclement weather
Feb 25 Generalized event knowledge
Reading: McRae and Matsuki, People use their knowledge of common events to understand language, and do so as quickly as possible. Available on Canvas
Additional reading, if you are curious: Here is the Elman paper that I mentioned in class:
Elman, On the meaning of words and dinosaur bones: Lexical knowledge without a lexicon.
Week 7
Mar 2 Compositionality and meaning in context
Reading: Pelletier, Compositionality and Concepts- a perspective from formal semantics and philosophy of language, chapter 3 in Hampton/Winter, Compositionality and concepts in linguistics and psychology
Mar 4 Background: type theory and proof theory. (lecture)
Additional readings (not required, but good if you want to know more):
On proof theory and model theory and what thoughts about meaning they insprire:
Szabolcsi, Questions about Proof Theory vis-à-vis Natural Language Semantics
Week 8
Mar 9 Asher's type-compositional semantics.
Reading: Nicholas Asher, Selectional restrictions, types and categories, Journal of Applied Logic 2014. Available on Canvas. We read pages 74-83, ending with the paragraph "So to conclude".
Due: Initial project descriptions, around 2 pages.
Mar 11 Zeevat et al: Representing the Lexicon. 2017, In: Bridging Formal and Conceptual Semantics. Selected papers of BRIDGE-14.
Week 9
Spring break
Week 10
Mar 23 Background: machine learning and classification (lecture).
Mar 25 Larsson: perception as classification.
Reading: Larsson, "Formal semantics for perceptual classification", Journal of Logic and Computation 2015.Click on "PDF" to get the full paper. We read sections 1, 2, 3.1, and 4.
Week 11
Mar 30: Gärdenfors' conceptual spaces
April 1: Background: distributional semantics a.k.a. semantic spaces a.k.a. word embeddings in computational linguistics (lecture)
Extra pointers, based on our discussion:
the Context Atlas: visualizations of word-in-context embeddings
Week 12
Apr 6 Semantic spaces in cognition
Reading: Kintsch, Predication
Due: Project report draft
Apr 8 Baroni and Sadrzadeh: Compositional distributional semantics
Reading: Baroni and Zamparelli, Nouns are vectors, adjectives are matrices: representing adjective-noun constructions in semantic space
Additional papers I will refer to:
Datasets of phrase similarity, and simple phrase representations:
Mitchell and Lapata, first paper, second paper, Grefenstette and Sadrzadeh
Vectors and matrices for other parts of speech: Grefenstette and Sadrzadeh again, Sadrzadeh on relative clauses
Paperno et al: A simplified approach that avoids high-order tensors
Week 13
Apr 13 Background: Bayesian generative models in computational linguistics, in semantics and pragmatics (lecture)
No assigned reading, but if you would like to read more, check out section 2 of the "Probabilistic Trend in Semantics and Pragmatics" draft on Canvas
Apr 15 Situtation description systems (lecture)
No assigned reading, but this is based on Erk and Herbelot (draft): "How to marry a star: probabilistic constraints for meaning in context". I'll put a current draft on Canvas.
Week 14
Part 3: Challenges
Apr 20 Can the meaning of a word vary arbitrarily?
Reading: Recanati, Literal Meaning, chapter 9 -- available on Canvas. We'll focus on sections 9.1 through 9.3 and 9.7.
Apr 22 The immediacy of world knowledge
Reading: Hagoort and Van Berkum, Beyond the Sentence Given. We focus on sections 2 and 4, with 6 summing everything up.
Week 15
Apr 27 What kind of meaning should we try to characterize?
Reading: Westera and Boleda, Don't blame distributional semantics if it can't do entailment
This paper touches on several themes we have discussed in the semester: whether it is possible to distinguish different senses of a word; what distributional models of meaning say at the lexical level, and whether they can be helpful at the level of larger expressions. It also touches on the problem of reference, which we haven't discussed a lot so far.
Apr 29 Conventions in a community of speakers, and how they affect the lexicon
Reading: Noble and Fernandez, Semantic Approximation And Its Effect On The Development Of Lexical Conventions
This paper uses simulations to explore semantic approximation, where meanings of words are extended to cover new situations, and how semantic approximation plays out in a community of speakers
Note: Here is a slideset that explains Erev/Roth reinforcement learning
Week 16
May 4: Discussion of course projects:
Gabriella Chronis, Venkat Govindarajan
May 6: Discussion of course projects
Ellen Jones, Yejin Cho, Samuel Cantor
Final project report due: Monday May 17, end of day