What six problems must every conversation solve?
Schegloff's Conversation Analysis identifies six universal organizational challenges that speakers navigate in all talk-in-interaction. Understanding these helps explain why current AI dialogue systems fall short of human fluency.
Schegloff's Conversation Analysis identifies six "generic orders of organization" — problems that every conversation must solve for orderly interaction to proceed:
Turn-taking — who should talk next and when? How does this affect the construction and understanding of turns themselves?
Action-formation — how are the resources of language, body, environment, and position fashioned into recognizable actions (requesting, inviting, complaining, agreeing, rejecting) in a class of unknown size?
Sequence-organization — how are successive turns formed to be "coherent" with prior turns, and what is the nature of that coherence?
Trouble-handling — how to deal with problems in speaking, hearing, and understanding so that interaction doesn't freeze, intersubjectivity is maintained or restored, and the sequence can progress?
Word-selection — how are the components of a turn selected, and how does that selection shape understanding by recipients?
Overall structural organization — how does the overall composition of an interaction get structured, and how does placement inform the construction and understanding of talk?
These are not theoretical constructs — they have empirical support and appear to be language-universal. Properties of sequence organization also generalize to text-based chats, though digital interaction may follow slightly different patterns.
The practical relevance for AI: current dialogue systems explicitly address only turn-taking (who responds) and action-formation (intent classification). Sequence-organization is partially addressed through context windows. Trouble-handling (repair) is almost entirely absent — since Do language models actually build shared understanding in conversation?, models skip the repair sequences that humans use to maintain intersubjectivity. Word-selection receives some attention through style control. Overall structural organization is ignored.
This means current AI dialogue addresses roughly 2 of 6 fundamental conversational requirements. The other 4 represent a design space that remains largely unexplored.
Since What three layers must discourse systems actually track?, Schegloff's six orders provide the conversational-level complement to Grosz & Sidner's discourse-level framework. The two are not competing — they describe different levels of the same phenomenon.
Related concepts in this collection 9
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
What three layers must discourse systems actually track?
Grosz and Sidner's 1986 framework proposes that discourse requires simultaneously tracking linguistic segments, speaker purposes, and salient objects. Understanding why all three are necessary helps explain where current AI systems structurally fail.
discourse-level complement to Schegloff's conversational-level framework
-
Do language models actually build shared understanding in conversation?
When LLMs respond fluently to prompts, do they perform the communicative work humans do to establish mutual understanding? Research suggests they skip the grounding acts that make dialogue reliable.
repair (trouble-handling) is the specific mechanism for maintaining intersubjectivity that LLMs skip
-
Why do language models sound fluent without grounding?
Explores whether LLM fluency masks the absence of communicative work—the clarifying questions, acknowledgments, and understanding checks that humans perform. Why does skipping these acts make models sound more confident?
grounding acts span multiple of Schegloff's six orders
-
Why do language models skip the calibration step?
Current LLMs assume shared understanding rather than building it through dialogue. This explores why that design choice persists and what breaks when it fails.
dynamic grounding addresses trouble-handling specifically
-
Can AI systems detect and correct misunderstandings after responding?
How do conversational systems recognize when their previous response was based on a misunderstanding, and what mechanism allows them to correct it retroactively rather than restart?
TPR is a specific instantiation of the trouble-handling generic order: reactive correction at T3 after T2 reveals misunderstanding
-
What semantic failures break dialogue coherence most realistically?
Can we distinguish distinct types of incoherence by manipulating semantic structure rather than surface text? This matters because text-level evaluations miss the semantic failures that actually occur in dialogue systems.
DEAM's semantic-level failure modes map to specific generic orders: contradiction/coreference → trouble-handling, decreased engagement → action-formation, irrelevancy → sequence-organization
-
Why do dialogue systems lose context when topics return?
Stack-based dialogue management removes topics after they're resolved, making it hard for systems to reference them later. Does this structural rigidity explain why conversational AI struggles with topic revisitation?
topic management is a specific instantiation of the "overall structural organization" generic order; the stack-vs-attention debate addresses how this order should be solved architecturally
-
Can models learn when NOT to speak in conversations?
Does training AI to explicitly predict silence—through a dedicated silent token—help models understand when intervention adds value versus when they should stay quiet? This matters for building conversational agents that feel naturally helpful rather than intrusive.
DiscussLLM directly addresses the turn-taking generic order: the silent token mechanism trains models to solve the "who should talk next and when" problem, which Schegloff identifies as the first organizational requirement of conversation
-
When should AI agents ask users instead of just searching?
Explores whether tool-enabled LLMs should probe users for clarification when uncertain, rather than silently chaining tool calls that drift from intent. Examines conversation analysis patterns as a formal alternative.
insert-expansions instantiate sequence-organization and trouble-handling simultaneously: they create embedded pair structures within the main sequence (sequence-organization) to repair potential misunderstanding before it compounds (trouble-handling)
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- Sequence Organization in Interaction: A Primer in Conversation Analysis
- No that's not what I meant: Handling Third Position Repair in Conversational Question Answering
- Intent Mismatch Causes LLMs to Get Lost in Multi-Turn Conversation
- Conversational DNA: A New Visual Language for Understanding Dialogue Structure in Human and AI
- Probing the Multi-turn Planning Capabilities of LLMs via 20 Question Games
- Task-Oriented Dialogue with In-Context Learning
- Conversational Alignment with Artificial Intelligence in Context
- MultiChallenge: A Realistic Multi-Turn Conversation Evaluation Benchmark Challenging to Frontier LLMs
Original note title
sequence organization in talk-in-interaction has six generic orders that all conversation must solve — turn-taking action-formation sequence-organization trouble-handling word-selection and overall structure