Retraces and Repetitions

Status: Current Last updated: 2026-10-20 (commit f7e32e22)

Retraces mark content that the speaker said but then corrected, repeated, or abandoned. They are one of the most consequential constructs in CHAT because they affect how every dependent tier aligns to the main tier.

CHAT Syntax

A retrace has two parts: the retraced content (what the speaker said first) and the correction (what follows). The retraced content is marked with a trailing bracket code:

The CHAT manual supplies these surface meanings. Chatter’s historical Rust variant names differ: Multiple represents [///], and Reformulation represents [/-]. The variant name is not a linguistic classification.

Single-Word Retraces

When only one word is retraced, no angle brackets are needed:

*CHI: I [/] I want that .
*CHI: ana [//] an .
*MOT: the book [/-] the magazine is here .

Group Retraces

When multiple words are retraced, angle brackets delimit the scope:

*MOT: <the dog> [//] the cat ran .
*CHI: <I want> [/] I need cookie .
*CHI: <I want the> [///] give me that .

Retraces with Replacements

A retraced word often has a replacement [: target] and/or error code [* code]. This is common in aphasia and child language corpora where the speaker produces an incorrect form:

*PAR: tika@u [: kitty] [* p:n] [//] kitty is nice .
%mor: noun|kitty aux|be-Fin-Ind-Pres-S3 adj|nice-S1 .

*PAR: lɛɾɪ@u [: later] [* p:n] [//] later in the day .
%mor: adv|late adp|in det|the-Def-Art noun|day .

*CHI: male [: female] [* s:r] [/] male [: female] [* s:r] .
%mor: adj|female-S1 .

In each case, the retraced word (before the [//] or [/]) is excluded from %mor alignment. Only the correction (after the marker) is counted.

Data Model

Retraces are a first-class variant of UtteranceContent:

flowchart TD
    UC["UtteranceContent"]
    UC --> Word
    UC --> RW["ReplacedWord"]
    UC --> Retrace
    UC --> AG["AnnotatedGroup"]
    UC --> Other["Other content variants"]

    Retrace --> BC["BracketedContent"]
    Retrace --> RK["RetraceKind"]
    BC --> BIW["BracketedItem::Word"]
    BC --> BIRW["BracketedItem::ReplacedWord"]

    style Retrace fill:#f96,stroke:#333

The Retrace struct wraps the retraced content in a BracketedContent container, which can hold any combination of words, replaced words, and other content items:

The source definition is model/content/retrace.rs in talkbank-model. Besides content and kind, is_group preserves angle-bracket spelling, span locates the construct, and optional marker_span locates the marker itself when the parser retained that evidence. Source-free values must not invent marker offsets.

Where the annotations live

There is deliberately no annotations field. A content item followed by a run of scoped markers is a LEFT-ASSOCIATIVE CHAIN: each marker scopes over everything to its left, so these two lines are different claims about the same two words.

dog [* p:w] [/] dog     the error is on the abandoned attempt
dog [/] [* p:w] dog     the error is on the retrace

Annotations written BEFORE the marker therefore annotate the retraced material and live inside content; annotations written AFTER it annotate the retrace and live on an AnnotatedRetrace(Box<Annotated<Retrace>>) wrapper, exactly parallel to Group / AnnotatedGroup.

A single flat annotations field holding both would make the two indistinguishable: dog [* p:w] [/] would be silently written back as dog [/] [* p:w], and a second adjacent marker would overwrite the first, so на [//] [/] на would become на [/] на. Neither would be visible to validate, to --roundtrip (which tests idempotence of serialize(parse(x)), not fidelity to the input) or to SemanticEq (the two orderings would be the same model).

Why First-Class?

Representing retraces as annotations on words or groups would mean every match on content had to inspect annotation lists to determine whether a word was retraced, a class of bugs where retraced content is accidentally included in alignment counting, word extraction, or retokenization.

Making Retrace a top-level UtteranceContent variant means:

1. Exhaustive matches expose unhandled variants. Catch-all arms can hide a newly added variant, so first-class representation alone is not proof that every consumer handles it correctly. Shared structural views and scoped wildcard-match lints help keep that handling explicit.
2. Domain-aware gating is centralized. The content walker checks the Retrace variant once, not at every annotation-inspection site.
3. Alignment counting is simple. The count function returns 0 for Retrace in Mor domain, no annotation inspection needed.

Parser Conversion

The tree-sitter grammar parses retrace markers ([/], [//], etc.) as annotations on word_with_optional_annotations. The Rust parser converts them to structural Retrace nodes in parse_word_content():

flowchart LR
    subgraph "Tree-sitter CST"
        WOA["word_with_optional_annotations"]
        SW["standalone_word"]
        BA["base_annotations"]
        RP["retrace_partial / retrace_complete / ..."]
        WOA --> SW
        WOA --> BA
        BA --> RP
    end

    subgraph "Rust Model"
        RET["UtteranceContent::Retrace"]
        BC2["BracketedContent"]
        W2["Word or ReplacedWord"]
        RET --> BC2
        BC2 --> W2
    end

    WOA -->|"parse_word_content()\n(word.rs)"| RET

parse_word_content() first builds the word or replaced-word core, then calls fold_marker_chain on the ordered markers. That fold preserves the distinction between annotations before and after a retrace, including adjacent retraces. parse_group_content() in group/parser.rs uses the same fold with a group core. These are the primary parser’s conversion owners, not a claim about every backend’s implementation.

Alignment Behavior

Retraces interact differently with each dependent tier domain:

flowchart TD
    RT["Retrace node\n(e.g. 'tika@u [: kitty] [* p:n] [//]')"]

    RT -->|"Mor domain"| SKIP["SKIP\n(return 0)\nNot morphologically analyzed"]
    RT -->|"Pho domain"| COUNT["COUNT\nPhonologically produced"]
    RT -->|"Sin domain"| COUNT2["COUNT\nGesturally produced"]
    RT -->|"Wor domain"| COUNT3["RECURSE\napply FilteredLexicalV1 leaf rule"]

    style SKIP fill:#faa,stroke:#333
    style COUNT fill:#afa,stroke:#333
    style COUNT2 fill:#afa,stroke:#333
    style COUNT3 fill:#afa,stroke:#333

Why %mor skips retraces: The %mor tier represents the morphological analysis of what the speaker meant to say. Retraced content is a false start or error; it was produced phonologically but is not part of the intended linguistic structure. The correction after the retrace marker carries the morphological analysis.

Why traversal enters retraces outside %mor: Retraced material was produced, so %pho, %sin and %wor can select positions within it. Entering a container does not mean every leaf participates in every domain. In particular, %wor selects only the current timing policy’s eligible words.

For %wor, retrace ancestry does not change leaf-level membership:

• eligible regular words and fillers count both inside and outside retraces;
• phonological fragments (&+), nonwords (&~), untranscribed markers (xxx, yyy, www) and omissions are excluded in either location;
• overlap annotations do not change this leaf-level policy.

For example, the retraced one below remains eligible, but &+ss does not:

*CHI:	<one &+ss> [/] one play ground .

The selected lexical sequence is one, one, play, ground. This is membership, not evidence of available or acoustically correct timings. A parsed legacy %wor tier can use a different convention; do not delete its words or declare the CHAT invalid just to force a count match. See %wor timing semantics for binding, lexical corroboration and timing admission.

Implementation

Both the counter and the walker ask one owner, alignment/helpers/descent.rs, what a domain does with a retrace: %mor excludes it, every other domain enters it. The counter takes a PositionalDomain and converts it for the descent rule; the walker takes the TierDomain directly.

Counting and extraction: walk_alignable_item() in alignment/helpers/count.rs, one walk whose sink either counts or collects:

UtteranceContent::Retrace(_) | UtteranceContent::AnnotatedRetrace(_) | /* other containers */ => {
    match descend(item.structure(), Some(domain.into())) {
        Descent::Into(entered) => walk_alignable_bracketed(entered.content(), domain, sink),
        Descent::Atomic(unit) => sink(AlignablePosition::Atomic(unit)),
        Descent::Excluded => {} // %mor: a retrace is not aligned
    }
}

Walking: walk_words() in alignment/helpers/walk/mod.rs:

UtteranceContent::Retrace(_) | UtteranceContent::AnnotatedRetrace(_) | /* other containers */ => {
    if let Some(into) = descend(item.structure(), domain).entered() {
        walk_bracketed_content(&into.content().content, domain, f);
    }
}

%wor generation and overlap counting still use dedicated recursive helpers, but only for %wor-specific sequencing details like replacement handling, not for retrace-sensitive membership.

Validation

The three retrace rules

validation/retrace/ separates node visitation (visit.rs) from the serialized-order leaf/checkpoint collection (collection/) needed to inspect what follows a marker. Both use ContentStructure; the visitor reaches nested retraces rather than treating an outer retrace as a leaf.

E377 and E378 are disjoint despite the neighbouring names. In a [//] [/] a the inner retrace still holds a word, so E378 stays silent; in <&=sigh> [/] &=sigh there is no second marker, so E377 stays silent. The repairs differ too: drop a marker for E377, retrace the words rather than the vocalization for E378.

E378 tests for absent words, not for a present event: an event alongside words does not by itself invalidate a retrace. Words can occur deeper inside an annotated group or quotation (<<the dog> [?]> [/] the dog), so testing only immediate children would incorrectly reject such content. Untranscribed material counts as words on purpose: xxx, yyy and www lower as words, so retracing speech nobody could make out stays valid.

Which variants are containers is owned in one place for these rules, model::content::structure::ContentStructure. It has one owner because two hand-written copies of that knowledge can disagree, as about PhoGroup and SinGroup, which silently stops E377 firing inside ‹...› with no test able to see it.

Alignment walkers also use ContentStructure, through the domain policy in alignment/helpers/descent.rs. Structural classification answers what an item contains; domain policy separately decides whether to enter it, count it as an atomic position, or exclude it.

Alignment Validation (E705)

Correctly represented retraces are excluded from the %mor count. This prevents retraced words from causing a false E705 mismatch; it does not guarantee that the remaining main-tier and %mor positions agree.

Regression Tests

Source diagnostic cases live in crates/talkbank-parser-tests/tests/integration/retrace_diagnostics_from_source.rs; the reference example is corpus/reference/annotation/retrace.cha. The error specifications for E370, E377 and E378 own the diagnostic contracts. These are evidence entry points, not a claim of complete retrace coverage.

See Also

• Alignment Architecture: full alignment system docs
• The %mor Tier: morphological tier format and alignment rules
• CHAT Manual: Retracing

This page last changed: 2026-10-20 (commit f7e32e22). The whole book last changed: 2026-10-27 (commit 5bfee2f9).