Authority caveat. Nobody on this site claims to know Navajo authoritatively. Every Navajo form is copied from a named published source with its example number; none is invented. Formal rules are hypotheses to be tested against Grant's texts.

SOURCED stated in a published source we opened HYPOTHESIS our formalization or inference GRANT'S CONJECTURE Grant's own claim OPEN needs data or reading

'Someone' has no person but is still human: the animacy ranking reads [human], not person, and Willie's "PASS 3.3" is a paradigm label (T74d)

2026-11-11 audit pass 2. Apostrophe fixes only. Navajo glottal stops typed as ' or ’ (for example in prefix names like ʼa- and ʼeʼe-, or words like Awééʼ) now print as ʼ (U+02BC), and a build step no longer turns the closing quote of an English translation inside a Navajo row into a glottal stop (‘…the girls.’, not ‘…the girls.ʼ). No form, gloss or translation wording changed.

Working note (newest). Not yet filed into a section; listed under Working notes.

Verbs on this page → dictionary: yiztał ‘kick’

Restated on the unified structure (2026-11-15). The indefinites (ʼa-, ʼadi-, areal ho-) are restated as personless pronouns in ordinary argument positions (H60): ʼadi- as the silent indefinite agent in Spec,vP that raises to Spec,TP by the default (H10b), ʼa- in VP. The 3i data on this page have not yet been individually restated. The head-direction mismatch argued below as rule sets A/B or a changeover is now stated as Grant’s Merge-site hypothesis and tested in §3.1–3.2. Status on the spine: see unified structure, §4, indefinite and passive rows. The data, judgments, translations and citations below are unchanged; the analysis below is kept as the record of what was argued then.

T74d asked which of two things in Willie 1991 the animacy ranking uses for INDF, the ʼadi- 'someone' agent of the agentive passive. On pp. 50–51 Willie lists ʼadi among the indefinite prefixes, apart from the first to fourth persons. In chapter 6 she glosses biʼdistał=yę́ę as "PASS 3.3.kicked", which looks like it files the agent as a third person. This run read the page images again and set the glosses side by side. In chapter 2 the same prefix is glossed with the patient's person only: "2s.PASS", "1s.PASS" (pp. 37, (21), (23)). In chapter 6 "3.3" is the label Willie puts on every 3-on-3 form of 'kick', Direct, Inverse and passive alike (p. 152, (1c), (1d), (1f)). And the ranking itself is stated over the animacy scale, not over persons (p. 86, (70)). So the ranking can read the [human] that p. 36 gives INDF, and INDF needs no person value at all HYPOTHESIS.

On this page: 1 Question · 2 Sources · 3 Rules first · 4 The data · 5 Answer to T74d · 6 What is left of p. 86 · 7 Rulesets A and B · 8 Derivations · 9 The role schema · 10 Fidelity notes · 11 Open questions

Notation is the same as on agent-ext and person-head. No Syntax · Semantics · Interface conventions file is in the repo or the context folder. A = the word level (prefixal, left-headed). B = the clause level (right-headed). Σ = the role schema, with Dowty's proto-roles as the baseline. EXT/INT = Willie's external/internal argument. INDF = the ʼadi- agent. Merge builds structure: it puts two pieces together into a bigger one. Agree builds nothing. It copies feature values such as person and number from one place to another. Move is Merge again of a piece that is already in the structure.

1. Question

EI-10 (ergative-indefinite) says INDF is on the animacy ranking, because it is human and indefinite, and an argument is off the ranking only if it is definite and human (EI-5's criterion). It also leaned on two supports that INDF counts as a third person: the gloss "PASS 3.3" (p. 152, and "PASS.3.3" on p. 157 (10)), and p. 86's line that "only third person arguments can be indefinite". agent-ext §8 found that pp. 50–51 put ʼadi in a separate class from the persons, which weakened the second support. T74d asks which of these the ranking actually uses. The question matters because the case against reading biʼdistał as an inverse (EI-8) rests on EI-10 (T73d).

2. Sources

SOURCED Willie, MaryAnn 1991. Navajo Pronouns and Obviation. PhD dissertation, University of Arizona. Page images rendered from the PDF and read this run: pp. 35–38 (PDF 40–43; offset printed + 5 in this part), 50–51 (PDF 55–56), 152 and 157 (PDF 158 and 163; printed + 6 here). P. 86 (PDF 92) was read from the PDF's text layer this run and matches the image-based quote already on ergative-indefinite (EI-4).

Already on the site and not new: (20)–(23) on pleonastic-psych and reflexive-reciprocal; the p. 36 "must be understood as human" quote (IX-5, indefinite-agent-b); (1c), (1d), (1f) and (10) on relatives. New here: setting the chapter 2 and chapter 6 glosses side by side, and the p. 35 boundary line "This completes the survey of the definite direct pronominals" before §2.3.

Not opened: Young & Morgan 1987 (the source of the term "agentive passive"), Hale et al. 1977 and Witherspoon 1977 (Willie's sources for the animacy hierarchy, p. 86).

3. Rules first

AP-1  §2.2 ends "This completes the survey of the definite direct pronominals"     SOURCED (Willie p. 35, image)
      and §2.3 opens "a group of pronominal prefixes that mark indefinite and
      non-referential arguments", which includes the agentive-passive prefix.
AP-2  "the indefinite agent must be understood as human, but no noun that          SOURCED (Willie p. 36, image; = IX-5)
      describes this agent may be adjoined."
AP-3  Chapter 2 glosses the agentive passive with the patient's person and          SOURCED (Willie p. 37, (21), (23), image)
      PASS, and nothing for the agent: (21) "2s.PASS.shaving", (23)
      "1s.PASS.painting".
AP-4  Chapter 6 glosses three 3-on-3 forms of 'kick' with the same "3.3":           SOURCED (Willie p. 152, (1c), (1d), (1f), image)
      (1c) "3.3.kicked" (Direct), (1d) "INV 3.3.kicked", (1f) "PASS 3.3.kicked".
      P. 157 (10) "PASS.3.3.will kiss" follows the same pattern.
AP-5  (70): the voice is chosen by "the animacy scale": agent higher → Direct;       SOURCED (Willie p. 86; = EI-4)
      patient higher → Inverse; equal → discourse. "The voice alternation can
      only be used if all arguments are third person. This clearly is
      connected with the fact that only third person arguments can be
      indefinite; first and second person arguments are always definite."
AP-6  "3.3" in chapter 6 is a label for the 3-on-3 paradigm cell that PASS or       HYPOTHESIS (from AP-3, AP-4)
      INV modifies. It is not a person value for INDF: in chapter 2, where the
      patient is 1st or 2nd, the agent slot gets PASS and no number.
AP-7  The ranking compares animacy (AP-5), and INDF's animacy is given directly    HYPOTHESIS (from AP-2, AP-5)
      by AP-2 ([human]). So the ranking needs no person value for INDF.
AP-8  Feature split for INDF: [human] yes, [definite] no, [person] none.           HYPOTHESIS (from AP-1, AP-2, AP-6, EI-1)
      Person and animacy are separate features in Willie's system, and the
      ranking reads only animacy.
AP-9  EI-10′ (replaces the support clause of EI-10; EI-10's criterion is            HYPOTHESIS (from AP-7, AP-8, EI-5)
      kept): an argument is off the ranking iff it is definite AND human.
      INDF is on the ranking because it is indefinite (AP-1) and human
      (AP-2). Neither "PASS 3.3" nor "only third person arguments can be
      indefinite" is needed.
AP-10 The precondition "all arguments are third person" (AP-5) is read as            HYPOTHESIS (reading of AP-5 with AP-1)
      "no argument is 1st, 2nd or 4th" (no definite human person). A
      personless INDF meets it, so a 3rd patient with INDF can be ranked.
      Whether Willie means this or a positive [3] on every argument is OPEN.

4. The data

All forms are copied from Willie's page images. Willie's gloss is given first, then a Leipzig line. Willie segments (21) herself; the other forms are not segmented further here.

(21), p. 37niʼdilzhééh (ni-ʼadi-l-zheeh in underlying form)
Willie: 2s.PASS.shaving · Leipzig: 2SG.OBJ-INDF.AGT-CL-shave
'You are being shaved.'
(23), p. 37shiʼdidleesh
Willie: 1s.PASS.painting · Leipzig: 1SG.OBJ.INDF.AGT.paint
'I am being painted.'
(1c), p. 152yiztał=yę́ę
Willie: 3.3.kicked=REL · Leipzig: 3OBJ.3SBJ.kick.PFV=REL
'the one he kicked' (Willie: "Direct patient; internal argument")
(1d), p. 152biztał=yę́ę
Willie: INV 3.3.kicked=REL · Leipzig: INV.3OBJ.3SBJ.kick.PFV=REL
'the one who kicked him' (Willie: "Inverse agent; internal argument")
(1f), p. 152biʼdistał=yę́ę
Willie: PASS 3.3.kicked=REL · Leipzig: 3OBJ.INDF.AGT.kick.PFV=REL
'the one who was kicked' (Willie: "subject of agentive passive")

In the Leipzig line for (1f) the "3.3" is not copied, because AP-6 reads it as a paradigm label. That reading is a HYPOTHESIS; Willie's own gloss is printed beside it.

5. Answer to T74d

The gloss leg. If "PASS 3.3" meant "a 3rd patient and a 3rd agent", chapter 2 should gloss (21) as something like "2s.3.PASS". It does not. It gives the patient's person and then PASS, which takes the agent's place in the gloss (AP-3). In chapter 6 the "3.3" is shared by a Direct, an Inverse and a passive form of the same verb (AP-4), so it reads as the name of the cell (third-person patient, transitive 'kick') that INV or PASS then modifies (AP-6) HYPOTHESIS. This leg of EI-10 is now as weak as the person-class leg.

What the ranking reads. Willie states the ranking over "the animacy scale" (AP-5) SOURCED. Animacy is exactly what p. 36 fixes for INDF: "must be understood as human" (AP-2) SOURCED. So the ranking can place INDF without asking what person it is (AP-7). INDF then has [human] and no [definite], and no person (AP-8) HYPOTHESIS.

Effect on EI-10. EI-10's criterion survives, and its two person-based supports can be dropped (AP-9, EI-10′). The case against EI-8 (reading biʼdistał as an inverse) still depends on EI-10′, which still has no direct evidence. That would be a form where an agentive passive's choice is visibly decided by ranking INDF against a non-human patient (T70a). T73d's request for direct evidence therefore stays OPEN, but T74d's "which one" is answered: animacy HYPOTHESIS.

6. What is left of p. 86

P. 86 says "only third person arguments can be indefinite" (AP-5). Read with a positive [3] feature, this files INDF as a 3 and clashes with AP-1 and AP-8. Read as "only arguments that are not 1st or 2nd can be indefinite", it is a statement about the definite persons and says nothing positive about INDF's person (AP-10) HYPOTHESIS. Willie's next clause, "first and second person arguments are always definite", is the contrast she draws, which fits the second reading. Which reading she intends is OPEN (T75a).

7. Rulesets A and B

StepLevelRulesetWhat happens
1vPBΣ assigns proto-agent and proto-patient. INDF carries [human, −def] and no person (AP-8).
2ClauseBChoose the voice. Precondition: no 1st/2nd/4th argument (AP-10). Rank by animacy (AP-5, AP-7). INDF agent: stays INT; the patient becomes EXT (VC-13 with a 1st/2nd patient; with a 3rd patient, Willie p. 152 "subject of agentive passive").
3WordAAgree copies person only from arguments that have it: the patient's into the object slot. The agent slot is spelled out as ʼadi-/ʼ-d-, not by a person value.
4WordASpell-out, prefixes left of the stem (left-headed).

The word is again the changeover point. Every ranking decision is made in B on features (animacy, definiteness), and A spells out person where there is a person to spell out HYPOTHESIS.

8. Derivations

D1  (21) ni'dilzhééh 'You are being shaved.' (Willie p. 37)
1  Merge  [VP pro(2sg)  shave ]                            B; head-right VP
2  Merge  [vP INDF[hum,−def,no person] [VP ...] v ]         B; VPISH (AP-8)
3  Check  voice: a 2nd-person argument is present, so       B; AP-5, AP-10; = VC-13
          no Direct/Inverse choice; ranking not consulted
4  Move   pro(2sg) → EXT; INDF stays INT                     B; VC-13 (third voice)
5  Agree  object slot ← [2sg] from pro(2sg)                  A
          agent slot: no person to copy (AP-8)
6  Spell  ni- (2SG.OBJ) + 'adi- (INDF) + l (CL) + zheeh      A; left-headed prefix array;
                                                              Willie's underlying form
7  Output ni'dilzhééh

D2  (1f) bi'distał=yę́ę 'the one who was kicked' (Willie p. 152)
1  Merge  [VP pro(3)  kick ]                                B
2  Merge  [vP INDF[hum,−def,no person] [VP ...] v ]         B; AP-8
3  Check  voice: no 1st/2nd/4th argument → ranking applies   B; AP-10
4  Rank   INDF [hum] vs pro(3) on the animacy scale          B; AP-5, AP-7 (no person read)
5  Move   pro(3) → EXT; INDF stays INT                       B; Willie: "subject of agentive passive"
6  Agree  object slot ← [3] from pro(3)                      A
7  Spell  bi- + 'adi-/'d- + stem                             A; not segmented further
8  Merge  [ clause ] + =yę́ę (REL, binds EXT)               A/B boundary; XA-1 (p. 181)
9  Output bi'distał=yę́ę

Step 4 of D2 shows where the ranking applies, not its outcome. Willie gives no judgment on which voice is preferred when a human 'someone' kicks a non-human patient. That is T70a, and it stays OPEN.

9. The role schema (Focus 2)

This run separates two kinds of information that a verb argument carries. Its role (proto-agent, proto-patient) comes from Σ. Its features ([human], [definite], person) come from the item filling the slot. The ranking that chooses the voice reads only features, and here only animacy (AP-7). That fits a weak schema thesis: roles are assigned schematically and the grammar then consults features of what fills them HYPOTHESIS. INDF is a useful test case because it has animacy but no person and no noun can be added to it (AP-2). It is about as close to a bare role filler as Navajo gets, and it still carries a lexical-type feature ([human]) that the clause uses. So the strong thesis (roles, not lexical items, are the arguments) gets no support from it and stays OPEN. Grant's own revised view, that the strong version does not hold cleanly, is GRANT'S CONJECTURE.

10. Fidelity notes

All Navajo forms are copied from the page images. Willie's "unspec." and "PASS" are rendered INDF.AGT in the Leipzig lines. (23) and the chapter 6 forms are not segmented, because Willie does not segment them. The p. 86 quote is from the PDF's text layer, which has OCR errors elsewhere on the page ("Heirarchy" is Willie's own spelling); the sentences quoted match the image-based quote on ergative-indefinite. No forms or judgments were invented, and D2 step 4 states no outcome.

11. Open questions

IDQuestionWhat would settle it
T75aDoes p. 86 "only third person arguments can be indefinite" give INDF a positive [3], or only say that indefinites are not 1st/2nd (AP-10)?A passage in Willie ch. 3–4 or Willie 1990 that ranks or agrees with INDF as a 3. OPEN
T75bDoes any source gloss an agentive passive with a 3rd patient as "3.PASS" (chapter 2 style), which would confirm AP-6?Young & Morgan 1987 (not opened); other Willie chapters on images. OPEN
T75cIs [human] on INDF a feature the ranking reads, or only an interpretive restriction (AP-7)?T70a: an agentive passive with a non-human patient, with a voice judgment. OPEN
T75dDo the other indefinite prefixes of pp. 50–51 ('a, ho-) also lack person while carrying animacy (p. 38: 'a "marks an indefinite inanimate argument")?Willie §2.3.2 on images (pp. 38–49). OPEN

Still open from earlier: T73b, T73c, T73d (direct evidence), T73e, T74a–c, T74e. Sources: Willie 1991 pp. 35–38, 50–51, 152, 157 (page images, this run); p. 86 (text layer, this run). See also agent-ext, ergative-indefinite, third-voice, indefinite-agent-b, open questions, sources.