2 episodes · follows chapter 26 of the textbook
Conversation and Its Structure
What conversation is made of: turns and endpointing, speech acts, grounding, adjacency pairs, initiative and implicature, and dialogue acts tagged by a local classifier combined with an act-bigram discourse grammar through Viterbi. Then Appendix K's frame-based systems: dialogue state tracking as an explicit, tested state machine (a pure reducer with invariants and replay), confidence-tiered confirmation measured in a closed-loop simulation, and what LLM chat changed (fluent NLU/NLG) and did not (typed state, validation, confirmation, task-level evaluation).
Progress — loading…
Objectives
Language
- Describe turns, speech acts (Austin; Bach & Harnish), grounding (Clark & Schaefer), adjacency pairs, initiative, implicature
- Describe frames, slots, domain/intent, dialogue state, confirmation policies and task-oriented evaluation
Math & data structures
- Scaled likelihoods (Bayes turned around) and Viterbi over acts; interval arithmetic for turn offsets
- State machines, pure reducers, folds, invariants and replay; slot error rate and joint goal accuracy
Code
- A contextual act tagger with Viterbi from scratch, turn statistics, and a batched PyTorch port
- A frame tracker reducer with invariant checks; ablations and a confirmation-policy probe; an LLM-tracker harness
Episodes
- Properties of Human Conversation and Dialogue Acts15 min · 31 cues
Turns, adjacency pairs, grounding, initiative, implicature, dialogue acts.
- Frame-Based Dialogue Systems (Appendix K) and the LLM Era15 min · 32 cues
Slots, frames, dialogue state tracking as a state machine; what LLM chat changed and didn't.
Exercises
| Exercise | Tier | Ladder | Time |
|---|---|---|---|
| Dialogue acts in context: a local tagger, an act-bigram discourse grammar, Viterbi, and turn-taking statistics | browser | Predict · Scratch · Harden · Compose | 90 min |
| Dialogue state tracking as an explicit, tested state machine (a pure reducer with invariants) | browser | Scratch · Harden · Compose | 90 min |
| Break the tracker (no carry-over, no context), then probe 'explicit confirmation is reliable but slow' | browser | Break it · Probe | 90 min |
| Port: the contextual dialogue-act tagger in PyTorch (softmax tagger + tensor Viterbi) matching NumPy | gpu | Port · Harden | 75 min |
Labs
- Properties of Human Conversation and Dialogue Acts browser C26.01 · 70 min
- Frame-Based Dialogue Systems (Appendix K) and the LLM Era browser C26.02 · 75 min
Library contributions
Files in nlpcore this module teaches you to write.
Sources
- J. L. Austin, How to Do Things with Words (William James Lectures, Harvard, 1955), Oxford 1962
- K. Bach, R. M. Harnish, Linguistic Communication and Speech Acts, MIT Press, 1979
- D. G. Bobrow, R. M. Kaplan, M. Kay, D. A. Norman, H. Thompson, T. Winograd, GUS, a Frame-Driven Dialog System, Artificial Intelligence 8(2):155-173, 1977 · link
- P. Budzianowski et al., MultiWOZ - A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling, EMNLP 2018 · arXiv:1810.00278
- H. H. Clark, E. F. Schaefer, Contributing to Discourse, Cognitive Science 13:259-294, 1989 · link
- H. P. Grice, Logic and Conversation, in P. Cole & J. L. Morgan (eds.), Syntax and Semantics 3: Speech Acts, Academic Press, 1975, pp. 41-58
- J. Kun, A Programmer's Introduction to Mathematics
- N. Nayak, D. Hakkani-Tur, M. Walker, L. Heck, To Plan or not to Plan? Discourse Planning in Slot-Value Informed Sequence to Sequence Models for Language Generation, Interspeech 2017, pp. 3339-3343 · link
- H. Sacks, E. A. Schegloff, G. Jefferson, A Simplest Systematics for the Organization of Turn-Taking for Conversation, Language 50:696-735, 1974
- J. R. Searle, Speech Acts: An Essay in the Philosophy of Language, Cambridge University Press, 1969
- Jurafsky & Martin, SLP3 draft, Appendix K: Frame-Based Dialogue Systems · link · §K.0.1, K.0.2, K.1.1, K.1.2, K.1.3, K.1.4, Historical Notes
- Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft (Aug 19, 2026), Ch. 26 Conversation and its Structure · link · §26.1.1, 26.1.2, 26.1.3, 26.1.4, 26.1.5, 26.1.6, 26.2
- T. Stivers et al., Universals and cultural variation in turn-taking in conversation, PNAS 106(26):10587-10592, 2009
- A. Stolcke et al., Dialogue Act Modeling for Automatic Tagging and Recognition of Conversational Speech, Computational Linguistics 26(3), 2000 · arXiv:cs/0006023
- J. Weizenbaum, ELIZA - A Computer Program for the Study of Natural Language Communication Between Man and Machine, Communications of the ACM 9(1), 1966 · link
- J. D. Williams, A. Raux, M. Henderson, The Dialog State Tracking Challenge Series: A Review, Dialogue & Discourse 7, 2016