Lang Theory Building

---
name: lang-theory-building
description: Use when turning a linguistic pattern into a theoretically grounded claim for a Language (LSA) manuscript — an analysis with explicit assumptions, observable predictions, and engagement across competing frameworks rather than inside one formalism. Structures the argument; it does not gloss data or run the statistics.
---

# Theory & Analysis Building (lang-theory-building)

At *Language* the analysis is the contribution. A glossed pattern, a corpus trend, or an experimental
effect is not a *Language* paper until it is attached to an **explicit account** — what the data are a
*case of*, what the analysis predicts, and how it fares against rival analyses. The journal's defining
demand is **theoretical grounding without framework parochialism**: state your assumptions, but engage
the alternatives a reader from another tradition would raise.

## When to trigger

- The data are solid but the "so what for linguistic theory" is thin
- A reader called the paper "descriptive," "stipulative," or "framework-internal"
- You need to make the analysis's predictions explicit and testable
- Fieldwork or corpus work whose theoretical claim is implicit and needs surfacing

## Build the analysis (by mode of work)

### Formal (phonology / syntax / semantics / morphology)
1. **Generalization** — state the empirical generalization crisply, in theory-neutral terms first, so
   any reader can see the pattern before any formalism.
2. **Analysis** — give the account (e.g., a constraint ranking, a feature geometry, a derivation, a
   type-driven composition); make each assumption explicit and non-vacuous.
3. **Predictions** — what should and should *not* occur if the analysis is right; these become the
   probes in `lang-research-design`.
4. **Adjudication** — show where a leading rival framework makes a different prediction, and which the
   data support.

### Variationist / sociolinguistic
- Name the **linguistic and social constraints** and the mechanism of change or variation; connect the
  variable to a general model, not just a local correlation.

### Historical / typological
- State the **generalization across languages** and the diachronic or structural mechanism behind it;
  say what the sample is a sample *of*.

### Psycholinguistic / computational
- Tie the model or experiment to a **claim about grammar or processing**; distinguish what the result
  shows about the representation from what it shows about the task or architecture.

## The cross-framework test (Language-specific)

Ask: *Would a linguist working in a different framework accept the generalization and see why my
analysis is a contender?* If the argument only makes sense once your formalism is granted, you have a
framework-internal exercise, not a *Language* contribution. Restate the generalization framework-neutrally
and let the analysis compete on predictions.

## The analytic-depth ladder (calibration, hedged)

An orienting heuristic, not an editorial rubric; confirm expectations against the current author pages.

| Rung | What the paper offers | Likely Language verdict |
|------|-----------------------|-------------------------|
| 0 | Glossed data, no analysis | desk-returned as description |
| 1 | Data + an analysis in one framework | promising but parochial |
| 2 | Analysis with explicit predictions, tested | the modal strong submission |
| 3 | Predictions that adjudicate between rival frameworks | the Language target |

Aim for rung 2 minimum; rung 3 wins. This premium on adjudication distinguishes *Language* from a
subfield venue that accepts a clean derivation inside the house framework.

## Engaging the tradition (a Language expectation)

Reviewers expect the analysis located against foundational work — the relevant classic account
(structuralist, generative, functionalist, Optimality-theoretic, usage-based) used, not name-dropped.
If your account revises a standard analysis, say exactly what it keeps and what it overturns.

## Referee-pushback patterns and the venue-specific fix

| Referee writes… | The Language-specific fix |
|-----------------|---------------------------|
| "Descriptively adequate, theoretically thin." | climb to rung 2–3; state the general claim + predictions |
| "Stipulative / analysis does no work." | show the analysis rules out attested-but-unwanted patterns |
| "Ignores the [rival] account." | make the adjudicating prediction explicit and test it |
| "Reads as framework-internal." | restate the generalization framework-neutrally |

Illustrative: a study of a reduplication pattern draws "competent description, thin analysis." Climbing
the ladder, the author states a general claim (an illustrative constraint interaction at the
morphology–phonology interface), derives a prediction about which bases block reduplication, and shows
a rival prosodic-template account predicts the opposite — moving from rung 1 to rung 3.

## Anti-patterns

- Presenting a gloss table or a corpus trend *as* the contribution, analysis left implicit
- "Analysis" that merely restates the data in formal notation (does no predictive work)
- Assuming one framework and never engaging the alternatives (parochialism)
- Predictions stated so loosely nothing could falsify them
- Ignoring the foundational account the reviewer will expect you to build on

## Output format

```
【Generalization】the pattern, stated theory-neutrally
【Analysis】the account + its explicit assumptions
【Predictions】what should / should not occur → research-design
【Rival adjudicated】the competing framework + the deciding prediction
【Scope】languages / constructions where the claim holds
【Cross-framework legibility】would an outsider accept the generalization? [Y/N]
【Next】lang-literature-positioning
```

## Supplementary resources

- [`../../resources/external_tools.md`](../../resources/external_tools.md) — analysis tooling across subfields
- [`../../resources/official-source-map.md`](../../resources/official-source-map.md) — Language theory-grounding and general-audience expectations