---
title: 'Mostly Agree: Merge and Agree inside the Navajo verb word'
date: '2026-10-27'
updated: '2026-10-27'
authors: Grant Golland (research by Grok Bot)
abstract: 'Grant asked whether the Navajo verb word has Merge sites, places where
  something is built or moved, and not only agreement. This essay argues that most
  of what happens inside the word is Agree. The few Merge sites are at the left edges
  of domains. The ''start to'' placeholder is the closest match anywhere on the site
  between a Navajo fact and a mid-level Minimalist term, an expletive filling an EPP
  or Edge requirement. Checking Chomsky''s 2000/2001 definitions forced one revision:
  the placeholder cannot sit on the word''s phase head.'
status: HYPOTHESIS (Agree inside, Merge at edges; one Merge site downgraded to OPEN;
  MA-2′(c) moved to P_INCH)
---

# Mostly Agree: Merge and Agree inside the Navajo verb word

*Grant Golland (research by Grok Bot)*, 2026-10-27. Status: HYPOTHESIS (Agree inside, Merge at edges; one Merge site downgraded to OPEN; MA-2′(c) moved to P_INCH).

**Abstract.** Grant asked whether the Navajo verb word has Merge sites, places where something is built or moved, and not only agreement. This essay argues that most of what happens inside the word is Agree. The few Merge sites are at the left edges of domains. The 'start to' placeholder is the closest match anywhere on the site between a Navajo fact and a mid-level Minimalist term, an expletive filling an EPP or Edge requirement. Checking Chomsky's 2000/2001 definitions forced one revision: the placeholder cannot sit on the word's phase head.


*Restated on the unified structure (2026-11-15): Merge sites and Agree sites inside the word are now the basis of Grant’s hypothesis about the headedness mismatch (GRANT’S HYPOTHESIS, tested on the sourced template in §3.2). See <https://navajo-decoded.grant-golland.workers.dev/unified-structure/#us-word>. The essay’s data, judgments and citations are unchanged; it is kept as the record of what was argued then.*

<h2 id="2026-11-11-audit-pass-1">2026-11-11 audit pass 1</h2>

“Needs nothing to move” described a possible realization model, not evidence that V stays put. Earlier source phase/Agree conditions are labeled contrasts, not axioms. Light d, expletive placement, the MA/EF rule family and the identity of a Navajo T remain hypotheses. Test Grant’s V→v→T and separate subject raising against PF merger/template filling on the same left-specifier vP/TP skeleton, preserving an explicit event LF. The old essay’s full derivations remain OPEN for pass 2.


<h2 id="2026-11-13-audit-pass-7">2026-11-13 audit pass 7</h2>

The pass-1 note above treats V→v→T and subject raising as a comparator to test against PF/template rivals. Since 2026-11-13 they are the working default in tensed clauses: VPISH frames where the arguments start so that syntax and semantics line up; it is not a word-order account. PF merger and template filling are kept only as labelled contrast, for how exponents line up inside the word, and the open question is what else has to move. The pass-1 note is kept as written. See the [2026-11-13 note](/derivation/#2026-11-13-vto-t-default) and the [pass 7 report](/derivation/#2026-11-13-audit-pass-7).

## 1. The question

Grant's note asks which Navajo template positions are Merge sites and which are Agree sites, and whether mid-level notions from Chomsky (EPP, feature inheritance, Multiple Agree) apply inside the word **[GRANT'S CONJECTURE]**.

## 2. The data

A fourth-person subject leaves a third-person object with no prefix:



> **(1)** *jishxash 3.4.bit ‘That person bit it.’*  
>   
>   
> — Willie 1991:124–125, (23)–(25); copied from the working note “Fourth person: escape from the animacy scale” <https://navajo-decoded.grant-golland.workers.dev/fourth-person/>, ex. (23b)



The causative *b-* sits on the incorporated postposition:



> **(2)** *ʼAwééʼ d-ee-za’.*  
> baby d-SPF:3-belch:PERF  
> ‘The baby burped.’  
> — Hale 1997, PDF p. 53 (page image), (41); “¬” restored as ł; copied from the working note “Attached or separate? Procliticized vs detached postpositions, where the word ends, and which verb + postposition pairs are idioms (T43e)” <https://navajo-decoded.grant-golland.workers.dev/procliticized-postpositions/>, ex. (41a)





> **(3)** *ʼAwééʼ bi-di-y-é-sa’. (< b-d-y-s-´-sh-ł-za’)*  
> baby 3-d-y-SPF:1s-ł-belch:PERF  
> ‘I burped the baby.’  
> — Hale 1997, PDF p. 53 (page image), (41); “¬” restored as ł; copied from the working note “Attached or separate? Procliticized vs detached postpositions, where the word ends, and which verb + postposition pairs are idioms (T43e)” <https://navajo-decoded.grant-golland.workers.dev/procliticized-postpositions/>, ex. (41b)



The 'start to' forms of HMP and Hale (*shiná'niidlóóh*, *'iná'niichah*, *'i'niichííl*, *bi'niichx'ǫ'*) are quoted in the working notes on the inchoative, from HMP (2b), (7b), (10a), (10d), (16a) **[SOURCED]**.

## 3. The argument

**Agree inside** **[HYPOTHESIS]**. Three facts need nothing to move inside the word:

1. The ordinary subject prefix is pronounced on Mode/Subject, which copies person and number from the slot where it was built (Hale 2003 n. 11).
2. A postposition's pronoun, including the inverse *bi-* in W&J (50b) *bitah*, is valued inside the PP.
3. The choice among *yi-*, *bi-* and Ø is made inside the word, before any noun arrives.

**Merge at edges** **[HYPOTHESIS]**. There are three candidates, all at the left edge of a domain:

- HMP's "bumping" of an object into disjunct Position I;
- the deictic piece of a subject (*ji-*, *'-*, *ho-*);
- the role-less placeholder *'* that 'start to' adds when nothing fills its object requirement.

After the originals were read, the deictic piece was downgraded to open. Chomsky's feature inheritance allows it to be either Agree morphology on the top head or Merge by the top head's edge feature, and the two readings give the same strings.

**The placeholder is an expletive** **[HYPOTHESIS]**. Chomsky (2008:157) says an edge feature that cannot be satisfied by movement "must be satisfied by EM, necessarily of an expletive since no argument role can be assigned", and that a head with no accessible goal shows "default morphology" **[SOURCED]**. HMP describe exactly this.

## 4. Counter-evidence and revision

**MA-2′(c) moves to P_INCH** **[HYPOTHESIS]**. MI and DbP allow pure Merge of an expletive only at a non-phase head like T: "If H is v/C, XP is not introduced by pure Merge" (MI (5a), p. 102) **[SOURCED]**. The rule MA-2′(c) had put the placeholder in the qualifier layer, the word's phase head. That is exactly the excluded case. The rule moves back to P_INCH, where IN-8 first put it. Two facts fit a T-like head there: INCH's requirement is obligatory, and the position carries no role.

**Not Hiraiwa's Multiple Agree** **[SOURCED]** **[HYPOTHESIS]**. The object prefix reads two different things from two different goals. Hiraiwa's operation has one probe valuing one feature on several goals, and the probe shows only the closer goal (p. 79 n. 12). The site's "Multiple Agree" candidate is therefore something else.

## 5. Status

Whether the deictic subject piece is Agree or Merge is open. So is whether the low participant under 'start to' raises or is reached in place (T37b), since no printed form rules out either **[OPEN]**.


## References

1. Chomsky, Noam 2000. Minimalist inquiries: the framework. In Martin, Michaels & Uriagereka (eds.), Step by Step, 89–155. MIT Press. [opened]
2. Chomsky, Noam 2001. Derivation by phase (DbP). [opened]
3. Chomsky 2008 (p. 157 as recorded on the site). [opened]
4. Hiraiwa 2001. [opened]
5. Hale, Munro & Platero (inchoative manuscript; HMP), nn. 1–4, exx. (2), (7), (10), (16). [opened]
6. Hale, Ken 2003 (manuscript), n. 11, p. 10. [opened]
7. Hale, Munro & Platero 2000, MITWPL 3, pp. 97–116. [not opened]
8. Young & Morgan 1980. [not opened]

Canonical HTML: https://navajo-decoded.grant-golland.workers.dev/essays/merge-agree/
