Voice, causatives and event structure

Legacy link recovery
Legacy section links (scroll sideways); old articles remain available.
theta-schemaderivationsderivationmaintained-statecausative-splitinchoativeinchoative-hmppleonastic-psychevent-structure

On the spine (2026-11-14 summary)

2026-11-14 overhaul (approved by Grant for the live site, 2026-11-14). This summary states the topic on the spine (V→v→T and subject raising to Spec,TP assumed; each pattern is what else moves). It supersedes the framing of the dated sections below, whose data, judgments and citations are unchanged.

  • HYPOTHESIS Every theta role is assigned inside vP: the internal argument in VP, the external argument in Spec,vP (VPISH as a syntax–meaning device). The method, LF conventions and tests are on the spine.
  • HYPOTHESIS The passive keeps a silent indefinite agent in Spec,vP (H10b/H16a): biʼdistał=yę́ę (NX-315). Results §12.
  • OPEN Telicity: no pair in thirteen reviews differed only in telicity. Two narrow claims are contradicted: =yę́ę attaches to states and to the telic ‘fix’ alike (NX-367), and the form of =go is the same with bounded and activity verbs (review 13). Mode and viewpoint labels are not telicity tests. Results §13.
  • OPEN Causatives and the classifiers have not yet been restated on the spine; the sections below keep their earlier analyses as contrast.

Theta roles as arguments: the proto-role schema

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. Kept as written (URL, data and status labels unchanged). Filed under The roles-vs-words thesis; the consolidated argument is the essay roles-not-words. Rule IDs: rules index.

Verbs on this page → dictionary: yiztał ‘kick’ · yishxash ‘bite’

Restated on the unified structure (2026-11-15). The framework is restated in §1: VPISH (arguments merged in the verb’s scope, subject in Spec,vP) with theta roles, and V→v→T with the subject raised to Spec,TP as the working default in tensed clauses; each pattern is what else moves, with an LF whose scope matches c-command. Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. Status on the spine: see unified structure, §1. The data, judgments, translations and citations below are unchanged; the analysis below is kept as the record of what was argued then.

What would it mean, formally, for the Navajo verb's argument domain to hold "only a schema of the theta roles … prior to realization as a lexical item"?

On this page: 1 Claim · 2 Definitions · 3 Well-formedness · 4 The yi-/bi- rule · 5 Fourth person · 6 Valence operations · 7 Three theories compared · 8 Strongest version

1. The claim GRANT'S CONJECTURE

"their arguments aren't set within the domain of the VP, but rather only a schema of the theta roles, which can properly be called proto roles here, since they are prior to realization as a lexical item."

Two parts. (a) What occupies the argument positions inside the verb's projection is a role, not a lexical DP. (b) Those roles are proto-roles in Dowty's sense: clusters of entailments, fixed by the event before any lexical item is chosen.

2. Definitions HYPOTHESIS (built from sourced pieces)

Entailment inventory. SOURCED Dowty 1991:572, exx. (27)–(28) (page image checked).

Proto-Agent (27)Proto-Patient (28)
vol: volitional involvement in the event or state
sent: sentience and/or perception
caus: causing an event or change of state in another participant
mov: movement relative to the position of another participant
(indep: exists independently of the event named by the verb)
chg: undergoes change of state
inc: incremental theme
aff: causally affected by another participant
stat: stationary relative to movement of another participant
(dep: does not exist independently of the event, or not at all)

Dowty's Argument Selection Principle (1991:576, (31)): the argument with the most Proto-Agent entailments is lexicalized as subject, the one with the most Proto-Patient entailments as object. Corollary 1, (32), and Nondiscreteness, (34), are on the same page. Dowty adds that the ASP constrains which lexical predicates are possible; it is not a step in a syntactic derivation. That matters below: if the schema is going to do syntactic work, it needs more than the ASP.

Proto-role slot. A slot is a triple

ρ = ⟨ E_ρ , pos_ρ , φ_ρ ⟩
  E_ρ   ⊆ PA ∪ PP         entailment profile, fixed by the event (root + v)
  pos_ρ ∈ {Spec,V ; Comp,V ; Spec,v ; Obl(P)}     first-Merge site
  φ_ρ   = [π: _ , #: _ , rank: _ ]               unvalued at first Merge

Schema. For a root √ and voice/classifier head v, Σ(√, v) = {ρ₁ … ρₙ}, n ≤ 3. First-Merge sites follow Hale's (2003) structural argument positions:

  • sole argument in Spec,V for the labile/unaccusative type (Hale 2003 ex. (22));
  • Spec,v for the unergative type, e.g. cha ‘cry’ (Hale 2003 ex. (23));
  • Comp,V for the object of a transitive (Hale 2003 tree (8)).

Grant's distinctive restriction.

(G)  No lexical DP is ever first-Merged at pos_ρ for any ρ ∈ Σ.

Exponents. The values of φρ are spelled out by verb-internal pronominal exponents. In the sources these are the conjunct object and subject markers: sh- ‘1s’ and ni- ‘2s’ in Hale 2003 ex. (1); the third-person object forms yi-/bi-; the fourth person ji- (subject) and ho- (object) (Willie 1991 §4.1). They also include null exponents, e.g. the Ø 3rd-person subject in the imperfective (NLA Navajo Verb Modes).

3. Well-formedness conditions HYPOTHESIS

A clause built on Σ is well formed iff conditions C1–C9 hold.

#ConditionStatus / basis
C1Exactly-once valuation. There is a bijection between Σ and the pronominal exponent occurrences in the verb word (null exponents included). Every slot is valued, and no slot is valued twice.HYPOTHESIS The PAH counterpart: argument positions "are satisfied always and only … by clitic pronouns" (Jelinek 1984:44, of Warlpiri). For Navajo, see Hale 2003:29: arguments are "satisfied in situ".
C2Event-fixity. Eρ does not change under the yi-/bi- choice or under fourth-person choice.SOURCED in effect: Hale gives the same translation for the direct and inverse members of a pair ("Conflicting Truths" ex. (2), prior run). The NLA grammar draft, ch. 3 "Basic Clause Structure" (Fernald, prior run), says the two "have the same meaning but have different uses in discourse".
C3Adjunct linking. Each overt DP links to exactly one ρ, and its linear position is fixed by rule set B (topic first), not by posρ.HYPOTHESIS, consistent with W&J 2000 (nominals as topic/focus adjuncts). Refined 2026-10-11 (evening): Willie 1991:51–70 confirms that nouns are adjuncts licensed by pronouns, but position fixes only a default link. The verb's and the noun's meaning can choose another slot (nominal adjunction §11, ADJ-SEM, ENT-LINK).
C4Ranking domain. R(Σ) = {ρ ∈ Σ : π(ρ) = 3}. Voice alternation is defined iff |R(Σ)| = 2 and Σ is transitive.SOURCED W&J 2000:264: the alternation "is excluded where there is a first-, second-, or 'fourth'-person argument, and in intransitive sentences"
C5Agent identification. ρA = the slot in R with the most PA entailments, ρP = the other.HYPOTHESIS (Dowty's ASP applied to the ranked pair)
C6Rank preference. The DP linked to the topic slot should rank at least as high on the status scale as the other DP.SOURCED W&J 2000:273; Creamer 1974 (eight groups); Willie 1991:122 ("must")
C7Inverse definiteness. Under bi-, the DP linked to ρP (the topic) is definite.SOURCED W&J 2000:271, (54)–(56); Willie 1991:126
C8Uniqueness. At most one bi- (inverse) per clause.SOURCED W&J 2000:270
C9Fourth-person exit. If π(ρ) = 4, then ρ ∉ R.SOURCED Willie 1991:122: fourth person is "not involved in the animacy hierarchy"

4. The yi-/bi- linking rule

HYPOTHESIS Given C4–C5, and writing TOP for the slot linked to the clause-initial topic DP:

VOICE(Σ) = yi-   if TOP = ρ_A        (direct)
VOICE(Σ) = bi-   if TOP = ρ_P        (inverse)
subject to  C6: rank(DP_TOP) ≥ rank(DP_other)

Update 2026-10-08, T5: Willie 1991 (39), (72) show that the voice differs, and can be starred, with no nominals at all. So VOICE is restated over the external slot instead of over the topic DP: yi- iff the external slot is ρA, and the external slot is the higher-ranked one. The topic facts follow from linking. The outputs below do not change. See the inverse page, §3.

This is Hale's parsing generalization, restated over slots. In (NP₁) NP₂ X AGROBJ V: if AGROBJ = yi-, NP₂ is the object; if bi-, NP₂ is the subject ("Conflicting Truths" ex. (4), prior run).

Corollary HYPOTHESIS. Let ρrank be the slot whose DP is highest on the status scale. When the two ranks differ, C6 forces TOP = ρrank, so:

bi-  ⟺  ρ_rank ≠ ρ_A

In words: the inverse appears exactly when the referent order (animacy/status, fixed by the lexical DPs) and the proto-agent order (entailments, fixed by the event) pick different slots. When the ranks are equal, both voices are available (Creamer 1974). Qualified 2026-10-14: this holds for animates only. Two equal-ranked inanimates allow no inverse in either order (Witherspoon 1977:66, (11)–(14)); see INV-CAP on the who acts on whom page. Restored 12:46pm: the qualification is withdrawn. In those sentences the moving inanimate outranks the static one (Creamer 1974:36), so they are not equal-rank pairs, and INV-CAP is falsified by Creamer 1974:37 (Eight ranks). The inverse thus diagnoses a mismatch between the schema and its lexical realization. That is about as close as published data come to showing that the slots exist independently of the DPs that later fill them.

Test on the sourced data (forms and glosses as printed):

Ex.FormρA / ρP (by entailments)RankPredictedSource judgment
W&J (61a)łį́į́ʼ tsé yiztałhorse / rockhorse > rockyi-, horse topic✓ ‘The horse kicked the/a rock.’
W&J (61b)tsé łį́į́ʼ biztałhorse / rockhorse > rock✗ (rock topic violates C6)?*
W&J (62a)łį́į́ʼ ʼashkii yiztałhorse / boyboy > horse✗ (horse topic violates C6)?*
W&J (62b)ʼashkii łį́į́ʼ biztałhorse / boyboy > horsebi-, boy topic✓ ‘The boy was kicked by the/a horse.’
Willie (20a)ʼashkii łééʼchąąʼí yishxashboy / dogboy > dogyi-✓ ‘The boy bit the dog.’
Willie (20b)łééʼchąąʼí ʼashkii yishxashdog / boyboy > dog✗*
Willie (20c)ʼashkii łééʼchąąʼí bishxashdog / boyboy > dogbi-✓ ‘The boy was bitten by the dog.’

W&J = Willie & Jelinek 2000:273 (page image checked this run). Willie = Willie 1991:122 (page image checked this run; the typescript renders the nasal hooks as cedillas, and łééʼchąąʼí transcribes its łééʼcha̧a̧ʼí). W&J note: "Speakers vary in respecting the hierarchy, and its use is declining now under language loss" (p. 273). C6 is therefore a preference whose strength varies by speaker. That is not an exceptionless rule, and the ?* marks reflect it.

Where the agent is hosted. SOURCED Willie (1991:123) distinguishes the direct-voice third-person agent, "the external Nominative argument", from "an internal (Ergative) agent, as we see in the Inverse". HYPOTHESIS In slot terms: bi- relinks ρA from the external position to an internal one, and Eρ stays the same (C2). This is a change of pos, not of role. It is the one place where C2 has bite: the voice alternation may move a slot, but it may never rewrite its entailments.

5. The fourth person as an exit from the ranking

Willie 1991 (21)
łééʼchąąʼíshishxash
dog1.3.bit

‘The dog bit me.’

Willie 1991 (22)
łééʼchąąʼíhashxash
dog4.3.bit

‘The dog bit that person.’

Willie 1991:123 (page image checked this run). Compare (20b), * with two third persons.

HYPOTHESIS By C9, valuing ρP as 4 removes it from R, so |R| = 1 and VOICE is undefined (C4). The default (direct) linking then applies even though the agent is lower in animacy. This matches Willie's "the speaker can use a fourth person patient pronoun and avoid having to use an Inverse form" (p. 123; prior-run note).

The schema analysis predicts that exponent choice (π value) feeds the ranking, which feeds voice. So the order of operations is Σ valuation of π, then computation of R, then VOICE. In a derivation, π-valuation must therefore come before the step that fixes yi-/bi-. See D5. Full treatment (Willie §4.2.7, KEEP-DIR vs EDGE, Q–A, coreference, derivations F1–F5): fourth-person page (T8, 2026-10-08 evening).

6. Where the schema could earn its keep: valence operations HYPOTHESIS

If arguments really are slots with entailment profiles, valence operations should be stated over Σ, with no reference to lexical DPs. Hale & Platero (1996) give the sourced contrast:

  • SOURCED Simple transitivity alternation: intransitive vs transitive with only the ł-classifier added, and "no 'meaning' attached to the ł-classifier" (H&P 1996, prior run). Example: rope pair, Gessner 2003 ex. (6), below.
  • SOURCED Complex causative: ł plus a position-VI y-, plus a causee object that is overt bi- even with a 1st/2nd-person subject "as if attached to an incorporated postposition" (H&P 1996, prior run; Gessner 2003 exx. (10)–(11), page images checked this run).
  • SOURCED The split follows unaccusative vs unergative (H&P 1996; Gessner 2003). Alternating, by English gloss: float, whiten, dry, fall, crack, increase, flow, burn, roll, bend, cook, darken, harden, yellow, stretch, wear away, become gentle. Non-alternating: swim/bathe, cry, play, sing, spit, gasp, pant, inhale (H&P list, prior run; Navajo stems not reprinted).
ADD-CAUSER(Σ):  Σ' = Σ ∪ {ρ_C},  E_ρC ⊇ {caus, vol},  pos_ρC = Spec,v(outer)
  (a) if Σ = {ρ₁} with pos_ρ1 = Spec,V        → spell-out: ł only             (simple)
  (b) if Σ = {ρ₁} with pos_ρ1 = Spec,v        → ρ₁ relinked to Obl;
                                                 spell-out: Qual y + ł + oblique bi-   (complex)
  (c) if |Σ| ≥ 2 (transitive base)            → morphological causative rare;
                                                 analytic (periphrastic) causative     (Kibrik 1993:51 via Gessner)

Update 2026-10-07 (evening), T4: tested against Hale & Platero's lists on the plain ł-causer page. Mov is dropped as a trigger for (b), since motion verbs alternate. The position condition is kept. The entailment claim now sits in a separate linking rule, LINK-1: the sole slot is external iff it is volitional, or sentient with an atelic predicate. That rule fits H&P's whole sample but has not yet been shown to beat a lexical diacritic. Case (c) is relaxed: H&P print a complex causative on a transitive theme.

Update 2026-10-07: a stative base (no chg, no vol) falls outside cases (a)–(c). See MAINTAIN and conditions C10–C11 on the keep vs bring about page.

Case (b) uses Hale 2003's structure for the complex causative: an outer classifier ł, Qualifier y, an embedded vP, and an oblique object (Hale 2003 exx. (17b), (18c), (19); structure read from the prose and trees, Navajo strings not reprinted because the text layer is font-garbled). HYPOTHESIS In entailment terms, (a) applies when the sole slot's profile is P-Patient-dominant (chg), and (b) when it is P-Agent-dominant (vol/mov). OPEN The H&P lists could test whether the entailment profile predicts the split better than a lexical unaccusative/unergative diacritic. For example, are there volitional verbs among the alternating ones? On the English glosses, "roll" and "float" are the cases to check.

7. Three theories compared

(A) Grant's proto-role schema(B) Pronominal Argument Hypothesis (Jelinek 1984; W&J 2000; Hale 2003)(C) Lexical-DP / agreement analysis (Perkins, Platero, Speas, per Fernald 2000 §3.1.1)
What sits in the argument positionρ: an entailment profile with unvalued φa pronominal (D) element, the verb-internal prefixa lexical DP or pro; the prefix is agreement
Overt nominalsadjuncts linked to ρ (C3)adjuncts ("simply optional adjuncts", Jelinek 1984:44, of Warlpiri)arguments that A-move
Inverseρrank ≠ ρA (§4)topic/focus voice over weak pronouns (W&J)object fronting (OSV) plus agreement change
Fourth personπ = 4 exits Rpronoun outside the hierarchy (Willie 1991)as (B), stated over agreement
VPISH movementnone for ρ; DPs are linked, not movednone (in situ; Hale 2003:29)Spec-to-Spec raising of DP
Distinguishing predictionvalence operations and alternations stated over entailments; no D-properties at the θ-position itselfD-properties (definiteness, Condition C behaviour) at the prefixreconstruction of DPs into VP; Condition C effects from VP-internal copies

Honest assessment. HYPOTHESIS (A) and (B) are close to notational variants. The difference is whether what is merged at posρ is a role (A) or a pronoun (B). The sourced facts give the pronominal exponents D-like properties. The fourth person is "exclusively definite" (W&J 2000:278). Under bi-, the topic must be definite (C7). (B) reads this directly. (A) has to say that φ/D features come in with the exponent at valuation, not with the slot. That works, but it adds a step. (A) earns its keep only if some generalization refers to Eρ and cannot be stated over pronouns. The candidates here are the inverse corollary (§4) and ADD-CAUSER (§6). For ADD-CAUSER, see LINK-1 (E) vs (D) on the plain ł-causer page.

8. Strongest formalization reached in v0

HYPOTHESIS The verb's argument domain is a set Σ of slots, each with an event-fixed Dowty profile and an unvalued φ-bundle. Each slot is valued exactly once by a verb-internal exponent (C1), and lexical DPs are linked to slots only as clause-level adjuncts (G, C3). Voice is a function from Σ plus topic choice: bi- appears exactly when the referent-rank maximum and the proto-agent maximum fall on different slots, and only when exactly two third-person slots are ranked (C4–C9). The fourth person takes a slot out of the ranking.

VPISH derivations, step by step

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.

2026-11-11 audit pass 3. Status badges now use only the label set SOURCED, HYPOTHESIS, SPECULATIVE, OPEN, NONE and GRANT'S CONJECTURE; verdict or task words that had been styled as badges (for example Opened, PRE-VPISH, DONE, SUPPORTS, PARTLY ANSWERED) are now plain text or a label with a qualifier. See the pass 3 report.

Working note. Kept as written (URL, data and status labels unchanged). Filed under Merge and Agree inside the word; the consolidated argument is the essay merge-agree. Rule IDs: rules index.

Verbs on this page → dictionary: yiztał ‘kick’ · yiztsʼǫs ‘kiss’ · yishxash ‘bite’ · dabiiłtsą́ ‘see’ · nishhozh ‘tickle’ · nayiisniiʼ ‘buy’ · kʼí-ní-dláád ‘break (rope)’ · naa-ghá ‘walk around (one)’ · d-ee-zaʼ ‘belch, burp’ · yi-sh-cha ‘cry’ · hólǫ́ ‘exist; (with -í ‘with’) have’

Restated on the unified structure (2026-11-15). The framework is restated in §1: VPISH (arguments merged in the verb’s scope, subject in Spec,vP) with theta roles, and V→v→T with the subject raised to Spec,TP as the working default in tensed clauses; each pattern is what else moves, with an LF whose scope matches c-command. Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. 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. Position-class template statements below describe the order of prefixes; they are not the derivation (§3). Derivations below in which the verb or subject does not raise are labelled contrast; the working default raises V→v→T and the subject to Spec,TP. Status on the spine: see unified structure, §1. The data, judgments, translations and citations below are unchanged; the analysis below is kept as the record of what was argued then.

2026-11-11 audit pass 1. Obvious errors corrected: D1 copied a person value from an “unvalued” argument and displayed a constructed Ø string as output; its English row was translation, not a source gloss. D5 moved before any target-creation step. D4 mislabeled complement-to-specifier movement. The slot/role ASCII tree is withdrawn rather than passed off as categories. C-head assignments, A/B cycles, blanket inverse focus bans and their timing are HYPOTHESES, not sourced facts or axioms. Each construction now has an editorial LF check; complete scope/reconstruction proofs remain OPEN. The current primary comparator is V→v→T plus separate subject raising.

2026-11-11 audit pass 4. D8 gloss fix: Hale & Platero 1996 gloss the mode of naa-ghá and na-b-ii-sh-ł-á as IMPF (imperfective). This page printed IMP, which in Leipzig glossing means imperative; now IMPF, also in the derivation step. See the pass 4 report.

Interlinear rows whose form and gloss do not line up one-to-one were re-checked against the source page. Each is marked † with a note under its table: SOURCED where the source prints it that way, OPEN where the source cannot confirm it.

2026-11-13 audit pass 7. An earlier dated note on this page treats V→v→T and subject raising as Grant’s conjecture or 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/template accounts are kept only as labelled contrast, and the open question is what else has to move. The earlier note is kept as written. See the 2026-11-13 note and the pass 7 report.

Eight construction types. Each derivation is a numbered sequence of Merge, Agree, Move and spell-out steps. Each step is tagged with the rule set it uses. The examples retain earlier source citations. Source transcriptions, English translations and editorial analyses are distinguished below; this audit does not re-verify every old source image.

D1 intransitive · D2 local transitive · D3 direct yi- · D4 inverse bi- · D5 fourth person · D6 polar question · D7 wh-question · D8 causatives · what the derivations show

Conventions

Legacy notation (historical hypotheses, not the current core categories). Qual/M/S are morphological or author-specific labels, and ρ names a role/argument, not a phrase category. The current subject base is left Spec,vP.

  • Heads, bottom-up: V (stem), v = Voice (the "classifier": Ø, ł, d, l), M/S = Mode/Subject portmanteau (≈ Infl/T; Hale 2003:10), Qual = Qualifier (position VI), then Top/Foc and C above the word.
  • Selection: Qual selects M/S, and M/S selects v+V (Hale 2003:32).
  • ρ = proto-role slot (see schema definitions). Under the Pronominal Argument reading, read ρ as a pronominal D. Copies are written ⟨x⟩. Agree is written Probe ⟶ Goal and described in words.

Rule-set tags. Σ schema step (choice of slots, valuation of φ). A word-internal linearization: functional heads precede their complements, lexical V follows its complement (refined conjecture (i′) on the headedness page). B clause-level linearization and linking: topic/focus DPs, particles, postpositional phrases.

Restated 2026-10-16 (12:46pm, T27j): tags A and B are now read as spell-out statements: rule set A = Hale 2003's skeleton-filling SA1–SA5, rule set B = SB1–SB5, and both are one rule CYC applied to the word (D1) and to the clause (D2). Hale's (1), (13), (20b), (31a), (34) are derived step by step under CYC and under an antisymmetric alternative on A/B as spell-out. The tags are retained as historical bookkeeping. They do not independently establish a word boundary, a head direction or a compulsory order of operations.

Two variants where they differ.

  • V-H (Hale 2003): pronominal arguments are licensed in situ, with no A-movement (p. 29). The object prefix ends up leftmost through phonological filling of the bipartite skeleton (pp. 30–36). SOURCED
  • V-P (VPISH overlay): Grant's two movements are added. (a) Spec-to-Spec A-movement of the first argument, Spec,v → Spec,M/SP. (b) "Another type of movement possible for arguments beyond the first": the object goes to a higher Spec. Hale (2003:11–12) considered (b), as raising to Spec,M for Case, and rejected it because it conflicts with the PA hypothesis. HYPOTHESIS

Glossing. Source glosses are kept separate from editorial interpretation. The D1 English row is translation, not a morpheme gloss. Words the source does not segment (e.g. yiztał, glossed 3ACC-3NOM-kicked) are not segmented here. A morpheme the source does not gloss is marked "unglossed".

D1. Simple intransitive (unergative)

NLA Verb Modes, imperfective paradigm
yi-sh-cha yi-ni-cha yi-cha
‘I’m crying.’‘You’re crying.’‘S/he is crying.’

Fernald, Denny et al., Navajo Verb Modes (NLA, talkingdictionary.swarthmore.edu), imperfective section; the segmentation is theirs. Gloss compiled from their prose: "imperfective verbs have sh-, ni-, and Ø- (which is not pronounced) for their first, second, and third person markers". So -sh- = 1, -ni- = 2, Ø = 3. yi- is unglossed on that page. cha ‘cry’ is Hale's example of an unergative, with its argument in Spec,v (Hale 2003 ex. (23)).

  1. Σ Σ(√CHA, v) = {ρ₁}. Eρ₁: experiencer role proposed; crying does not entail volition. This role assignment is our HYPOTHESIS. posρ₁ = Spec,v (unergative; Hale 2003 (23)).
  2. Merge V = cha.
  3. Merge v (Voice/classifier) with VP. A v⌢VP.
  4. Merge subject DP (ρ₁) in left Spec,vP with interpretable [person:1] in this first-person example. The agreement probe, not that person value, starts unvalued.
  5. Merge M/S (imperfective) with vP. A M/S⌢vP.
  6. Agree M/S ⟶ ρ₁. The probe on M/S copies ρ₁'s person value [1]. The probe receives [1] from the argument; -sh- realizes agreement. An exponent is not the source that values the argument’s semantic person.
  7. Move (V-P only): ρ₁ goes Spec,v → Spec,M/SP, leaving ⟨ρ₁⟩. Spec-to-Spec, Grant's "given". HYPOTHESIS The higher copy is silent. The only overt reflex of the argument is the φ-exponent on M/S, which Agree provides with or without movement.
  8. Spell-out A Retain the cited source form above, without inserting a classifier into its transcription. The earlier constructed string with an extra Ø is withdrawn; a silent classifier was only our assumption. This site does not assign yi- to a template position.
  9. B No independent overt DPs. The hypothetical subject DP is already present in the syntax; “no overt noun” does not mean “no argument”.

Result. D1 cannot tell V-H from V-P: the movement has no exponent. OPEN Position and category of yi- in this paradigm (Young & Morgan's paradigm would settle it; not opened).

HYPOTHESIS (LF check) ∃e[cry(e) ∧ Experiencer(e,speaker)]. Volition is not entailed. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D2. Transitive, local persons (no voice alternation)

Hale 2003 (1)
ni-sh-hozh
2s-1s-tickle

‘I tickle you.’

Hale 2003 ex. (6) analyses it as Obj ni / Mode Ø / Subj sh / Classifier ł / Stem ghozh; ex. (7) bracketing [Obj [Mode [Subj [Cl [Stem]]]]]. Ex. (2) Ni ni-sh-hozh ‘I tickle YOU’ adds an overt contrastive pronoun.

  1. Σ Σ(√GHOZH, ł) = {ρA, ρP}. posρA = Spec,v, posρP = Comp,V (Hale 2003 tree (8)). EρA ⊇ {vol, caus}, EρP ⊇ {aff} (HYPOTHESIS).
  2. Merge V ghozh with ρP. A lexical V is complement-final: ρP⌢V.
  3. Merge v = ł. A v⌢VP. Merge ρA in Spec,v.
  4. Merge M/S (Mode Ø). Agree M/S ⟶ ρA: [1sg], exponent sh.
  5. Σ ρP valued [2sg], exponent ni. C1 holds: two slots, two exponents. C4: R = ∅ (no third persons), so VOICE is undefined and there is no yi-/bi-.
  6. Move (V-P): ρA goes Spec,v → Spec,M/SP (silent). ρP goes Comp,V → outer Spec,M/SP: the second movement type, crossing ρA. (V-H: no movement. ni, a complement, is placed leftmost by skeleton filling, Hale 2003:33–36.)
  7. Spell-out A → ni-Ø-sh-ł-ghozh, surfacing as nishhozh. The ł + gh → h fusion is Hale's phonology, not derived here.
  8. B Optional overt pronoun Ni adjoined at the left edge for contrast (ex. (2)). It links to ρP and does not value it again; it is an adjunct.

Earlier slot diagram withdrawn (audit pass 1). Its M/SP/ρ labels did not distinguish categories from role indices. The generic transitive vP/TP figure supplies the core only. Extra object raising and its LF/scope are OPEN, not inferred from prefix order.

Result. Rule set A derives Mode–Subj–Cl–Stem with no movement of heads. Placing the object prefix first needs either Hale's phonological skeleton filling or V-P's object raising. V-P makes the order Obj > Subj a derived c-command relation. That conflicts with Rice's scope generalization, under which subject scopes over object (headedness (iii)). OPEN A scope or binding diagnostic for Obj vs Subj inside the word is exactly Grant's open question 6.

HYPOTHESIS (LF check) ∃e[tickle(e) ∧ Agent(e,speaker) ∧ Theme(e,addressee)]. Extra object raising cannot change the theta-role map. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D3. Transitive, two third persons, direct (yi-)

Willie & Jelinek 2000:273, (61a)
łį́į́ʼtséyiztał
horserock3ACC-3NOM-kicked (Direct)

‘The horse kicked the/a rock.’

  1. Σ Σ(√KICK) = {ρA (Spec,v), ρP (Comp,V)}. EρA ⊇ {vol, caus, mov}, EρP ⊇ {aff, stat} (HYPOTHESIS).
  2. Merge V, ρP; v; ρA; M/S, as in D2, steps 2–4. A as before.
  3. Σ π(ρA) = π(ρP) = 3. R = {ρA, ρP}, |R| = 2, so VOICE is defined (C4).
  4. B Linking. DP łį́į́ʼ ‘horse’ links to ρA, DP tsé ‘rock’ to ρP. TOP = the first DP = horse = ρA.
  5. Σ VOICE(Σ) = yi-, since TOP = ρA. C6 is satisfied: horse (Creamer group 2) outranks rock (group 7).
  6. Move (V-P): ρA Spec-to-Spec to Spec,M/SP, as "external Nominative" (Willie 1991:123's term for the direct-voice agent). ρP: second-type movement, as in D2.
  7. Spell-out A one verb word, yiztał. Segmentation not attempted: the source glosses it as a whole, 3ACC-3NOM-kicked.
  8. B DPTOP ⌢ DP ⌢ [verb word]. Under (G) both DPs are adjuncts, base-generated at the left edge and coindexed with their slots. Lexical-DP alternative (not Grant's, not Hale's): both DPs first-merge in ρ positions and A-move. The subject reaches Spec,TP/M/SP, giving SOV with no inverse logic.

HYPOTHESIS (LF check) ∃e[kick(e) ∧ Agent(e,horse) ∧ Theme(e,rock)]. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D4. Transitive, two third persons, inverse (bi-)

Willie & Jelinek 2000:273, (62)
a. ?*łį́į́ʼʼashkiiyiztał
horseboy3ACC-3NOM-kicked (Direct)
[‘the horse kicked the/a boy’]
b.ʼashkiiłį́į́ʼbiztał
boyhorse3ACC-3NOM-kicked (Inverse)

‘The boy was kicked by the/a horse.’

Willie 1991:122, (20)
a.ʼashkiiłééʼchąąʼíyishxash‘The boy bit the dog.’
boydog3.3.bit
b. *łééʼchąąʼíʼashkiiyishxash(The dog bit the boy.)
c.ʼashkiiłééʼchąąʼíbishxash‘The boy was bitten by the dog.’
boydogINV3.3.bit

Typescript łééʼcha̧a̧ʼí, with cedillas for nasal hooks. (20b)'s parenthesized translation is Willie's.

  1. Σ Same Σ as D3. C2: same Eρ. The horse is still the kicker; only linking changes.
  2. Merge as in D3, steps 2–3. R = {ρA, ρP}.
  3. B DP ʼashkii ‘boy’ links to ρP and is TOP; DP łį́į́ʼ ‘horse’ links to ρA.
  4. Σ VOICE(Σ) = bi-, since TOP = ρP. C6: boy (group 1) > horse (group 2) ✓. C7: the topic ’ashkii is definite (W&J 2000:271 requires this). C8: one bi- ✓.
  5. Move (V-P) HYPOTHESIS: the older inverse hypothesis raises ρP from V’s complement, not from a specifier. Calling that “Spec-to-Spec” was incorrect. Any intermediate outer Spec,vP, final Spec,TP and relation to the primary agent-raising analysis remain OPEN. ρA stays low, as an "internal (Ergative) agent, as we see in the Inverse" (Willie 1991:123). Tension: W&J gloss both (62a) and (62b) as 3ACC-3NOM, which keeps the case labels constant across voice. Willie's "internal (Ergative)" agent says the inverse changes the agent's position. This is a brief source-theory contrast, not the derivation our analysis must reproduce; the actual movement path is unresolved.
  6. Spell-out A biztał, unsegmented.
  7. B DPTOP(boy) ⌢ DP(horse) ⌢ V.

Crashes. (62a) ?*: TOP = horse = ρA gives VOICE = yi-, but rank(boy) > rank(horse) violates C6. The gradient mark matches C6 being a preference that speakers apply to different degrees (W&J p. 273). (20b) *: TOP = dog = ρA gives yi-, with rank(boy) > rank(dog), so C6 is violated. Willie: "When a lower-ranked agent acts upon a higher-ranked patient, and all arguments are third person, then the Inverse voice must be used" (p. 122).

Update 2026-10-08: D3/D4 fix VOICE from the topic DP (steps 4–5 and 3–4). The inverse page, §5 moves that choice earlier, to Σ (the external slot is the higher-ranked one), because bare yiztał/biztał already differ in meaning. It also re-derives the inverse as I1–I5 on Willie's and Sandoval & Jelinek's pairs. The strings and judgments here are unchanged.

HYPOTHESIS (LF check) ∃e[kick(e) ∧ Agent(e,horse) ∧ Theme(e,boy)] for (62b); inverse order does not exchange these roles. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D5. Fourth-person object: leaving the ranking

Willie 1991:123, (21)–(22)
(21)łééʼchąąʼíshishxash‘The dog bit me.’
dog1.3.bit
(22)łééʼchąąʼíhashxash‘The dog bit that person.’
dog4.3.bit
  1. Merge prerequisite (HYPOTHESIS): build VP with the theme DP; merge v⁰ and the agent DP in left Spec,vP; merge T⁰ (the older M/S≈T identification remains OPEN) to create the Spec,TP target. Only after this can the agent raise or V head-move through v to T.
  2. Σ Σ(√BITE) = {ρA, ρP}, with E as in D4 (Willie 20) unchanged (C2).
  3. Σ π-valuation first. ρP valued [4] (in (21), [1]). By C9, ρP ∉ R. R = {ρA}, |R| = 1, so VOICE is undefined (C4).
  4. B DP łééʼchąąʼí ‘dog’ links to ρA. No DP for ρP: the fourth person "does not ordinarily cooccur with an NP" (W&J 2000:277).
  5. Move (V-P): ρA Spec-to-Spec to Spec,M/SP. Willie: "the lower ranked third person agent is the subject, the external Nominative argument" (p. 123). The direct-voice configuration survives even though rank(human) > rank(dog).
  6. Spell-out A hashxash, unsegmented. Willie names the 4th-person object ho-; the ha- shape is Young & Morgan's phonology, not derived here.

Result. HYPOTHESIS Ordering matters. If VOICE were computed before ρP's π-value was fixed, a human patient with a dog agent would force bi-, as in (20c). Fourth-person valuation bleeds the inverse. So in the schema, Σ-valuation precedes the step that fixes voice. The same holds for (21), with a 1st-person patient.

Update 2026-10-08 evening: the fourth-person page restates this as KEEP-DIR / EXIT, compares fourth with Grant's ranking edge (escape hatch, not a second inverse), and derives F1–F5 (baseline inverse, fourth-patient rescue, indefinite agent, fourth agent / two fourths, W&J coreference).

HYPOTHESIS (LF check) ∃e[bite(e) ∧ Agent(e,dog) ∧ Theme(e,h)] for (22), where h is the contextually intended fourth-person referent. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D6. Polar question

NLA Navajo Questions, yes/no section
(1)Dibéneehólǫ́.‘You have sheep.’
sheep2-withexist.NI
(2)Daʼdibéneehólǫ́?‘Do you have sheep?’
Qsheep2-withexist.NI
(3a–c)*Dibé daʼ nee hólǫ́? · *Dibé nee daʼ hólǫ́? · *Dibé nee hólǫ́ daʼ?
(4)Dibééshneehólǫ́?‘Do you have sheep?’
sheep-Q2-withexist.NI
(15)Dibéneehólǫ́, yaʼ?‘You have sheep, don’t you?’
sheep2-with3-exist.NI right

Fernald, Chavez et al., NLA, version 3.2022, exx. (1)–(4), (15). Glosses as printed: (15) has 3-exist.NI where (1) has exist.NI. "It is ungrammatical for da’ to appear in any other position."

  1. Σ Verb hólǫ́ ‘exist’ with one slot ρth, for what exists (HYPOTHESIS; the source gives no argument analysis). The possessor is not in Σ of the verb. It sits inside the postpositional word nee ‘2-with’.
  2. A/lexical P: inside nee, the person prefix precedes P ("2-with"). That is lexical-head-final, like V (Hale 2003:32: postpositional projections take their argument on the left). Not segmented further.
  3. B DP dibé links to ρth. PP nee is adjoined. Order: DP ⌢ PP ⌢ V.
  4. Merge C⁰[Q] above a built TP HYPOTHESIS. The three following surface strategies are sourced; their analysis as exponents of the same C head is not:
    • (2): da’ at the left edge only (3a–c *).
    • (4): enclitic -ísh on the first phrase, neutral. On a later phrase it focuses that phrase (yes/no (10)–(11)). The cited simple-clause examples reject repetition (8) and hosting on da’ itself (13). This is not a universal one-per-question theorem; coordination has its own restrictions.
    • (15): the tag ya’ at the right edge.
  5. B Linearize C. HYPOTHESIS If da’ is C⁰, then C is head-initial, contra (i′). If da’ is an operator in Spec,CP, its initial position says nothing about head direction, and C⁰ could be the null or final element (ya’ in tags). A uniform-final clause predicts the second analysis only if that head/specifier assignment is independently supported; it is not a consequence of VPISH.

Result. OPEN D6 is the cleanest test case for rule set B, but it depends on a category decision (head vs specifier) that the sources do not make. Diagnostics to look for in texts: does da’ co-occur with an overt complementizer-like element? Can it head embedded questions?

HYPOTHESIS (LF check) ASK_POL(∃s∃x[sheep(x) ∧ HAVE(s,you,x)]) for neutral (2)/(4). Tag bias in (15) is not derived. Q scopes over the proposition; Q’s pronounced host does not fix C’s position. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D7. Wh-question

Willie & Jelinek 2000:272–273, (57)–(60)
(57a)háí-shyiztsʼǫs‘Who did he kiss?’
who-Q3ACC-3NOM-kissed (Direct)
(57b)háí-shbiztsʼǫs‘Who kissed him?’
who-Q3ACC-3NOM-kissed (Inverse)
(58a)Jáanháí-shyiztsʼǫs‘As for John, he kissed who?’
(58b)háí-shJáanyiztsʼǫs1) ‘Who was it, that kissed John?’ Or: 2) ‘Who was it, that John kissed?’
who-QJohn3ACC-3NOM-kissed (Direct)
(59)háí-shJáanbiztsʼǫs‘Which one (of you) was it, that was kissed by John?’
(60a)Jáanháí-shbiztsʼǫs‘As for John, he was kissed by whom?’
(60b)háí-shJáanbiztsʼǫs1) ‘Which one (of you) was kissed by John?’ 2) *‘John was kissed by whom/which one?’

Page images checked this run (printed pp. 272–273). W&J: "There is no obligatory Wh-movement in Navajo, but there may be focus (leftward) movement of the Wh-word. The particle -sh encliticized to Wh-words has inherent focus, and usually occurs after the first word in the sentence." "The second reading of (58b) shows optional focus movement of the Wh-word + Q to the leftmost position." A wh-word coindexed with a bi-pronoun gets a "D-linked" "which one" reading. "The inverse does not permit optional focus movement of the Wh-word." "The bi-pronoun marks a topic/focus inversion that is incompatible with focus movement of the Wh-word."

NLA Navajo Questions, content (1)
Maryhaʼátʼíínayiisniiʼ?
Marywhat3-3-buy.P

‘What did Mary buy?’ (wh-word in its linked position after the topic; no fronting)

Derivation of (58b), both readings.

  1. Σ Σ(√KISS) = {ρA, ρP}, both π = 3, so R is defined. VOICE = yi- (direct).
  2. B Linking of two adjuncts, wh-DP háí and DP Jáan. Base order: TOP ⌢ other ⌢ V. Under yi-, TOP = ρA.
  3. Reading 1: háí links to ρA and is in the TOP position in situ; Jáan links to ρP. No movement. ‘Who kissed John?’
  4. Reading 2: Jáan links to ρA (base TOP), and háí links to ρP, base position after Jáan (cf. (58a)). A′-Move: focus movement of [háí-sh] to the left edge, ⟨háí-sh⟩ left behind: háí-sh Jáan ⟨háí-sh⟩ yiztsʼǫs. ‘Who did John kiss?’ HYPOTHESIS This is a candidate for Grant's "another type of movement possible for arguments beyond the first". It is A′, optional, and available only to the non-topic argument.

Derivation of (60b): why reading 2 crashes.

  1. Σ VOICE = bi-, so TOP = ρP (§4 of the schema page).
  2. Reading 1: háí links to ρP = TOP, in situ, D-linked (‘which one’), consistent with C7 (the inverse topic is definite/presupposed). Jáan links to ρA. ✓
  3. Reading 2 would need Jáan = TOP = ρP with háí (ρA) focus-moved over it. HYPOTHESIS Historical candidate filter: reject this particular inverse reading/host configuration. A blanket ban on every inverse focus movement is not established. W&J’s analysis and Willie’s conflicting cases must be separated (see the dated note below and the matched host tests). ✗

Result. HYPOTHESIS A serial schema can order valuation, inverse choice and host/focus licensing. A parallel compatibility filter can reject the same configuration. The data do not uniquely prove that serial timing or a general V-movement block. The NLA (1) order Mary haʼátʼíí nayiisniiʼ is the in-situ base (topic ⌢ wh ⌢ V). OPEN Multiple wh-questions: see open questions.

Note added 2026-10-13: Willie 1991:207–208 says focus movement of a question word “is not possible when the agent and patient are of equal rank”, and gives háísh John yiyiiłtsą́ only the reading ‘Who saw John?’. That conflicts with reading 2 of W&J's (58b) above. Willie 1991 allows the fronting only when meaning rules out the other reading (‘What did John see?’). See the wh-questions page §6.3 (T19a).

HYPOTHESIS (LF check) (58b)-1: ASK_x:person(x)[∃e[kiss(e) ∧ Agent(e,x) ∧ Theme(e,John)]]. (58b)-2: ASK_x:person(x)[∃e[kiss(e) ∧ Agent(e,John) ∧ Theme(e,x)]]. (60b)-1 restricts x to a discourse set and retains Agent=John, Theme=x; starred (60b)-2 intends Agent=x, Theme=John. The source rejects that parse, not an incoherent theta-role map. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

D8. Causatives: simple alternation vs complex causative

Hale & Platero 1996:4, via Gessner 2003 ex. (6)
Tłʼóółkʼí-ní-dláád‘The rope broke.’
ropekʼí-NPF:3-break:PERF
Tłʼóółkʼí-i-ní-ł-dláád‘He broke the rope.’
ropekʼí-3-NPF:3-ł-break:PERF
Gessner 2003 ex. (10), citing Hale & Platero 1996:4
ʼAwééʼnaa-ghá‘The baby is walking around.’
babyna-IMPF:3-walk:sg:CI†
ʼAwééʼna-b-ii-sh-ł-á‘I am walking the baby around (i.e. making it walk).’
babyna-3-y-IMPF:1s-ł-walk:sg:CI

SOURCED † Segmentation note (2026-11-11 audit pass 4): Hale & Platero 1996 prints naa-ghá with the gloss “na-IMPF:3-walk:sg:CI”: the gloss names more morphemes than the form shows as separate pieces (fused or unwritten morphemes), so the two lines do not match one-to-one in the source. Kept exactly as printed. The mode is glossed IMPF (imperfective) as in Hale & Platero 1996; this site had printed IMP, which in Leipzig glossing means imperative. Corrected in audit pass 4.

Gessner 2003 ex. (11), citing Hale & Platero 1996:5
ʼAwééʼd-ee-zaʼ‘The baby burped.’
babyd-SPF:3-belch:PERF
ʼAwééʼbi-di-y-é-saʼ (< … -ł-zaʼ)‘I burped the baby.’
baby3-d-y-SPF:1s-ł-belch:PERF

Page images of Gessner 2003 checked this run. Gessner: with 1st/2nd-person subjects the causee object is overt bi-, and y- is required. Abbreviations are Gessner's (NPF, SPF = perfective mode types; IMP = imperfective; CI = classificatory).

Simple (rope): ADD-CAUSER case (a).

  1. Σ Base Σ = {ρ₁}, pos = Spec,V (labile/unaccusative; Hale 2003 (22)). Eρ₁ ⊇ {chg}.
  2. Merge V dláád ‘break’, ρ₁ in Spec,V; v = Ø (intransitive) or v = ł + ρC in Spec,v (transitive).
  3. Σ Transitive: Σ' = {ρC, ρ₁}. ρ₁ is now the object, valued 3 with exponent -i- per Gessner's segmentation, glossed "3".
  4. Spell-out A preverb kʼí (disjunct) ⌢ [object] ⌢ M/S ní ⌢ v ł ⌢ V. H&P: ł is "nothing more than … a simple marking of the transitivity alternation".

Complex (walk, burp): ADD-CAUSER case (b).

  1. Σ Base Σ = {ρ₁}, pos = Spec,v (legacy unergative classification: walking supplies a mover, but burping does not entail volition; the exact theta-role/structure assignment is HYPOTHESIS).
  2. Merge inner vP₂: v₂ + V, ρ₁ in Spec,v₂.
  3. Merge Qual y (Hale 2003 exx. (17b), (18c): Qualifier y; Gessner puts it in position 6c).
  4. Merge outer v₁ = ł with ρC in Spec,v₁. Then M/S, with Agree M/S ⟶ ρC [1sg] = sh (walk) / 1s fused in é (burp; SPF:1s per Gessner's gloss).
  5. Σ ρ₁ relinked to Obl: exponent overt b(i)- even with a 1st-person subject. That is "as if attached to an incorporated postposition" (H&P). Ordinary position-IV 3rd-person objects would be null here.
  6. Spell-out A na (preverb) ⌢ b (3, oblique causee) ⌢ ii (glossed y) ⌢ sh (IMPF:1s) ⌢ ł ⌢ á (walk:sg:CI), Gessner's segmentation. Qual y sits left of M/S, as rule set A (head-initial Qual⌢M/S) requires.

Crash. Case (a) applied to an unergative base, ł alone, is impossible according to H&P's (13) (*ʼAwééʼ d-é-saʼ). Re-checked 2026-10-07 (evening) against the text of H&P 1996 itself (vdoc.pub copy, characters restored; no page image available). Form and gloss match: *ʼAwééʼ d-é-saʼ (< … -ł-zaʼ), baby d-SPF:1s-ł-belch, ‘I burped the baby.’ A full step-by-step crash is derivation C4. HYPOTHESIS In schema terms, ADD-CAUSER(a) requires posρ₁ = Spec,V, so that v's specifier is free for ρC. With an unergative, Spec,v is occupied, so a second v layer (vP embedding) is needed, and the displaced slot can only be valued obliquely.

See also derivations C1–C4 on the plain ł-causer page: the simple alternation and the complex causative from H&P 1996's own examples, with H&P's raising-to-Spec,I version of VPISH shown beside in-situ licensing.

See also derivations M1–M3 on the keep vs bring about page: posture causatives, including a causer who holds a state rather than bringing about a change.

HYPOTHESIS (LF check) Rope intransitive: ∃e[break(e) ∧ Theme(e,rope)]. Transitive: ∃e[break(e) ∧ Causer(e,a) ∧ Theme(e,rope)]. Complex burp: ∃e∃e′[CAUSE(e,a,e′) ∧ burp(e′) ∧ Experiencer(e′,baby)]. This semantic hypothesis entails no volition or aspectual endpoint; a complete subject-base derivation of the old Spec,V analysis remains OPEN. Event/state closure is below Q where present. These LFs constrain a future full derivation; they do not certify every old step list. A new Move requires an already merged target.

What the derivations show HYPOTHESIS

Point 4 below (“rule set A does no work on DPs, rule set B none on prefixes”) is sharpened on A/B as spell-out §3: noninterleaving follows only within that proposed cycle model, not independently from the data.

  1. The added subject-raising step has no independently isolated overt reflex in these word forms. VPISH names the base-position hypothesis, not the raising operation. Subject φ is realized on M/S by Agree whether or not the argument moves (D1–D5). The movement Grant takes as given is therefore neither confirmed nor refuted by morphology. Its evidence would have to come from scope and reconstruction.
  2. The "second type" of movement has two distinct candidates. (a) Object raising over the subject inside the word (V-P, D2), which Hale considered and rejected. (b) Optional A′ focus movement at clause level (D7), for which the sources report specific readings and restrictions. Those observations do not establish a single landing site or a blanket inverse movement ban.
  3. A serial order is one hypothesis, not forced: φ-valuation of slots, then voice, then A′ movement (D5, D7). Matched scope/reconstruction judgments must separate that order from parallel filtering; the construction of these lists does not do so.
  4. Rule set A does no work on DPs at all, and rule set B does no work on prefixes. The lists impose that division. This is not independent evidence for B2, and B1/B2/B3 remain OPEN; see the boundary tests.

The spine: how every Navajo clause is built (Merge, theta roles, V→v→T, and what else moves)

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

Verbs on this page → dictionary: yiztał ‘kick’ · nishhozh ‘tickle’ · naa-ghá ‘walk around (one)’ · yi-sh-cha ‘cry’

2026-11-14 overhaul (approved by Grant for the live site, 2026-11-14). This page is rebuilt from Grant’s premises of 2026-11-13. VPISH lines up syntax and semantics: the arguments start inside the verb’s projections, where they get their roles. It is not an account of word order. In ordinary tensed clauses the verb moves V→v→T and the subject moves from Spec,vP to Spec,TP (on to Spec,CP or C when needed); that is near-universal and is assumed. So the question for every Navajo pattern is what else moves, and why. The older Navajo literature supplies the data; its analyses appear only as labelled contrast (§S6). The earlier text of this page, with all audit notes, is kept unchanged under History. No Navajo form, gloss, judgment or citation changed.

This page sets out the one clause structure, the spine, that every derivation on the site is built on. It says what is merged where, where each theta role is assigned, which movements are assumed in every tensed clause, how the verb word is pronounced, how logical forms are written, and what counts as a test. The findings of the first-principles reviews, restated on this spine, are on Results on the spine; every phenomenon placed on one structure for the clause, the noun phrase and the verb word is on One structure for Navajo (2026-11-15).

On this page: The spine in brief · S1 Premises · S2 Step by step · S3 Logical forms · S4 What else moves · S5 What counts as a test · S6 Older analyses as contrast · S7 Results · History

The spine in brief

A Navajo sentence is built from the bottom up. The verb is merged first with the participant it directly affects, which gets its role there (Theme, Patient). A light verb v is merged on top, and the participant that acts or experiences is merged in its specifier, Spec,vP, and gets its role there (Agent, Experiencer). This is VPISH: every argument starts inside the verb’s own projections, so the meaning can read each role off the place where the argument was merged. Then tense, T, is merged. In a tensed clause the verb moves up through v into T, and the subject moves up from Spec,vP into Spec,TP. If the sentence is negated, a negation head is merged above TP; if it is a question, a C head with Q is merged above that. The verb word is the V+v+T complex, pronounced where it ends up. Up to here nothing is special to Navajo. What is special to Navajo is what else has to move to give its orders: the verb last, the patient before the agent in the inverse, a phrase before doo that is still negated. Each of those extra steps is a hypothesis that has to be argued from sourced examples.

[CP [C′ [NegP [Neg′ [TP DP_subj [T′ [vP ⟨DP_subj⟩ [v′ [VP [V′ DP_obj ⟨V⟩]] ⟨v⟩]] V+v+T]] Neg⁰]] C⁰]]
Specifiers on the left. NegP and CP only when the clause is negated or a question.
Angle brackets = silent lower copies. Drawn here with T⁰, Neg⁰ and C⁰ final (HS-1); see S4.

S1. Premises

Only these are assumed. Everything else on the site, including every Navajo-specific rule, is a hypothesis to be tested.

  1. Theta roles (assumed). Each argument gets exactly one theta role, from the head it is merged with, and each role goes to exactly one argument.
  2. VPISH (assumed). The internal arguments are merged inside VP and the external argument in Spec,vP, so every argument starts within the scope of its verb and gets its role there. This lines up syntax and semantics. It is not a word-order claim, and VPISH is never a node label.
  3. The working default in tensed clauses (assumed; Grant, 2026-11-13). V⁰ moves to v⁰ and on to T⁰ (head movement, no head skipped), and the subject moves from Spec,vP to Spec,TP. When a question, focus or topic needs it, a phrase moves on to Spec,CP or a head to C. An example that seems to resist this is flagged OPEN, not smoothed over.
  4. Syntax matches logic (assumed). Every derivation ends in an explicit logical form, and an operator’s scope is what it c-commands where it is interpreted. A reading that needs different scope needs a different structure, or an explicitly stated extra step (reconstruction, covert movement).
  5. Specifiers on the left, real X-bar labels (CP, NegP, TP, vP, VP, DP, PP). Head direction is not assumed (S4).
  6. Two negations. Sentential negation (doo … da) and constituent negation (hanii) are kept apart unless data show they are one thing.

HYPOTHESIS Everything beyond S1 is labelled where it is used: the direction of each head; the inverse Move; NegP’s internal structure; Q placement; covert scope; the internal order of the verb word; any Navajo-particular rule. Light d is no longer among them: Grant withdrew it on 2026-11-14 (below). Telicity is OPEN unless a minimal pair decides it.

S2. Building a clause, step by step

StepOperationWhat it doesStatus
1MergeMerge V⁰ with its internal argument(s) to form VP. V assigns their roles: Theme or Patient to its complement; a Goal or a postposition’s object where the verb selects one (in a ditransitive, which internal argument is higher is a hypothesis, H3-loc).assumed (VPISH, theta)
2MergeMerge v⁰ with VP; merge the external argument in left Spec,vP. v assigns its role (Agent, Causer, Experiencer). In a passive this argument is silent and indefinite (H10b/H16a, a hypothesis). A causative adds a further v with its own causer (hypothesis).assumed (VPISH, theta)
3MergeMerge T⁰ with vP. Which prefix, if any, spells out T is OPEN.assumed
4Head-moveV⁰→v⁰→T⁰, after both targets exist. Any head merged between v and T (for example the distributive head of H21a) is passed through, not skipped.working default
5MoveThe subject moves from Spec,vP to left Spec,TP; the lower copy is silent. It keeps the role it got in Spec,vP; the Move assigns none.working default
6MergeSentential negation only: merge Neg⁰ above TP. On present evidence Neg⁰ is the final da, and doo is a lower marker adjoined inside its complement (H23). Constituent negation hanii is not merged here; it attaches to the phrase it negates (H20).HYPOTHESIS
7MergeQuestions and other clause types only: merge C⁰ above NegP or TP. Polar Q is merged at the clause edge (H8 Q-edge); a question word’s Q may be displaced to it (H12).HYPOTHESIS
8MoveOnly when required: a phrase moves to Spec,CP (a fronted question word, H14; a host for =sh). Not obligatory in every question: question words can stay where they are merged.working default when triggered; each trigger HYPOTHESIS
9Spell-outThe verb word is the V+v+T complex, pronounced at T with its pronominal and other prefixes. Which exponent spells out which head, and why the prefixes come before the stem, is OPEN (see S4, last paragraph). Enclitics (-ísh, =sh) are written on a host at PF; the host is not their scope position.OPEN
10InterpretLF: read each role at the argument’s first-merge position and each operator’s scope at its interpreted position. Generic negative yes/no question: ASK_POL(¬∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)]).assumed (S3)
Analysis A: Working default with T⁰ final (HS-1): nothing else has to move for verb-final order Hypothesis
Analysis A: Working default with T⁰ final (HS-1): nothing else has to move for verb-final orderThe theme merges with V⁰ and gets its role there; the agent merges in left Spec,vP and gets its role from v⁰ (VPISH: arguments start inside the verb's projections, so roles are read off first-merge positions). V→v→T and subject Spec,vP→Spec,TP are the working default in tensed clauses: V⁰ head-moves through v⁰ to T⁰, and the agent raises to left Spec,TP, leaving a silent lower copy. In this panel T⁰, Neg⁰ and C⁰ are all final (HS-1, HYPOTHESIS), so the raised verb is pronounced after everything in TP and before the final negation and question heads; nothing else has to move to give verb-final order. The lower negative marker is adjoined to T′ (H23; its exact site is OPEN). Scope order = c-command: C⁰[Q] over Neg⁰, Neg⁰ over the whole TP, so the question asks about the negative proposition. One negation, from Neg⁰. LF: ASK_POL(¬∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)]) Full structure in the list below the figure.CPC′NegPC⁰[Q, polar]Neg′TPNeg⁰DPagent aT′AdvPlower negative markerT′vPT⁰DPagent av′VPv⁰V′DPtheme tV⁰predicateverb wordhead-movehead-moveAgree φ (optional)Move
Analysis B: Working default with T⁰ initial (HS-2): the remnant VP must also move above T (HYPOTHESIS) Hypothesis
Analysis B: Working default with T⁰ initial (HS-2): the remnant VP must also move above T (HYPOTHESIS)Same Merge steps, roles and working default as panel A: V→v→T and subject Spec,vP→Spec,TP. Here T⁰ and v⁰ are initial (HS-2, HYPOTHESIS), so the raised verb in T⁰ would be pronounced before the object left in VP. To give the printed verb-final order, something else has to move: the remnant VP (theme, low PPs and adverbs, without the verb, which has already left) left-adjoins to T′. That extra movement has no independent trigger yet and is HYPOTHESIS; the lower negative marker is omitted here. Neg⁰ and C⁰ stay final, so this panel mixes head directions. Scope order = c-command, as in panel A: the moved VP is interpreted at its lower copy for theta purposes, and it stays under Neg⁰ in either position. LF: ASK_POL(¬∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)]) Full structure in the list below the figure.CPC′NegPC⁰[Q, polar]Neg′TPNeg⁰DPagent aT′VPT′T⁰vPDPagent av′v⁰VPV′DPtheme tV⁰predicateverb wordhead-movehead-moveAgree φ (optional)MoveMove
(1) Schematic (placeholder leaves, no Navajo example): the whole clause spine with every optional layer present Hypothesis 2026-11-14 overhaul draft (Grant, 2026-11-13: VPISH aligns syntax and semantics; V→v→T and subject raising are assumed in tensed clauses). All leaves are generic placeholders; no Navajo form. The layers drawn are the ones the first-principles reviews needed: vP with the external argument in Spec,vP, TP, NegP with final Neg⁰ (H23, review 8) and CP with a polar question feature (reviews 4 and 8). Head directions of T⁰, Neg⁰ and C⁰ are hypotheses; see /headedness/#spine-2026-11-14.
Bracket notation (copy)

A: [CP [C′ [NegP [Neg′ [TP [DP agent a] [T′ [AdvP lower negative marker] [T′ [vP ⟨DP⟩ [v′ [VP [V′ [DP theme t] [V⁰ predicate]]] v⁰]] T⁰]]] Neg⁰]] C⁰]] B: [CP [C′ [NegP [Neg′ [TP [DP agent a] [T′ VP [T′ T⁰ [vP ⟨DP⟩ [v′ v⁰ ⟨VP⟩]]]]] Neg⁰]] C⁰]]

Structure as a list

Analysis A

  • CP
    • C′
      • NegP
        • Neg′
          • TP
            • DP (id S-high; landing site of Move from subject): agent a
            • T′
              • AdvP (id negmark): lower negative marker
              • T′
                • vP
                  • DP (id subject; goal of Agree φ (optional); struck copy; moved to S-high): agent a
                  • v′
                    • VP (id lowVP)
                      • V′
                        • DP (id object): theme t
                        • V⁰ (head; id V; head-moves to v): predicate
                    • v⁰ (head; id v; head-moves to T)
                • T⁰ (head; id T; probe for Agree φ (optional) with subject; hosts the “verb word” (V + v + T))
          • Neg⁰ (head; id Neg)
      • C⁰ (head; id C; [Q, polar])

Analysis B

  • CP
    • C′
      • NegP
        • Neg′
          • TP
            • DP (id S-high; landing site of Move from subject): agent a
            • T′
              • VP (id VP-high; landing site of Move from lowVP)
              • T′
                • T⁰ (head; id T; probe for Agree φ (optional) with subject; hosts the “verb word” (V + v + T))
                • vP
                  • DP (id subject; goal of Agree φ (optional); struck copy; moved to S-high): agent a
                  • v′
                    • v⁰ (head; id v; head-moves to T)
                    • VP (id lowVP; struck copy; moved to VP-high)
                      • V′
                        • DP (id object): theme t
                        • V⁰ (head; id V; head-moves to v): predicate
          • Neg⁰ (head; id Neg)
      • C⁰ (head; id C; [Q, polar])

Figure: the spine with every optional layer present, drawn twice. Panel A has T⁰ final (HS-1), so nothing else moves. Panel B has T⁰ initial (HS-2), so the remnant VP must also move above T. Placeholder leaves only, no Navajo forms. Every step beyond S1 is a HYPOTHESIS.

S3. Logical forms

  • e ranges over eventualities, states included (s is used for states). Event closure ∃e applies at vP.
  • Roles are written Agent(e,x), Patient(e,y), Theme(e,y), Exp(e,x), Goal(e,z). Each argument bears one role; a result or path relation such as OUT counts as a role when its argument is an individual.
  • Sentential negation contributes one ¬, from Neg⁰, scoping over the event: ¬∃e[…]. A result with ∃e[… ∧ ¬P(e)] is a different meaning and does not stand in for it.
  • Questions: ASK_POL(p) asks whether p; ASK_WH(λx …) or ?x.… asks for values of x. Q above Neg gives ASK_POL(¬p).
  • Constituent negation: assertion ¬[focus = value] plus a presupposition for the rest of the clause.
  • Relatives and nominalized clauses: a linked enclitic gives ιx … (the individual); an unlinked one names the fact (H34).
  • Scope = c-command at the interpreted position. Reconstruction (reading a moved phrase at a lower copy), covert movement (H22a) and choice functions (H22b) are extra steps and must be stated. Source translations are SOURCED; role labels and LFs are HYPOTHESIS.

S4. What else moves

With the working default in place, the Navajo-particular claims are all claims about additional movement, or its absence. The site’s current hypotheses, each argued on Results on the spine:

PatternWhat else moves (HYPOTHESIS)Main evidenceStatus
Verb-final orderHS-1: nothing (T⁰ final). HS-2: the remnant VP, above T⁰ (T⁰ initial).Headedness; clause spineOPEN
Inverse (bi-)The patient moves above the raised agent (H3), only if definite or presuppositional (H6′)NX-029/030; NX-085 vs *NX-086HYPOTHESIS; trigger and landing OPEN
Fronted question wordMoves to Spec,CP when animacy keeps the roles apart (H14)Willie 1991:206–207HYPOTHESIS
Displaced =shQ moves to the clause edge (H12-move), after the inverse MoveNX-085 vs *NX-086HYPOTHESIS; H12-pf OPEN
Phrase before doo, still negatedMoves above doo but stays under Neg⁰ da (H23)NX-287, NX-288HYPOTHESIS
Missing argument with -í-The argument moves into Neg⁰’s domain; written on the verb at PF (H18a)NX-228 vs NX-229HYPOTHESIS
Wide-scope indefiniteCovert Move above the subject (H22a), or no Move and a choice function (H22b)NX-267OPEN between the two
Relative clauseNothing: D⁰ (=yę́ę, =ígíí) takes CP and is linked by Agree (H27b)NX-314; NX-312/313HYPOTHESIS
Question with “you think / she said”Nothing: two clauses side by side (H40a)NX-378, NX-384, NX-385HYPOTHESIS; H40b not excluded
Single argument of ʼádinMust reconstruct, or not raiseNX-292OPEN (resists the default)

GRANT'S HYPOTHESIS The verb word (restated 2026-11-15 on Grant’s correction). Once V has raised through v to T, the verb word is one complex head. Grant’s hypothesis is that headedness is fixed locally by Merge site, and that Navajo linearizes the sublexical domain opposite to the phrasal one: the clause is head-final, the word’s prefix heads are initial. The order of heads inside the word is then predicted to mirror their order in the clause (a higher head further left), which is the order the Mirror Principle gives for prefixes; sourced orders that go the other way are misfits for it, not confirmations (headedness fact-check, row 14). Whether every sourced prefix order fits is OPEN: the conjunct core fits; aspect, object/subject, distributive and preverb orders do not yet (unified structure §3.2; Grant’s two-tier linking proposal in §3.3, and the earlier, untested sublexical V-to-T in §3.4). The older position-class template (S6) describes the order; it does not derive it.

S5. What counts as a test

  • A test is a sourced form with its judgment and reading, copied with its source and example number, set against a hypothesis that predicts that judgment and that reading. Minimal or near-minimal pairs count most.
  • A hypothesis passes a test (“fits”, “consistent”), it is not “confirmed” by it. One starred example is weak evidence. Contradicting data count against a hypothesis; no older account is treated as a judge.
  • Not evidence: a translation alone (an English passive does not show a Navajo passive), a mode or viewpoint label (not a telicity test), a silent head proposed to fit the order, an enclitic’s written host (not its scope position), or a grid cell with no example (not a judgment of ungrammaticality).
  • Every review runs the question grid: polar or wh × affirmative or negative × active (direct), inverse and passive. Cells are filled only with sourced examples.
  • No Navajo form or judgment is ever constructed. Where a test needs a sentence no source prints, it is listed as OPEN, for a speaker or another source.

S6. Older analyses, as labelled contrast

The Navajo literature was written before current syntax had these tools, so its analyses are not fully informed. Its data (forms, glosses, judgments, page and example numbers) are used exactly as printed. Its analyses appear on this site only as contrast, never as the structure a derivation is built on.

AnalysisWhere fromWhat we keepOn the spine instead
Position classes and the verb template (“Pos. IV”, “Pos. V”)Young & Morgan 1987; Kari 1976The order of prefixes, as dataThe order is something the spell-out of V+v+T must produce (S4, OPEN)
Qual > M/S > v > V, skeleton filling, no head movementHale 2001, 2003Examples, page numbers, segmentations as printedV→v→T is assumed; “M/S” is not a node label
S-adjunction: nouns as adjuncts to S, “Maximal Verb Complex”, topic and focus adjunctsWillie 1991; Willie & Jelinek 2000All examples, indices and judgmentsNouns are linked to participants (H4″); orders come from Move (H3, H14)
Pronominal Argument HypothesisJelinek; Willie 1991Data on noun omissionWhether overt nouns are arguments or linked adjuncts is OPEN, not presupposed
NP-PRO constraint, parsing principlesPlatero 1973, via Speas 1986The one-noun restrictionsH4″, a stated default with a theta override
Noun–clitic chainsSpeas 1986(4.6)–(4.8)Roles assigned at first Merge (S1)
A/B rule sets; the headedness changeover at the word; CYCThis site, 2026-10 (from Grant’s earlier proposal)The sourced orders they were tested onHead direction of T⁰, Neg⁰, C⁰ plus spell-out of the verb word (headedness)
ID-, RD-, EH-, FB-rules; “roles, not words”This site, 2026-10 (thesis dropped 2026-11-02)Data and rule IDs, as a recordNot used
Light d (d takes a DP, as v takes VP)Grant’s proposalA record of the proposal and its testsWithdrawn by Grant, 2026-11-14. Never needed in reviews 1–13; the always-definite version is contradicted by NX-372

2026-11-14: light d withdrawn; light n an open question. Grant’s words (2026-11-14): light d “was just a proposal to account for facts … its nothing i’m deeply attached to … the existence of this unit was simply posited and i’m willing to unposit it.” Light d is therefore withdrawn, not a hypothesis still under test. In thirteen first-principles reviews it was never needed; the determiner-like elements that do work are all overt (results §13). The light-determiners essay and the dated reviews that test light d stay as a record.

OPEN Light n. Grant also noted that if DP and TP are built alike, there may be a light n in the noun phrase, parallel to light v in the clause. This is a structural question, separate from light d’s job of covering particular facts, and it is not a premise of the spine. No review so far has found data that bear on it (review 13 says so explicitly); a review raises it only when its data do.

S7. Results

The findings of first-principles reviews 1–13, restated on this spine with NX numbers and logical forms, are on Results on the spine: nouns and participants, the inverse, question words, sentential and constituent negation, questions with negation, quantifier scope, relative clauses, questions with “you think”, reported speech, =go clauses, passives, telicity, light d (withdrawn) and light n (OPEN), and the question grid. Open problems: what resists, and what is open.

History: the earlier versions of this page and the audit notes

Everything below is the page as it stood before the 2026-11-14 rebuild, kept unchanged as a record: the audit notes of passes 1–7 and the earlier sections 1–8. Where it differs from the spine above, the spine above supersedes it.

2026-11-11 audit pass 1. The earlier bracket tree mixed categories (vP/VoiceP, ApplP/PP), and light-d/Asp layers were built into the default. The core and both figures are redrawn with real X-bar categories, a left Spec,vP subject and explicit LF. GRANT’S CONJECTURE V→v→T and subject→Spec,TP are tested first; VPISH fixes only the base position, not either movement. Optional d/Appl/Asp and prefix realization stay OPEN. The old blanket objection to head movement depended on left adjunction and is corrected below.

2026-11-11 audit pass 3: syntax accuracy. All labels below are HYPOTHESIS unless marked. No Navajo form was added, changed or reconstructed in this pass.

Worst errors fixed.

  1. English particle-verb trees, panels A and C (phrasal verbs): the object sits in Spec,PP, not in the complement of throw, yet the LF gave it two theta roles (Theme of throw and the OUT role). It now has one (OUT); being thrown follows from Cause(e,s).
  2. Navajo prefix/phrasal and changeover trees, panels A and B (headedness, phrasal verbs): the directional AdvP is a VP adjunct but its LF OUT(e,t) gave the object a second role besides Theme. The LF is now OUT(e), an event modifier.
  3. Five working-note pages stated the A/B headedness changeover as a fact (“the word is the changeover point”, “the changeover … is visible in step 6”). Each sentence now says it is a hypothesis, not established.
  4. Three rule/derivation blocks written with Hale 2003’s M/S head (subject-head SH-7, pl-move-reason, deictic-order) were not marked as older notation; they are now labeled as a historical record, not our trees.
  5. The home page still framed the dropped “roles, not words” thesis as the flagship question, and its “Newest” item was stale. Both are fixed.

Counts by finding type. Theta-role errors in tree LFs: 6 panels in 3 trees. Changeover stated as fact: 7 sentences on 5 pages. Other overclaims (“confirmed”): 2. Dropped-thesis framing: 6 spots on 2 pages. Unmarked older notation: 3 blocks. “(VPISH)” step annotations: 3. Light-d pages given an explicit HYPOTHESIS note for dP: 10. Verdict-word notes on historical thesis pages: 3. Status badges brought into the label set: 59 on 20 pages. Local box file paths removed from public text: 105 on 37 pages. Undefined rule IDs: 0 (see below).

Checked and found sound. All 11 tree files: real categories only, binary branching, specifiers on the left, each head projecting its own category, subject first merged in Spec,vP, every panel with an LF. Every Move and head movement now passes a c-command test, every head movement passes the Head Movement Constraint (a skip must be flagged), and LF scope matches c-command for Neg over the event and for Q over Neg. In the questions trees, doo…da stays one clause-wrapping negation, consistent with review 4 (NX-203–217). V-to-T stays neither proven nor refuted; the changeover stays not established; light d is a hypothesis; telicity is OPEN.

Correction to the pass 2 report. Pass 2 listed about 65 “undefined” rule IDs (for example PL-1–4, OBJ-1–4, REFL-1–3). That was a false positive: those strings are Leipzig glosses (PL-1-mother, REFL-1.SBJ, INDF.OBJ-4.SBJ) or labels defined inline in a derivation. A stricter check finds 653 rule IDs referenced and 0 undefined.

New automatic checks. site/check_tree_categories.py now also enforces banned labels (VPISH, S, W, M/S, SC, Prt…), c-command of every movement, the HMC, LF scope order for Neg and Q, and one theta role per argument with every role filled; site/test_tree_categories.py has 9 new tests. Badges outside the label set now fail the build. Full report: site/audits/2026-11-11-pass3.md.

Deferred to pass 4. (1) 66 interlinear rows on 28 pages whose segmented form and gloss have different numbers of hyphens; a spot check shows they copy the source’s own finer gloss, so each needs its source page reopened before any change (list in the audit file). (2) Free translations are not yet machine-checked against glosses. (3) The theta check reads named roles only (Agent, Theme, Exp, Goal…); result and path relations such as OUT are checked by hand. (4) 450 older formal units in working notes remain OPEN until re-derived one by one. (5) Verdict words in running text on the historical thesis pages are explained by a note, not rewritten. (6) Forms damaged in the PDF (¬, ²) stay flagged and are not reconstructed.

2026-11-11 audit pass 4: syntax accuracy (gloss alignment, translations, theta, tree data)

Authorized by Grant (“continue to revise and check for syntax accuracy”). Same rules as passes 1–3: no Navajo form is invented or reconstructed; every change below either restores what a source prints or corrects our own gloss. Full list: site/audits/2026-11-11-pass4.md in the repository.

What was checked

  • Form/gloss hyphen alignment. Re-extracted every interlinear row: 64 rows on 28 pages had a different number of hyphens or clitic marks in the form and the gloss. Each was reopened in its source. 53 rows (22 distinct cells) match their source exactly — Willie 1991, Hale & Platero 1996, Willie & Jelinek 2000, and the Navajo Language Academy notes sometimes gloss more finely (3.3.neck:sever) or more coarsely (Kinłání=jįʼ “to”) than the hyphens. These are kept as printed and marked with a † and a SOURCED note under the table. 4 rows (Hale 1997, y-/b-ní-Ø-Ø-¬-ʼ²í) segment a damaged print we cannot check: marked † OPEN, nothing reconstructed. 6 essay rows were a builder bug (below). 1 row was our own gloss missing hyphens (below). A new checker, check_igt_alignment.py, now fails the build check on any unmarked mismatch: 0 remain.
  • Free translations against glosses (person/number, tense/aspect, negation scope, direct/inverse, passive): 625 rows screened, 9 flagged, all 9 confirmed against the source as correct (idioms such as doo ʼashtʼéʼégóó “extremely”, chʼééh “in vain”, Witherspoon’s “let” readings, the postposition-based alternation in possessor raising per Willie p.196). No translation changed.
  • Theta checker extended to relation predicates (OUT, IN, UP, DOWN, AWAY, OFF, ON, INTO, ONTO, AT, TO, FROM, BECOME, CAUSE, RESULT). A relation whose argument is an individual, not an event or state, now counts as a role. Regression test: the pass-3 “throw the rat out” LF (Theme(e,r) plus OUT(s,r)) is caught. A panel can declare a deliberate composite role as a hypothesis (“θ-composite: r”), which Ramchand & Svenonius’s complex-predicate panel B now does. 0 errors on the current 11 trees.
  • Trees against their cited NX entries. All 6 trees that cite NX entries (9 entries) were checked: every Navajo leaf is a printed token of the cited form, and subject/object roles in each tree and LF agree with the entry’s gloss and translation (e.g. NX-153 1-2-know = you know me; NX-086 inverse biztał = agent is the wh-word). 0 mismatches. A new automatic check makes any future non-cited leaf fail.

Worst five

  1. Essays dropped the first Navajo word of every quoted example. The essay builder removed a cell that had already been removed, so e.g. Witherspoon’s Hastiin łį́į́ʼ yiztał showed as “łį́į́ʼ yiztał” with the glosses shifted one column. Fixed on all 11 essays: events, obviation and others.
  2. IMP for IMPF. Hale & Platero 1996 gloss the mode of naa-ghá as IMPF (imperfective); we printed IMP, which reads as imperative. Fixed on derivations and walk (sg).
  3. Our own prefix gloss did not line up with the prefixes in the nishhozh table on Navajo verbs. Hyphens added to the gloss.
  4. Hale 1997 segmentation is ours, not Hale’s, and one sign is damaged in the print. Now marked OPEN on inverse agree and the inverse/fourth-person topic.
  5. The pass-3 theta bug class is now caught automatically (object getting a role from both the verb and OUT): note on the prefix-order phrasal-verb section.

Still OPEN

  • Hale 1997 y-/b-ní-Ø-Ø-¬-ʼ²í: segmentation and gloss need the original print.
  • Gessner 2003 (cited for the walk-sg rows via H&P) is not available to us; the IMPF fix follows H&P 1996 directly.

Deferred to pass 5

  • Translation screen used heuristics (person, negation, passive, inverse words); a full manual read of all 625 rows against sources is still due.
  • Trees without NX citations use placeholders only; when real data is added they should cite NX entries so the new leaf check applies.
  • Legacy pages’ prose glosses outside tables are not covered by the alignment checker.

2026-11-11 audit pass 5: translations read by hand, prose glosses, our own derivations, schematic trees

Authorized by Grant (“yes please keep correcting”). Same rules as passes 1–4: no Navajo form is invented, reconstructed or changed. Every change below either restores what a source prints or corrects our own wording. Full list: site/audits/2026-11-11-pass5.md in the repository.

What was checked

  • Translations read by hand. All 164 interlinear rows on the syntax and derivation pages (derivations, syntax, clause spine, inverse as Agree, A/B spell-out, headedness, interrupted verb complex, =sh vs -ísh, and the where-nouns-sit and prefix-order topics) were read one by one for who acts on whom, person and number, negation, tense/aspect and passive. Every row used in a tree or verb-movement test was also checked against its source: NX-085 and NX-086 (Willie 1991 p.211, page image), NX-094 (Hale 2001 p.678), NX-153 and NX-159 (Perkins & Fernald ch.8 pp.194–195), NX-205 and NX-213 (Navajo Questions database entries 804 and 2199). Doubtful rows were reopened in the source: Willie & Jelinek 2000 (49), (50), (57)–(62) on page images; Willie 1991 (1), (18)–(21), (54)–(57), (67)–(69); Hale 2001 (15) on the page image. No translation was wrong. Two cases that look odd are printed that way in the source and are kept as printed: (59) and (60b) are the same sentence with different readings in W&J p.272–273, and Willie’s (1) glosses yiicha as “2d.crying” under ‘We … are crying’ (already marked [sic?]).
  • Glosses outside tables and abbreviations. A count of IMP, IMPF, IPFV, PERF, PFV, OPT, PROG, FUT, ITER, 3i, 3o and 4 across the site found one real clash: our own “Leipzig” lines for Willie (24) and (26) used IMP (Leipzig imperative) for her impersonal. They now use IMPERS, the label the same page already uses for (94). Willie’s own category name IMP in prose, and Oxford’s Algonquian INV.IMP, are their authors’ labels and are kept. PERF and PFV both occur, each as its source prints it. 77 inline prose glosses of the shape form (gloss) were checked for alignment; none is misaligned once citation labels are excluded. One gloss cell used a quotation mark for the glottal stop (ʼnii); fixed. A new checker, check_gloss_abbrev.py, now fails on IMP in a Navajo gloss row, a U+2019 glottal in an interlinear cell, or a misaligned inline gloss: 0 remain.
  • Our own derivations and LFs re-read (23 tree panels plus the formal blocks on derivation, clause spine and headedness). Scope matches c-command in every LF. In all 4 negation panels, NegP contains TP, the LF has one ¬ (doo…da is one clause-wrapping negation), and the question operator is above it. No page claims the headedness changeover is established. Fixed: 11 panels that move V to T now each say it is GRANT’S CONJECTURE, untested, neither proven nor refuted; the changeover figure now words the head-direction split as a hypothesis; the clause-spine formal statement says the same. The tree checker now enforces all of this (5 new tests).
  • Placeholder-only trees. 4 trees draw only placeholders (the two derivation skeletons, the light-determiner comparison, the assembly rivals). Their pages do not name a single example to draw from, so instead of attaching words they are now titled “Schematic”. The assembly-rival title says that NX-094 motivates the test but its words are not drawn.

Worst five

  1. Tree panels moved V to T without saying it is untested. 11 panels now state that V→v→T is GRANT’S CONJECTURE, neither proven nor refuted: headedness, verb as derivation, the where-nouns-sit and prefix-order topics, and the light-determiners essay.
  2. The changeover figure stated the head-direction split flatly. It now says this is a hypothesis and not established: headedness.
  3. IMP used for “impersonal” in our Leipzig lines, where IMP means imperative: absolutive indefinite, indefinite person, indefinites topic.
  4. Placeholder trees looked like analyses of real sentences. Now titled “Schematic”: verb as derivation, headedness, light determiners.
  5. A glottal stop typed as a quotation mark in a gloss line: how high is da-?.

Still OPEN

  • Hale 1997 y-/b-ní-Ø-Ø-¬-ʼ²í (from pass 4): needs the original print.
  • Gessner 2003 is still not available to us; the D8 causative rows rest on the page-image check recorded earlier and on Hale & Platero 1996.

Deferred to pass 6

  • Reopen the source for the remaining hand-read rows that were checked only for internal consistency this pass (NLA polar-question rows, Hale 2003 rows on A/B spell-out, Willie §3.2 postposition rows).
  • The other ~460 translation rows outside the syntax/derivation pages.
  • Prose glosses in other shapes (“glossed …”, quoted glosses) are not yet covered by the checker.

Tree panels that move V to T now each say that V→v→T is GRANT’S CONJECTURE: untested, neither proven nor refuted. (Superseded by audit pass 7, 2026-11-13: V→v→T is now the working default.)

Placeholder-only trees are now titled “Schematic” so they are not read as analyses of a particular Navajo example.

2026-11-11 audit pass 6 report

Scope. Same rules as passes 1–5: forms are never invented, reconstructed or changed; only the labels SOURCED, HYPOTHESIS, SPECULATIVE, OPEN, NONE and GRANT’S CONJECTURE are used.

1. Sources reopened for rows pass 5 checked only for consistency. NLA yes/no grammar (version 3.2022) exx. (1)–(4), (3a–c), (9), (10a, b), (11a, b), (15) and the NLA Postposition Lexicon sentence Q2; Hale 2003 spell-out rows N2–N5; Willie 1991 §3.2 postposition rows (26)–(36). All match their sources (forms, glosses, translations, brackets). 0 corrections.

2. Translation rows outside the syntax and derivation pages. 1,117 distinct rows with a translation were read: 234 first shown on the dictionary, 165 on topic pages, and 718 on the other notes and essays. About 45 doubtful-looking rows were checked against the sources (Willie 1991, Saxon 2003 via the note, the NLA lexicons, Smith, Perkins & Fernald 2007): who acts on whom, person and number, negation, tense/aspect and the inverse/passive. No translation differed from its source; every odd-looking one (e.g. Willie (55) ‘John was kissed by two girls’, (74b) lííʼ, Speas’s ‘bit’ under biztał) is printed that way and already flagged. One correction: instrument-voice and the postpositions essay said the instrument bee occurs “in both 3/3 sentences”, but NLA 92 is 3/3 only by its translation (its gloss line marks a first-person subject). Both now say so; the point is OPEN (T51a).

3. Checker. check_gloss_abbrev.py rule 4 applies the alignment test to the other prose-gloss shapes after a segmented Navajo form (quoted gloss, “glossed …”, “is glossed …”, “= …”). 38 such glosses on the built site; 0 mismatches. 5 new tests (13 in all).

4. Labels on our own claims, non-syntax pages. A paragraph screen for V-to-T, the changeover, light d and telicity found 377 unlabelled mentions; most are source reports, headings or text under a labelled section. Fixed: three visible tree-panel titles that move V to T now say GRANT’S CONJECTURE (the changeover and prefix/phrasal figures); §8 of absolutive-indef and indef-person and §12 of fourth-person-uses (statements of the A→B changeover) are now labelled HYPOTHESIS; the inchoative derivations that called ‘get cold’ telic now say this is our reading of the English gloss, with telicity OPEN. Light d and the changeover stay labelled as Grant’s proposals (GRANT’S CONJECTURE) or hypotheses.

Plateau. Pass 6 found no translation or form errors against the sources and only labelling slips. Corrections have plateaued for the translation layer. OPEN carried forward: NLA 92 subject (T51a); NLA A1 Jáan Mary bił baghan was not found again in the live lexicon this pass (unchanged, not re-verified).

2026-11-13: V-to-T and subject raising are the working default

What changed, in plain words. Until today the site treated two movements as Grant’s conjecture, to be tested against an analysis in which nothing moves: the verb moving up to T (V→v→T), and the subject moving from where it starts (Spec,vP) up to Spec,TP. Grant corrected this on 2026-11-13. VPISH, the VP-Internal Subject Hypothesis, is not a way to account for word order. It frames the verb’s arguments so that syntax and semantics line up: the arguments start inside the verb’s phrase, where they get their roles. In ordinary sentences with tense, in most languages, the verb’s head moves to T and the subject moves from Spec,vP to Spec,TP (and on to Spec,CP or C when a question or focus needs it). That is close enough to universal that it should be assumed.

What the site now does.

  • In every tensed clause, V→v→T and subject Spec,vP→Spec,TP are the working default. They are drawn in the trees and used in the derivations, not labelled GRANT’S CONJECTURE, “untested” or “neither proven nor refuted”.
  • The useful question is no longer whether the verb moves but what else has to move to give the Navajo order, where the verb comes last. If T⁰ is final, the raised verb is already last and nothing else has to move. If T⁰ is initial, the material left in the verb phrase (object, postpositional phrases, adverbs) has to move above T, for example as a remnant VP. Each such extra movement is labelled HYPOTHESIS; which head direction is right is OPEN, and the headedness changeover is still not established.
  • The analyses in which nothing moves (PF lowering, template filling, “no-movement rival”) are kept only as labelled contrast, mainly for how prefixes end up inside the verb word. Tree panels that only repeated a clause with no movement were removed or redrawn; they gave the same string and the same meaning, so no data were lost. On Grant’s instruction (2026-11-13) the last three panels that drew the verb word with no V raising were removed too, so no tree now draws a tensed clause without V in T.
  • Sourced examples that seemed to bear on verb movement are kept, now as tests of the default. *NX-246 (hanii after verb + object) is what the default predicts, conditional on H20; its two follow-up tests are OPEN. NX-292 (ʼádin “not exist”) resists the default unless its one argument is read in its lower position or does not raise; that is OPEN.
  • Unchanged: sentential negation (doo … da) is kept apart from constituent negation (hanii); telicity is OPEN; light d is a tested hypothesis; LF scope must match c-command; specifiers are on the left and VPISH is never a node label. No Navajo form, gloss, judgment or source citation changed.

The tree checker now enforces this: every T⁰ in a tensed-clause tree must be the target of V→v→T and the subject must be raised to Spec,TP, unless the panel carries an explicit “V not in T … OPEN” note explaining why. Full list of changes: the pass 7 report.

2026-11-13 audit pass 7 report: V→v→T as the working default

Authorized by Grant (2026-11-13: “This isn’t my syntax if you don’t do that for real”). Same rules as passes 1–6: no Navajo form, gloss, judgment or citation is changed. Full list: site/audits/2026-11-13-pass7.md in the repository. Plain-words summary: the 2026-11-13 note.

  • Trees (14 of 15 tree files; 28 panels). 8 panels that left V in place now draw V→v→T with the subject in Spec,TP (the second panels of the negative-polar NPI and passive-glossed wh figures; both panels, both clauses, of the long-distance wh figure; both panels of the inverse Q-host and negative-question figures). 3 panels that only repeated a clause with no verb or subject movement were removed (negative disjunct, the two relative-clause figures); each gave the same string and LF. 14 panels that already raised V are relabelled “working default” and say what else has to move. 3 panels that drew the verb word with no V raising (prefix/phrasal B; assembly B, PF lowering, and C, template filling) were first kept as labelled contrast with a “V not in T … OPEN” note, then removed on Grant’s instruction, 2026-11-13. They carried no Navajo form, gloss or judgment not shown in the remaining panels (prefix A; assembly A and D, which keep their letters so older dated notes still match). The trees now have 22 panels in these 14 files. The PF/template accounts remain only as table rows and prose contrast on the headedness page.
  • Dated review sections (9): first-principles reviews 4–11 and the phrasal-verbs update. Each has a pass-7 note; “GRANT’S CONJECTURE … tested against a no-movement rival” is now the working default, verdicts say what else has to move, and the no-movement analyses are labelled contrast.
  • Method and formal statements: this page’s core, step table and surface-order section; the clause-spine formal statement; the headedness movement section (§9); background and home-page summaries. Pages whose dated pass-1 notes called V→v→T a comparator to test get a pass-7 note; the old notes are kept as written.
  • Checker: the rule “V-to-T panels must be labelled GRANT’S CONJECTURE” is replaced by “tensed-clause trees show V in T and the subject in Spec,TP, or carry an explicit OPEN note”; panels that raise V must call it the working default, and calling V→v→T a conjecture, untested, or neither proven nor refuted now fails. 5 new tests replace 1.
  • Also fixed: the MIT manuscript hale014 on negative polarity is by Hale & Platero, not “Hale & Perkins”, on the queue and sources pages.

Left OPEN under the default

  • Head direction of T⁰: final T⁰ gives verb-final order with nothing else moving; initial T⁰ needs remnant-VP movement above T (HYPOTHESIS). The headedness changeover is not established.
  • NX-292 (ʼádin): its lone argument must reconstruct or not raise.
  • *NX-246: the default’s two follow-up predictions (whole-clause focus; hanii after the object) need speakers or a source.
  • How a low directional exponent (chʼí) and other prefixes reach the verb word once V has raised: the PF step, if any, is OPEN.
  • What triggers the second subject move above doo (NX-223/229, NX-288).

This page describes the Navajo verb word as a minimalist derivation. The subject DP starts in left Spec,vP. The core builds VP, vP and TP. In a tensed clause, V→v→T and subject raising from Spec,vP to Spec,TP are the working default (see the 2026-11-13 note); VPISH only fixes where the subject starts and gets its role. Additional aspect, applicative and nominal structures are hypotheses, not part of this core. Position classes such as Young & Morgan's "Pos. IV" and "Pos. V" are treated as data that the derivation has to account for, not as the description itself. The earlier skeleton came from a 2026-11-04 from-memory sketch without opened sources; the current generic core replaces it. HYPOTHESIS throughout, unless a claim is marked otherwise.

On this page: 1 Base structure · 2 Tree · 3 Which step is what · 4 The surface-order problem · 5 Word and clause · 6 Position labels as data · 7 References · 8 Open questions

1. Base structure, bottom to top

  1. VP. The root V merges with its internal argument. The core uses an object DP. An extra light determiner selecting a DP/NP, as in Light determiners (analysis D), remains an optional analysis rather than a required layer. GRANT'S CONJECTURE / HYPOTHESIS
  2. P/Appl (optional). When the verb has a postpositional object, a low head sits between V and v and takes its own dP from the same d series. Whether this head is an applicative or an incorporated postposition is OPEN (T92b).
  3. v/Voice. Introduces the external argument. The template's "classifier" (its voice and valence element) is the best candidate for the exponent of v/Voice, and it is listed with the root. HYPOTHESIS
  4. Asp (optional hypothesis). The aspect/mode head above vP. The stem-set alternation (the stem changes shape with aspect and mode) is a dependency between Asp and the stem that is listed with each stem. HYPOTHESIS
  5. T, then independently motivated C. HYPOTHESIS

Where the d heads sit. In the older light-d proposal, object d heads the nominal selected by V. The two subject-d proposals were: (a) It is the same kind of head as the object d, and its dP sits in Spec,Voice. (b) It is a light head in the verbal spine, like Voice: it introduces the subject and is listed with the stem, as light verbs are listed with their nominals. The older draft chose a split: object d taking DP and subject d as a spine head. That choice is not established. The corrected core does not identify d with either nominal D or verbal v. HYPOTHESIS, OPEN (T92a). The stem-listing of indefinite ʼa- (listed-d, Young & Morgan 1987 pp. 362, 364) fits option (b) for subjects SOURCED (listing) / HYPOTHESIS (fit).

2. Tree (schematic, no Navajo forms)

WORKING DEFAULT — generic transitive core, not a source tree
[TP [DP a_i] [T′ T⁰ [vP [DP a_i (lower copy)] [v′ v⁰ [VP [V′ [DP t] V⁰]]]]]]
Specifiers are left. T⁰ and v⁰ are initial; lexical V⁰ is final in this hypothesis.
V⁰ → v⁰ → T⁰; subject DP: Spec,vP → Spec,TP, lower copy unpronounced (working default in tensed clauses).
With T⁰ initial, the raised verb precedes the object; a verb-final clause then needs the remnant VP
to move above T (HYPOTHESIS). With T⁰ final, nothing else moves. Which is right is OPEN.
LF: ∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)]
No telicity, light d, Asp, applicative, C position or prefix exponent order is inferred.
Schematic (placeholder leaves, no Navajo example): a corrected vP-to-T skeletonGeneric transitive-event schematic, not a Navajo morpheme segmentation. Agent DP first merges at left Spec,vP and raises to left Spec,TP, leaving a struck lower copy. V⁰ head-moves through v⁰ to T⁰. T⁰/v⁰ are initial and lexical V⁰ is final; these directions are hypotheses. No Asp, Appl, Qual or light-d head is silently assumed. LF abstracts from tense/aspect, with no telicity entailment. Word-internal exponent order remains a separate OPEN problem. V→v→T and subject Spec,vP→Spec,TP are the working default in tensed clauses (audit pass 7, 2026-11-13). As drawn, T⁰ is initial (HYPOTHESIS), so the raised verb would be pronounced before everything left in vP. The Navajo verb-final order then needs something else to move above T: the remnant VP (object, PPs, adverbs), as in panel D of the assembly figure (HYPOTHESIS). The alternative is a final T⁰, where nothing else moves. Which is right is OPEN. LF: ∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)] Full structure in the list below the figure.TPDPagent aT′T⁰vPDPagent av′v⁰VPV′DPtheme tV⁰predicatehead-movehead-moveAgree φ (optional)Move
(2) Schematic (placeholder leaves, no Navajo example): a corrected vP-to-T skeleton Hypothesis 2026-11-11 audit pass 1: replaces the 2026-11-04 from-memory mixed-category sketch. All leaves are generic placeholders; no new Navajo form. These are our conjectural operations, not a tree attributed to Hale or a deduction from VPISH. VPISH fixes only the stipulated subject base position. Optional d, aspect and applicative structures from the old drawing require independent argument and are OPEN; see /derivation/#2026-11-11-audit-pass-1.
Bracket notation (copy)

[TP [DP agent a] [T′ T⁰ [vP ⟨DP⟩ [v′ v⁰ [VP [V′ [DP theme t] [V⁰ predicate]]]]]]]

Structure as a list
  • TP (id TP)
    • DP (id S-high; landing site of Move from subject): agent a
    • T′ (id Tbar)
      • T⁰ (head; id T; probe for Agree φ (optional) with subject)
      • vP (id vP)
        • DP (id subject; goal of Agree φ (optional); struck copy; moved to S-high): agent a
        • v′ (id vbar)
          • v⁰ (head; id v; head-moves to T)
          • VP (id lowVP)
            • V′ (id Vbar)
              • DP (id object): theme t
              • V⁰ (head; id V; head-moves to v): predicate
The same head chain with a word box
Schematic (placeholder leaves, no Navajo example): a corrected vP-to-T skeleton with a complex-head boxGeneric transitive-event schematic, not a Navajo morpheme segmentation. Agent DP first merges at left Spec,vP and raises to left Spec,TP, leaving a struck lower copy. V⁰ head-moves through v⁰ to T⁰. T⁰/v⁰ are initial and lexical V⁰ is final; these directions are hypotheses. No Asp, Appl, Qual or light-d head is silently assumed. LF abstracts from tense/aspect, with no telicity entailment. The box marks the V+v+T head chain, not a syntactic W node or an asserted PF prefix order. V→v→T and subject Spec,vP→Spec,TP are the working default in tensed clauses (audit pass 7, 2026-11-13). As drawn, T⁰ is initial (HYPOTHESIS), so the raised verb would be pronounced before everything left in vP. The Navajo verb-final order then needs something else to move above T: the remnant VP (object, PPs, adverbs), as in panel D of the assembly figure (HYPOTHESIS). The alternative is a final T⁰, where nothing else moves. Which is right is OPEN. LF: ∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)] Full structure in the list below the figure.TPDPagent aT′T⁰vPDPagent av′v⁰VPV′DPtheme tV⁰predicateV+v+Thead-movehead-moveAgree φ (optional)Move
(3) Schematic (placeholder leaves, no Navajo example): a corrected vP-to-T skeleton with a complex-head box Hypothesis 2026-11-11 audit pass 1: replaces the 2026-11-04 from-memory mixed-category sketch. All leaves are generic placeholders; no new Navajo form. These are our conjectural operations, not a tree attributed to Hale or a deduction from VPISH. VPISH fixes only the stipulated subject base position. Optional d, aspect and applicative structures from the old drawing require independent argument and are OPEN; see /derivation/#2026-11-11-audit-pass-1.
Bracket notation (copy)

[TP [DP agent a] [T′ T⁰ [vP ⟨DP⟩ [v′ v⁰ [VP [V′ [DP theme t] [V⁰ predicate]]]]]]]

Structure as a list
  • TP (id TP)
    • DP (id S-high; landing site of Move from subject): agent a
    • T′ (id Tbar)
      • T⁰ (head; id T; probe for Agree φ (optional) with subject; hosts the “V+v+T” (V + v + T))
      • vP (id vP)
        • DP (id subject; goal of Agree φ (optional); struck copy; moved to S-high): agent a
        • v′ (id vbar)
          • v⁰ (head; id v; head-moves to T)
          • VP (id lowVP)
            • V′ (id Vbar)
              • DP (id object): theme t
              • V⁰ (head; id V; head-moves to v): predicate

Figure: the same skeleton drawn as a tree, with Agree and head movement marked. Schematic labels only, no Navajo forms; every step shown is a HYPOTHESIS in our corrected core, not borrowed from the Syntax bot or read off earlier authors’ word-order analyses.

3. Which step is what

StepOperationWhat it doesStatus
1MergeMerge V⁰ with the theme DP, project V′ and VP.HYPOTHESIS
2MergeMerge v⁰ with VP; merge the agent DP in left Spec,vP.HYPOTHESIS
3MergeMerge T⁰ with vP, creating T′ and the left Spec,TP landing position.HYPOTHESIS
4AgreeOptional T-to-subject Agree copies already interpretable subject person/number; it creates no branch. The identification of T with a particular prefix is OPEN.HYPOTHESIS
5Head-moveOnly after both targets exist: V⁰→v⁰→T⁰. This is the working default in tensed clauses (2026-11-13), not an axiom: an example that resists it is flagged OPEN.HYPOTHESIS
6MoveSubject DP raises from left Spec,vP to left Spec,TP; strike the lower copy.HYPOTHESIS
7MergeIf a construction independently requires C, merge it above TP first. Only then can a separately motivated V→C or DP→Spec,CP operation apply.HYPOTHESIS
8Spell-outSpell out the V+v+T complex at T. If T⁰ is initial, a verb-final clause first needs the remnant VP above T (HYPOTHESIS; OPEN against a final T⁰). PF-lowering/template accounts are kept only as labelled contrast. No rule here derives all Navajo prefix positions.HYPOTHESIS
9InterpretLF: ∃e[PRED(e) ∧ Agent(e,a) ∧ Theme(e,t)]. No aspectual endpoint is entailed.HYPOTHESIS

The fourth person ji- is proposed to be a d of the same series (Light determiners). Where it attaches in the tree is not decided here. OPEN

4. The surface-order problem

In the verb word, the stem comes last and the higher heads come first, with the classifier next to the stem. HYPOTHESIS (stated from memory of the template, not yet checked against Kari 1976). If the spine C T Asp v V were head-initial and nothing moved, this would already give prefixes followed by the stem.

Problem 1 (Mirror Principle). The object prefix is farther from the stem than the subject prefix (Pos. IV before Pos. V; biʼdi-, Young & Morgan 212/3, on pos-labels) SOURCED. But the object merges lower than the subject. A head-initial spine with no movement would put the subject prefix farther left, which is the opposite order. So the object-before-subject order cannot come straight from the order of Merge. HYPOTHESIS

Problem 2 (head movement). A V-first complex follows if each raising head is always left-adjoined and no later reordering applies. Those extra assumptions would give the wrong prefix order. This does not refute V→v→T as such: adjunction direction and PF realization must be specified independently. Head raising is the working default; post-syntactic merger is kept only as labelled contrast for word-internal exponent order, which stays OPEN. HYPOTHESIS

Minimal template statement (a description, not an explanation). The core builds V < v < T. Optional P/Appl, Asp and light-d layers in the older model have not been established. A rule that applies after syntax (a position-class template) orders the spelled-out forms of these heads and of the d heads, and puts the heads listed with the stem right next to it. This keeps the syntax minimalist but gives up the Mirror Principle for the prefix chain. It states the order. It does not derive it. HYPOTHESIS

Getting the order to fall out of the derivation instead would take extra machinery, such as moving d_obj above Voice and building the word by moving heads out of specifiers (Baker 1988's problem). That is unverified and heavy, and this page does not adopt it. OPEN

5. Word and clause: an observed asymmetry

Many declaratives have arguments before the verb word, and many exponents precede its stem. Surface order alone does not establish that those exponents are selecting heads, or that every clause-level head is final. This page records that mismatch as an observation still to be explained, not as a setting of a head-direction parameter. It is the same A (word) / B (clause) split as on driver-3i. HYPOTHESIS

6. Position labels as data

Young & Morgan's "Pos. IV, obj. pron." and "Pos. V, subj. pron." are labels on prefix entries, not on verbs (pos-labels) SOURCED. In this framework they are the facts the derivation has to produce: one surface slot (IV) for a d that merges low, and one (V) for a d that merges high. The labels are not the analysis.

7. References

  • Opened earlier on this site: Young & Morgan 1987 (pp. 212, 362, 364, via listed-d and pos-labels). Aissen 1997 and Willie 1991, as cited in Light determiners.
  • Suggested by the Syntax bot, not opened: Kari 1976; Rice 2000; Hale 1973; Witherspoon 1977; Baker 1988; Speas 1990.

8. Open questions

T92a–d are on Open questions: (a) is the subject d a spine head or a dP in a specifier? (b) Appl or incorporated P? (c) head movement is the working default for V→v→T (2026-11-13); does word-internal prefix order still need post-syntactic merger? (d) does Navajo need phrasal movement of its argument dPs?

Update (2026-11-04, afternoon PT): Where the subject d sits tests options (a) and (b) against Hale 2001's spell-out procedure. Option (b) as stated here (a Voice-like spine head) gives the wrong order for (22b) shiʼdooʼniid. So does a plain dP in a specifier. Historical result under that particular procedure: a dP in Spec,vP with d displacement, or a spine head above Mode/Subj. Neither is a current result under all rivals; the dated audit above supersedes treating this procedure as the standard. HYPOTHESIS

Keeping a state vs bringing about a change

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. Kept as written (URL, data and status labels unchanged). Filed under Event structure and argument realization; the consolidated argument is the essay events. Rule IDs: rules index.

Verbs on this page → dictionary: d-ee-zaʼ ‘belch, burp’ · yiizį́ ‘stand up; stand (someone) up; hold (someone) standing’ · neezdá ‘sit down; make sit; keep (somewhere)’ · neezhtéézh ‘lie down (two)’ · siʼka ‘be there; have (it) for a purpose’ · sią́ ‘(solid round object) be on; keep from rolling’ · neezdá ‘sat down; with yikáá’ ‘on it’: sat down on it’

Restated on the unified structure (2026-11-15). Voice, causatives and events are restated with every theta role assigned inside vP: the classifier spells a value of v⁰ (H59, distribution NOT UNDERSTOOD), a causer is an external argument in Spec,vP, and the passive keeps a silent indefinite agent (H10b). This page’s causative analysis has not yet been restated on the unified structure (OPEN). Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. 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, causatives 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.

One Navajo stem, three event shapes: ‘he stood up’, ‘I stood him up’, ‘I hold him standing’. What the published forms show, how other languages draw the same line, and what it means for the proto-role schema.

On this page: 1 Question · 2 Navajo forms · 3 Other languages · 4 Entailments · 5 Rules · 6 Derivations M1–M3 · 7 Is the held object a problem? · 8 Open

1. The question

GRANT'S CONJECTURE (paraphrased from Grant's framing of this research track) Language tracks a fine-grained picture of how events unfold. Argument ranking should be studied at that level of detail, not at the level of a nickname for a theta role.

HYPOTHESIS Here is a test case at that level. A causer can bring about a change (‘stand X up’). A causer can also keep a state going (‘hold X standing’). In the second case nothing changes, yet the held participant is still the object. Navajo marks this difference inside the verb word (§2). If the proto-role schema is right that a verb's arguments are entailment profiles, then the cut should be statable as a difference in entailments (§4–§5), and the derivations should show where in the word that difference is made (§6).

2. The Navajo forms SOURCED

Source. Goertz, Grandon, Katy Lenté, Stefanie Adams, Jalon Begay & Melvatha Chee 2006. "Use of causatives in Navajo: Syntax and morphology." Santa Barbara Papers in Linguistics 18 (Proceedings of the Ninth Workshop on American Indigenous Languages). 14-page PDF from UCSB Linguistics. The PDF prints no page numbers, so pages are cited as PDF pages. Numbered examples come from the authors' consultants. Lettered examples come from published sources (here Young 2000:239, "verified by our consultants").

Verification. The PDF's text layer drops the nasal hook and the barred ł. Every form below was therefore read off page images rendered at 300 and 600 dpi (PDF pp. 9, 10, 11) and re-checked on 2026-10-07. Apostrophes are kept as printed: a typographic ’ in (17), (21a) and (22b), and a straight ' in (23) and (24).

Glossing. Goertz et al. give a free translation only. They give no segmentation and no morpheme gloss for any form in this section. Following the site's convention, the forms are left unsegmented, and no Leipzig gloss line is supplied. §2.4 records what is visible in the strings, and those observations are not a segmentation.

2.1 The posture set: change by the subject, change by a causer, state held by a causer

Goertz et al. 2006, ex. (O), PDF p. 11 (from Young 2000:239)
a.yiizį́
‘he/she/it stood up’
b.biiłzį́
‘I stood him/her/it up’
c.bisésį́
‘I have or hold him/her/it standing (standing position)’

The closing quotation mark of the (c) translation is missing on the printed page. Unsegmented; the source gives no gloss.

Goertz et al. 2006, exx. (M), (N), PDF p. 11 (from Young 2000:239)
M a.neezdá
‘he /she/ it sat down’
M b.binéłdá
‘I caused him/her/it to sit down’
N a.neezhtéézh
‘they two lay down’
N b.binéłtéézh
‘I made them two lie down’ (‘They are the possessor of my actions’, according to consultants.)
Goertz et al. 2006, ex. (17), PDF p. 9 (consultants)
a.neezdʼa
‘she sat down’
b.binełtʼa
‘I made her sit down (I caused her to sit down)’

Recorded as printed, not repaired. The same ‘sit down’ pair is spelled two ways in one paper: neezd’a / binełtʼa on p. 9 and neezdá / binéłdá on p. 11. Both were checked at 600 dpi. This site does not choose between them.

2.2 Stative bases: a state that holds by itself vs the same state held by the speaker

Goertz et al. 2006, exx. (21)–(22), PDF pp. 10–11 (consultants)
21a.siʼka
‘it’s there’
21b.sełka
‘I have it for a purpose’
22a.sią́
‘it (a solid, round object) is on the table without my doing anything’
22b.séłʼą́
‘I kept it from rolling around on the table’

The prose introduces these as "examples 18 and 19". The printed numbers are 21–22. The a-forms are "the plain stative verb", and the b-forms are "the derived causatives with ł- classifier" (PDF p. 10).

2.3 In sentences

Goertz et al. 2006, exx. (23)–(24), PDF p. 11 (native speakers)
23.shichidí hooghan bineʼ jí bisésį́
‘I keep a car behind the Hogan’ (car is animate)
24.tsésǫʼ gi tsídiiłtsooí biséłdá
‘I keep a canary in my window’ (I cause a canary to be in the window)

No word-by-word gloss is given by the source. The verb word of (23) is the same string as (O c).

What the authors say SOURCED (Goertz et al. §5.2, PDF pp. 10–11; §6 Conclusion, PDF p. 14):

  • "Two morphological causatives in Navajo are of special interest in that they do not signal causation in the usual ‘making something happen’ way, but rather signal intentionality, in the maintenance of a particular state."
  • They quote Kibrik (1993:55–56) on Athabaskan "possessives": the possessive meaning is "rather a side effect, typically but not obligatorily accompanying the inherent meaning that can be expressed as causing an existential/locative state to be maintained, controlling a goal in a position." (Kibrik 1993 itself not opened; secondhand.)
  • "When a person changes position, it is the subject that performs the action involved. Causatives of these verbs express that the change in position was done to the person by a causer."
  • Of (23)–(24): "the speaker is not only stating fact, but is stating that he or she has caused and is in control of the situation."
  • Conclusion: "Stative causative constructions (section 5.2) imply that the speaker is actually in physical control of the described event."

2.4 What is visible in the strings HYPOTHESIS

These are observations about letters on the page. They are not a segmentation. Navajo stems fuse with what precedes them. For example, Hale & Platero's ‘burp’ causative bi-di-y-é-saʼ is printed "< … -ł-zaʼ" (Gessner 2003 ex. (11), on the derivations page), so ł + z there surfaces as s. A missing ł in the string therefore does not show that ł is absent.

FormEvent shape (from translation)Initial bi…Letter ł visibleEnds in
yiizį́ (O a)change, by the subjectnono-zį́
biiłzį́ (O b)change, by a causeryesyes-zį́
bisésį́ (O c), (23)state, held by a causeryesno (cf. ł + z → s in H&P ‘burp’: OPEN)-sį́
binéłdá / binełtʼa (M b) / (17b)change, by a causeryesyes-dá / -t’a
binéłtéézh (N b)change, by a causeryesyes-téézh
biséłdá (24)state, held by a causeryesyes-dá
sełka (21b), séłʼą́ (22b)state, held by a causernoyes-ka / -ʼą́

Two patterns, both HYPOTHESIS:

  1. The stem stays and the frame changes. (O a–c) share a final -zį́/-sį́, and (M b) and (24) share -dá. (24) ‘keep a canary in my window’ looks like the held-state form of the ‘sit’ set, just as (O c) is the held-state form of the ‘stand’ set. If that is right, one posture stem appears in a change frame and a held-state frame. The difference between them is made by material to the left of the stem, which is rule set A territory.
  2. bi… tracks the kind of object, not change. Initial bi… appears when the held or moved participant is a person, a pair of people, a canary, or a car that Goertz et al. explicitly note "is animate". It is absent in (21b) and (22b), whose objects are inanimate ‘it’. All of these forms have first-person subjects. That matches Hale & Platero's special causee marking: third-person bi- is overt with a first- or second-person subject (schema §6). Goertz et al. do not make this connection, and the site does not treat this bi- as the yi-/bi- inverse.

3. The same cut in other languages

Only examples from sources opened on 2026-10-07 appear here, with their pages. Where a source reports another author's data, the row says so.

3.1 Talmy: onset causation vs extended causation SOURCED

Talmy, Leonard 2000. Toward a Cognitive Semantics, vol. 1, MIT Press. Chapter 7, "Force Dynamics in Language and Cognition", and chapter 8, "The Semantics of Causation". Author-hosted chapter PDFs, cited by the book's printed pages.

  • Steady-state causation, which includes causation of rest. (3a) The ball kept rolling because of the wind blowing on it. (3d) The log kept lying on the incline because of the ridge there. These two are "extended causation": (3a) of motion, (3d) of rest (ch. 7, pp. 415–416).
  • Onset causation. (5e) The ball's hitting it made the lamp topple from the table. (5f) The water's dripping on it made the fire die down. "If the two steady-state causative types, (3a,d), may be termed cases of extended causation, the present type can be called a case of onset causation" (p. 418).
  • One definition for both. What is "definitional for the concept of causation" is "that the Agonist's resultant state of activity is the opposite of its intrinsic actional tendency" (p. 418). Talmy therefore counts holding something at rest against its tendency as causation, with no change involved.
  • The causer as subject. (9b) The wind kept the ball rolling. / The wind is making the ball roll. Transitive keep puts the Antagonist in subject position and the Agonist in object position (p. 423). Talmy calls keep -ing "perhaps the single form most indicative of the presence of force dynamics". He says its equivalents have grammaticalized in other languages, but names none (p. 417).
  • Agentive version. (4) I pushed the box across the ice is ambiguous between "[I kept it in motion, going along with it.] (extended causation)" and "[I set it in motion and stayed put.] (onset causation)" (ch. 8, p. 473; §4.2, pp. 498–500). Onset causation is equated with "Shibatani's ‘ballistic causative’" (p. 500).
  • Counterfactual test. For point-durational causation, removing the cause "would entail the carton's remaining at rest". For extent-durational causation, removing it at any point "would entail the carton's coming to rest" (ch. 8 §3.7, p. 491).

3.2 Copley & Harley: keep and stay are dynamic but involve no change SOURCED

Copley, Bridget & Heidi Harley 2015. "A force-theoretic framework for event structure." Linguistics and Philosophy 38:103–158, §5.3 "Verbs of maintaining", pp. 146–149 (published PDF, text layer).

  • (66) The rock is keeping the door open. / The door is staying open. (67) The rock keeps the door open. / The door stays open. These verbs "are clearly dynamic … but which do not involve a change of any kind" (§5.3 opens on p. 146; the examples and quote are on p. 147).
  • Against event decompositions (CAUSE + BE, CAUSE + BECOME): "it is possible to keep something in a location without strictly being the cause of its being there or of its coming to be there" (p. 147).
  • Their proposal: ⟦vSTAY⟧ = λpλf . p(fin(f)), with p(init(f)) presupposed (p. 148; labelled (70) in the text layer). "keep includes a VoiceP projection which introduces an external argument and asserts that it is the source of the force; and stay does not". Keep is "a transitive suppletive variant of stay, in the same way that raise is allomorphically related to rise, lay to lie" (p. 149).

The prose on p. 147 refers to the four candidate decompositions as (68a–d), but the text layer prints them under the label (70). That is recorded as found; the page image was not checked, because no Navajo is involved.

3.3 Japanese and Korean: one verb, a pick-up reading and a keep-holding reading SOURCED

Ogihara, Toshiyuki & Eun-Hae Park 2019. "Aspect shift without coercion: Continuous causative verbs in Japanese and Korean." Journal of East Asian Linguistics 28(2):107–141. Read in the authors' pre-publication PDF (linked from Ogihara's page), cited by its manuscript pages.

Ogihara & Park (21), ms. p. 23
a.Hanako-wahandobaggu-ote-nimot-ta.
Hanako-TOPhandbag-ACChand-atpick.up-PAST
‘Hanako picked up her handbag (by hand).’ [Japanese]
b.Hanako-wahandobaggu-ote-nimot-te iru.
Hanako-TOPhandbag-ACChand-atpick.up-TEIRU.PRES
‘Hanako is carrying her handbag (by hand).’ [Japanese]
c.Jiho-nunhandbag-ul(son-ey)tul-ko iss-ta.
Jiho-TOPhandbag-ACChand-atpick.up-KOISS.PRES-DECL
‘Jiho is carrying her handbag.’ or ‘Jiho is picking up her handbag.’ [Korean]

Glosses as printed. The authors gloss -te iru and -ko iss as TEIRU and KOISS on purpose, "so as not to mislead the reader" (n. 1).

  • "maintaining this state requires Hanako's carrying the handbag in her arms or holding onto its handle by hand. Otherwise, it would fall to the ground" (ms. p. 23). Their (22): R1 = "x holds y (in x's hand) at t", the target state, and R2 = "x supplies appropriate force to keep y held in x's hand at t" (ms. p. 24).
  • Generalization (25), ms. p. 27: a -te iru/-ko iss sentence with an agentive subject has a target-state reading "if the agent is the controller of the target state, and the target state under discussion is physically identifiable, non-permanent and reversible". They call such verbs "continuous causative (CC) verbs".

3.4 Posture verbs without a causer: assume vs maintain SOURCED (secondhand data)

Newman, John 2022. "Experiential motivation and the linguistics of sitting, standing, and lying." WIREs Cognitive Science 13(4):e1592 (open access, PMC9539599; HTML, §4.1). The language data in this list are Newman's reports of other authors, which were not opened:

  • Chantyal (Tibeto-Burman), after Noonan & Grunow-Hårsta 2002:81–82: "‘stand’ can be expressed as a simplex verb …, either yep‐ ‘maintain a standing position’ or yes‐ ‘assume a standing position’". For ‘lie’ there are only complex forms, e.g. terso la‐ ‘maintain a horizontal position’ vs terso ta‐ ‘assume a horizontal position’.
  • Usan (Trans–New Guinea), after Reesink 1987:132: the action verbs (‘stand up’ naget, etc.) are basic, and "the at‐rest meanings are expressed by putting the action verbs into a continuative aspect".
  • German: causative setzen ‘to set something/someone down’ is "historically derived from the stative sitzen", and the inchoative is reflexive sich setzen.

HYPOTHESIS These are the causer-less half of the contrast: entering a posture vs staying in it. They line up with Navajo (O a) ‘stood up’ against a plain ‘is standing’. Goertz et al. do not print that plain stative for the ‘stand’ stem, so this cell is OPEN for Navajo.

3.5 Side by side

LanguageBring about a changeKeep a state (causer)How the difference is markedStatus
Navajobiiłzį́ ‘I stood him/her/it up’bisésį́ ‘I have or hold him/her/it standing’Different verb words on what looks like one posture stem; the material left of the stem differsSOURCED forms; HYPOTHESIS "one stem"
Englishmake … roll; onset readings of pushkeep … rolling; keep the door open; extended readings of pushDifferent light verb (make vs keep), or ambiguity of one verb (push)SOURCED Talmy; Copley & Harley
Japanesete-ni mot-ta ‘picked up’te-ni mot-te iru ‘is carrying (holding)’Same verb; aspect -te iru selects the maintained-state readingSOURCED Ogihara & Park
Koreantul- ‘pick up’tul-ko iss-ta ‘is carrying’ (also ‘is picking up’)Same verb; -ko iss, which is ambiguousSOURCED Ogihara & Park
Chantyalyes‐ ‘assume a standing position’(no causer) yep‐ ‘maintain a standing position’Different simplex verbsSOURCED via Newman (secondhand)
Athabaskan generallył-causative"possessive" causative: "causing an existential/locative state to be maintained"MorphologicalSOURCED Kibrik 1993 via Goertz (secondhand)

HYPOTHESIS In the material opened, Navajo is the only language where the keep-a-state meaning is a derived transitive verb built on a stative or posture stem, with the held participant marked as the object inside the word. In Japanese and Korean, aspect on the change verb carries the meaning. In English it is a separate verb, keep. Copley & Harley's analysis of English (keep = stay + an external-argument VoiceP) has the same shape the Navajo forms suggest: stative base + causer layer. That parallel is ours; no source draws it.

OPEN Not opened this run, so no claims are made from them: Dixon 2000 on causative parameters, Shibatani's ballistic/controlled causatives (cited here only through Talmy), Song 1996, Kulikov 2001, Nedjalkov & Silnitsky 1973, Kibrik 1993, Young 2000, and the Dutch/Germanic posture-verb literature (Lemmens; Newman ed. 2002).

4. The cut as proto-role entailments

SOURCED Dowty 1991:572 (page image re-checked 2026-10-07), (27) Proto-Agent: a. "volitional involvement in the event or state"; b. "sentience (and/or perception)"; c. "causing an event or change of state in another participant"; d. "movement (relative to the position of another participant)"; (e. "exists independently of the event named by the verb"). (28) Proto-Patient: a. "undergoes change of state"; b. "incremental theme"; c. "causally affected by another participant"; d. "stationary relative to movement of another participant"; (e. "does not exist independently of the event, or not at all"). Dowty offers these as "preliminary lists" and does not claim they are "necessarily exhaustive". On p. 576: ASP (31); Corollary 1 (32); and (34) Nondiscreteness: "some arguments have neither role".

Where the wording bites. HYPOTHESIS (27a) already covers states ("in the event or state"). (27c) does not obviously cover them: it says "causing an event or change of state". If keeping X standing counts as causing an event, the holder has (27c). Talmy (definition of causation, p. 418) and Copley & Harley (keep is dynamic, pp. 146–149) both support counting it. On a reading where an event requires a change, the holder lacks (27c). To make the choice explicit, the site splits the entailment in two:

caus_Δ   causes a change of state or position in another participant   (Talmy: onset causation)
caus_=   causes a state of another participant to continue, against
         its tendency otherwise                                       (Talmy: extended causation of rest;
                                                                       Copley & Harley: keep)
Dowty's (27c) caus  =  caus_Δ ∨ caus_=        [our reading, HYPOTHESIS]
M1 change by subject
(O a) yiizį́
M2 change by causer
(O b) biiłzį́
M3 state held by causer
(O c) bisésį́, (23), (24)
Slots{ρ₁}{ρC, ρ₁}{ρH, ρ₁}
Higher slot(none)ρC: vol, sent, caus_Δ, indepρH: vol, sent, caus_=, indep. Goertz: "in control of the situation"
ρ₁ (the one who stands)chg (posture) + mov; vol, sent if a person (‘he/she’), not if ‘it’chg + aff; mov (rises); vol is unclear (a person helped up may cooperate)¬chg, ¬mov; aff (only if (28c) counts maintenance); stat?
Subject by ASP (31)ρ₁ρC (more P-A)ρH (more P-A)
Object by ASP (31)(none)ρ₁ (chg, aff)ρ₁ (aff at most; possibly no P-P entailment at all, which (34) allows)

All entailment assignments are HYPOTHESIS, made from the source's English translations. No Navajo speaker judgment on entailments was available.

The cut, stated over slots HYPOTHESIS. M2 and M3 have the same number of slots, the same linking (higher slot = subject, ρ₁ = object), and, in the strings, the same object marker bi…. They differ in exactly two entailments, one on each slot: caus_Δ vs caus_= on the higher slot, and chg vs ¬chg on ρ₁. The two differences are linked. ρ₁ has chg exactly when the higher slot has caus_Δ. So the cut is one decision, about whether the event contains a change, and it shows up as entailments on both arguments.

5. Rules HYPOTHESIS

ADD-CAUSER on the schema page (§6) has two cases, keyed to the base slot: (a) chg-dominant base → ł only; (b) vol/mov-dominant base → ł + y + oblique bi-. A stative base like (21a) siʼka ‘it's there’ or (22a) sią́ ‘is on the table without my doing anything’ has a sole slot with neither chg nor vol. ADD-CAUSER as stated does not apply to it. The Navajo stative causatives need a third operation:

ADD-CAUSER_Δ(Σ)   (onset; = schema §6 cases a/b)
  input   Σ = {ρ₁},  E_ρ1 ∋ chg  (or ρ₁ is an agent-like mover, case b)
  output  Σ' = Σ ∪ {ρ_C},  E_ρC ⊇ {vol, caus_Δ},  pos_ρC = Spec,v(ł)
          E'_ρ1 = E_ρ1 ∪ {aff}

MAINTAIN(Σ)       (extended causation of a state; new)
  input   Σ = {ρ₁}, built on a STATIVE base (posture or location),  chg ∉ E_ρ1
  output  Σ' = Σ ∪ {ρ_H},  E_ρH ⊇ {vol, caus_=},  pos_ρH = Spec,v(ł)
          E'_ρ1 = E_ρ1 ∪ {aff}         chg ∉ E'_ρ1   (no change is added)
  spell-out  ł-causative on the stative stem (Goertz §5.2: "derived causatives with ł- classifier");
             object exponent bi- when ρ₁ is animate-like and the subject is 1st/2nd person
             (§2.4 pattern 2; OPEN)

Two conditions keep the rules tight:

#ConditionBasis
C10Change coherence. If the higher slot has caus_Δ, then chg ∈ Eρ1. If it has caus_=, then chg ∉ Eρ1. The two causal entailments are never both present on one slot of one verb word.HYPOTHESIS It follows the M2/M3 split in §4. Talmy's onset vs extended causation are distinct patterns (ch. 7, pp. 415–418). Copley & Harley's ambiguous push and Ogihara & Park's ambiguous Korean -ko iss are ambiguities, not a single reading with both.
C11Fixity of the cut. The choice between caus_Δ and caus_= is made in Σ (stem + operation). It does not change under linking, under the yi-/bi- choice, or under rule set B.HYPOTHESIS It extends C2 (event-fixity). It is testable: an inverse version of an (O c)-type verb, if one exists, should still mean ‘hold standing’.

6. Derivations M1–M3, step by step

Conventions are as on the derivations page. Tags: Σ schema step, A word-internal linearization, B clause level. V-H = Hale-style in-situ licensing; V-P = VPISH overlay. The source gives no segmentation, so the word-internal steps use schematic pieces: [MODE], [CL], [OBJ], √STAND. The surface string is printed whole and never cut into morphemes.

M1. Change by the subject: (O a) yiizį́ ‘he/she/it stood up’

  1. Σ Σ(√STAND, v) = {ρ₁}. Eρ1 ⊇ {chg, mov}, plus {vol, sent} if ρ₁ is a person. posρ1 is OPEN: Spec,V if the chg profile decides (unaccusative, Hale 2003 (22)), Spec,v if vol/mov decides (unergative, Hale 2003 (23)). Goertz et al.'s "it is the subject that performs the action involved" favours Spec,v for people. The ‘it’ reading favours Spec,V.
  2. Merge V = √STAND. Merge v = [CL] (which classifier is not readable from the string). A v⌢VP.
  3. Merge ρ₁ at posρ1. φ unvalued.
  4. Merge M/S = [MODE]. The translation ‘stood up’ is perfective; the mode is not given by the source. A M/S⌢vP.
  5. Agree M/S ⟶ ρ₁: [3]. Σ C1: one slot, one exponent (null or fused; not segmentable here).
  6. Move (V-P only): ρ₁ goes Spec,v (or Spec,V) → Spec,M/SP, leaving ⟨ρ₁⟩. It has no overt reflex, as in D1.
  7. Spell-out A [MODE/3]⌢[CL]⌢√STAND → yiizį́, unsegmented.
  8. B No DP. The verb word is the clause.

M2. Change brought about by a causer: (O b) biiłzį́ ‘I stood him/her/it up’

  1. Σ ADD-CAUSER_Δ(ΣM1) = {ρC, ρ₁}. EρC ⊇ {vol, sent, caus_Δ}. Eρ1 ⊇ {chg, aff}. C10 ✓.
  2. Merge V = √STAND with ρ₁ (in Spec,V, or in an inner Spec,v₂ if M1 is unergative; see step 4).
  3. Merge outer v = ł (a letter ł is visible in the string). Merge ρC in Spec,v. A v⌢VP.
  4. OPEN Branch. If ρ₁ was in Spec,v₂ (unergative), this is H&P's complex causative: Qual y is merged and ρ₁ is relinked to an oblique (D8). The initial bi… with a first-person subject fits that pattern. Whether a y is present cannot be read off biiłzį́ without a segmentation (Young 2000:239 or Young & Morgan would settle it).
  5. Merge M/S = [MODE]. Agree M/S ⟶ ρC: [1sg].
  6. Σ ρ₁ is valued [3], exponent [OBJ]. The string begins bi…. C1 ✓ (two slots, two exponents). C4: R = {ρ₁} only (ρC is 1st person), so VOICE is undefined. This bi… is not the inverse.
  7. Move (V-P): ρC Spec-to-Spec to Spec,M/SP. The movement is the same as in D2.
  8. Spell-out A [OBJ]⌢…⌢[MODE/1sg]⌢ł⌢√STAND → biiłzį́, unsegmented. The object exponent is leftmost, the classifier is next to the stem, and functional material precedes its complement.
  9. B No DP.

M3. State held by a causer: (O c) bisésį́ ‘I have or hold him/her/it standing’ and (23)

  1. Σ Base: Σ(√STAND, stative) = {ρ₁}, Eρ1 ∌ chg (‘be standing’). The plain stative form is not printed in the source, so it is schematic here.
  2. Σ MAINTAIN(Σ) = {ρH, ρ₁}. EρH ⊇ {vol, sent, caus_=}. Eρ1 gains at most aff and no chg. C10 ✓ (caus_= with ¬chg).
  3. Merge V = √STAND (stative) with ρ₁ in Spec,V. The source does not say whether the standing person is merged lower here than in M2. On the entailments it should be, since ρ₁ in M3 has no vol or mov; HYPOTHESIS.
  4. Merge v = ł-classifier, per Goertz et al.'s description of the stative causatives (PDF p. 10). Merge ρH in Spec,v. No letter ł is visible in bisésį́. Whether ł + z → s fusion applies, as in H&P's ‘burp’ (-sa’ < -ł-zaʼ), is OPEN. In (24) biséłdá the ł is visible.
  5. Merge M/S = [MODE]. ‘I have or hold’ is present and stative; the mode is not given. Agree M/S ⟶ ρH: [1sg].
  6. Σ ρ₁ is valued [3], exponent [OBJ], and the string begins bi…. C1 ✓. C4: VOICE is undefined (1st-person subject). C11: nothing later can turn caus_= into caus_Δ.
  7. Move (V-P): ρH Spec-to-Spec to Spec,M/SP. This step is identical to M2. VPISH cannot tell keeping from bringing about.
  8. Spell-out A [OBJ]⌢…⌢[MODE/1sg]⌢[ł]⌢√STAND → bisésį́, unsegmented.
  9. B For (23) shichidí hooghan bineʼ jí bisésį́: the material before the verb word is linked as clause-level adjuncts (C3). By the translation ‘I keep a car behind the Hogan’, shichidí links to ρ₁ (‘car’) and hooghan bineʼ jí is the locative ‘behind the Hogan’. That word-to-meaning match is ours (HYPOTHESIS); the source gives no word glosses. The locative names the place where the state holds. It is not a slot in Σ, so it is a rule-set-B adjunct, as in the postposition question on the open-questions page. Order: DP ⌢ locative ⌢ V. That is the ordinary verb-final clause. Goertz et al. note the car "is animate", and the verb carries bi…, consistent with §2.4 pattern 2.

What M1–M3 show HYPOTHESIS.

  1. The whole cut sits in steps Σ and A. M2 and M3 run through the same Merge, Agree and Move steps, with the same linking and the same clause-level syntax. They differ only in the Σ operation (ADD-CAUSER_Δ vs MAINTAIN), in the base (change vs stative), and in what rule set A spells out inside the word. Rule set B does no work on the distinction. So "keep" vs "bring about" is decided entirely on the word side of Grant's changeover point.
  2. VPISH is blind to it. The causer's Spec-to-Spec movement is the same in M2 and M3. Like the movement in D1–D5, it has no reflex that would separate the two.
  3. The schema has to see it. Under the Pronominal Argument Hypothesis, M2 and M3 are both "1st-person subject pronoun + 3rd-person object pronoun". The difference has to be stated somewhere other than the pronouns: over the base (stative vs change) or over the slots' entailments (caus_= vs caus_Δ). The schema states it over entailments. That is a statement about event structure, and a PAH analysis can state it equally well over the base predicate. So this case does not decide schema vs PAH (T7 on the open-questions page stays open).

7. Is the held object a problem for the schema or the PAH?

Short answer: no for the PAH and for the ASP; partly yes for the site's own rules, which this page fixes. HYPOTHESIS

  • Dowty's ASP: not a problem. The ASP counts entailments; it does not require the object to change. In M3 the holder has vol, sent and indep, plus caus_= if (27c) is read to include it, so it wins the subject. The held participant is the object with at most one Proto-Patient entailment (aff), and possibly none. Nondiscreteness (34), "some arguments have neither role", allows that. The held object is a weak object, not an impossible one.
  • Dowty's lists: one wording decision. (27c) says "event or change of state". Navajo M3 forces a choice: either a maintained state counts as an event (Talmy, Copley & Harley), or the holder's subjecthood rests only on vol, sent and indep. The site takes the first option and writes it as caus_= (§4). The held object also tests (28c) "causally affected". If keeping something in place does not count as affecting it, the M3 object has no Proto-Patient entailment at all, and it is the object only because the other argument outranks it.
  • The PAH: not a problem. The object prefix is a pronominal argument whatever its entailments. PAH says nothing about change.
  • The site's conditions C1–C9: not a problem. None of them mentions chg.
  • The site's ADD-CAUSER: a real gap, now patched. Its two cases were keyed to chg vs vol on the base slot, and a stative base has neither. MAINTAIN (§5) is the missing third case, with C10 and C11 to keep it from overgenerating.

8. Open questions this page adds

#QuestionWhat would settle it
K1Segmentation of (O a–c), (M), (N), (21)–(24): where are the object, mode, subject, classifier and stem? Is there a Qualifier y in biiłzį́? Is ł fused into -sį́ in bisésį́?Young 2000:239 itself; Young & Morgan 1987 paradigms for the ‘stand’, ‘sit’ and ‘lie’ stems. OPEN
K2What is the plain stative ‘is standing’ (the base of M3), and is (O c) built on it or on the (O a) change verb?Same sources; consultant check. OPEN
K3Does (O c) entail a prior (O b)? Can you ‘hold someone standing’ who stood up by themselves? Copley & Harley's point is that keeping need not involve causing the state to come about.Native-speaker judgment. OPEN
K4Does bi… in these forms track animacy (car "is animate"; canary) or H&P's causee marking with 1st/2nd subjects, or both? Third-person-subject versions of (O b)/(O c) would show which.Paradigms with 3rd-person subjects. OPEN
K5Do M3 forms behave like Copley & Harley's keep, dynamic but with no change? That is, are they compatible with ‘for an hour’ phrases, and how do they behave with the progressive?Grant's texts or elicitation. OPEN
K6Which spelling of ‘sit down’ is current: p. 9 neezd’a/binełtʼa or p. 11 neezdá/binéłdá?Young 2000; a dictionary; a speaker. OPEN

Which intransitive verbs take a plain ł-causer?

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.

2026-11-11 audit pass 4. Interlinear rows whose form and gloss do not line up one-to-one were re-checked against the source page. Each is marked † with a note under its table: SOURCED where the source prints it that way, OPEN where the source cannot confirm it. See the pass 4 report.

Working note. Kept as written (URL, data and status labels unchanged). Filed under Event structure and argument realization; the consolidated argument is the essay events. Rule IDs: rules index.
Verbs on this page 42 entries in the verb dictionary (show)

si-tsʼil ‘shatter’ · yí-yį́į́ʼ ‘melt’ · n-eez-tsiz ‘go out (fire); put out’ · kʼí-ní-dláád ‘break (rope)’ · yi-dloh ‘laugh’ · naa-ghá ‘walk around (one)’ · d-ee-zaʼ ‘belch, burp’ · yiizį́ ‘stand up; stand (someone) up; hold (someone) standing’ · neezdá ‘sit down; make sit; keep (somewhere)’ · neezhtéézh ‘lie down (two)’ · siʼka ‘be there; have (it) for a purpose’ · sią́ ‘(solid round object) be on; keep from rolling’ · OBJ-ʼ-y-ł-dlą́ ‘make OBJ drink something’ · ha-OBJ-y-ł-yeed ‘run OBJ up out’ · OBJ-ʼ-y-ł-haazh ‘make OBJ (go to) sleep’ · ʼ-(ł-)ʼeeł ‘float away’ · ii-(ł-)gááh ‘whiten’ · (ł-)gan ‘dry up’ · ʼ-(ł-)geeh ‘fall away [as person, animal]’ · ii-(ł-)kíísh ‘become spotted, put spots on’ · ii-(ł-)kʼis ‘crack’ · (ł-)lą́ąh ‘increase [in number or quantity]’ · ʼ-(ł-)lį́ ‘flow away’ · di-(ł-)lid ‘be burning’ · ʼ-(ł-)máás ‘roll away’ · ii-(ł-)táás ‘bend over, double’ · (ł-)tʼees ‘cook, roast, etc.’ · ʼ-(ł-)tʼééh ‘extend away [line, fence]’ · ii-(ł-)tłíísh ‘darken, turn brown’ · (ł-)tłʼis ‘harden [as mud, dough]’ · ii-(ł-)tsóóh ‘yellow’ · di-(ł-)tsʼǫǫd ‘stretch’ · di-(ł-)zháásh ‘begin to wear away, down’ · (ł-)zhǫǫh ‘become gentle, make gentle’ · na-bé ‘swim, bathe’ · di-lish ‘spurt urine [as of dog]’ · na-né ‘play’ · ho-taał ‘sing’ · di-zheeh ‘spit’ · ʼ-zhííł ‘gasp, inhale sharply’ · di-yih ‘pant, puff’ · ʼ-yóół ‘inhale’

Restated on the unified structure (2026-11-15). Voice, causatives and events are restated with every theta role assigned inside vP: the classifier spells a value of v⁰ (H59, distribution NOT UNDERSTOOD), a causer is an external argument in Spec,vP, and the passive keeps a silent indefinite agent (H10b). This page’s causative analysis has not yet been restated on the unified structure (OPEN). Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. 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. Derivations below in which the verb or subject does not raise are labelled contrast; the working default raises V→v→T and the subject to Spec,TP. Status on the spine: see unified structure, §4, causatives 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.

Some Navajo intransitive verbs become transitive just by adding the classifier ł (‘the rope broke’ → ‘he broke the rope’). Others cannot; they need the bigger causative with y- and an oblique bi- (‘the baby laughed’ → ‘I made the baby laugh’). This page tests whether the split can be predicted from what the verb entails about its one participant, or whether each verb has to be marked by hand. This is open question T4.

On this page: 1 Question · 2 Hale & Platero's data · 3 Dowty on intransitive classes · 4 Entailments of each verb · 5 Which rule fits · 6 Tightened rules · 7 Derivations C1–C4 · 8 Verdict · 9 Open

1. The question

The schema page, §6 states ADD-CAUSER with two cases. In case (a) the base's sole slot is first-Merged in Spec,V, and the causative is ł only (simple). In case (b) it is first-Merged in Spec,v, and the causative is y + ł + oblique bi- (complex). Those cases are keyed to a position. The schema page then adds a HYPOTHESIS that the position follows from the slot's entailments: chg-dominant → (a), vol/mov-dominant → (b). Open question T4 asks whether that is true, or whether the position is a lexical diacritic that has to be listed verb by verb.

GRANT'S CONJECTURE The verb's argument positions hold proto-roles (entailment clusters), not lexical items (schema §1). HYPOTHESIS If that is right, a valence operation should be predictable from entailments wherever possible. If the split has to be listed by hand, ADD-CAUSER gives the schema no advantage over the Pronominal Argument Hypothesis (T7).

2. Hale & Platero's data SOURCED

Source. Hale, Ken & Paul Platero 1996. "Navajo reflections of a general theory of lexical argument structure." In E. Jelinek, S. Midgette, K. Rice & L. Saxon (eds.), Athabaskan Language Studies: Essays in Honor of Robert W. Young, UNM Press, pp. 1–14. Opened 2026-10-07 (evening) as the text of the vdoc.pub copy. The page carries no page numbers, so citations here use example numbers. The PDF download sits behind a captcha and was not used. So no page image was checked, and the text ends partway through ex. (21).

How the characters were restored. The copy's font prints ł as ¬, Ł as Ò, and a nasal vowel as ² before the vowel (so y²í²íʼ = yį́į́ʼ). We restored every form below by that one mapping. We checked the mapping against the forms Gessner 2003 reprints from this paper, which were read off page images on 2026-10-06. The rope pair (7) and the ‘walk’ and ‘burp’ causatives (8b), (9b) come out letter for letter as in Gessner (6), (10), (11). Hyphens and glosses are H&P's own. Their abbreviations: SPF, YPF, NPF = perfective types; IMPF; PROG; CI = classificatory.

2.1 Simple transitivity alternation: ł only

H&P 1996, exx. (4), (5), (6), (16)
(4a)Tóshjeehsi-tsʼil.‘The barrel shattered, broke to pieces.’
barrelSPF:3-shatter:PERF
(4b)Łeetsʼaaʼsé-ł-tsʼil.‘I shattered the dish.’
dish3:SPF:1s-ł-shatter:PERF
(5a)Tinyí-yį́į́ʼ. (< -ghį́į́ʼ)‘The ice melted.’
iceYPF:3-melt:PERF
(5b)Yasyí-ł-hį́į́ʼ. (< -ghį́į́ʼ)‘I melted the snow.’
snow3:YPF:1s-ł-melt:PERF
(6a)Kǫʼn-eez-tsiz.‘The fire went out.’
firen-SPF:3-extinguish:PERF
(6b)Kǫʼn-é-ł-tsiz.‘I put the fire out.’
fire3:n-SPF:1s-ł-extinguish:PERF
(16a)Łeetsʼaaʼsi-tsʼil.‘The dish shattered, broke to pieces.’
dishSPF:3-shatter:PERF

The example labels (5) and (6) are run together in the text copy. The ‘melt’ pair comes first and the ‘extinguish’ pair second, so we number them (5) and (6). Each example carries a supporting sentence from Young & Morgan 1980 (e.g. "Y&M 80.804"); those sentences are not reprinted here. The rope pair (7) is on the derivations page, D8. The text prints the stem of (4) and (16) as -tsʼil and the theme in (14) as -ii-ł-tsʼił. Both are recorded as printed.

H&P on the ł: "We take this morphological reflection as being nothing more than that, a simple marking of the transitivity alternation. In particular, there is no “meaning” attached to the ł-classifier." The alternation is the "simple transitivity alternation", and in the intransitive member the classifier is "the so-called zero classifier Ø-".

2.2 Complex transitivization: the causative

H&P 1996, exx. (10), (13)
(10a)ʼAwééʼyi-dloh.‘The baby is laughing.
babyPROG:3-d:laugh:PROG
(10b)ʼAwééʼbi-y-eesh-dloh.‘I am making the baby laugh.’
baby3-y-PROG:1s-ł-d:laugh:PROG†
(13)*ʼAwééʼd-é-saʼ. (< … -ł-zaʼ)‘I burped the baby.’
babyd-SPF:1s-ł-belch

SOURCED † Segmentation note (2026-11-11 audit pass 4): Hale & Platero 1996 (10) prints bi-y-eesh-dloh. with the gloss “3-y-PROG:1s-ł-d:laugh:PROG”: the gloss names more morphemes than the form shows as separate pieces (fused or unwritten morphemes), so the two lines do not match one-to-one in the source. Kept exactly as printed.

The closing quotation mark of the (10a) translation is missing in the text copy. It is recorded as given, not repaired. Exx. (8) ‘walk’ and (9) ‘burp’ appear in Gessner's reprint (D8). (13) is starred in the source and was not re-checked on a page image (none is available). This is the first time the site has checked it against the paper's own text rather than the prior-run note. The form and gloss match.

H&P: the causative has "a position VI prefix -y-" and causee object markers that "differ from ordinary position IV object markers in that the third person is overt (bi-) in the presence of a first or second person subject, as if attached to an incorporated postposition". On (13): "There is no a priori reason why the verb form appearing here should not be possible, since the stem -za’ is perfectly possible with the ł-classifier, as the causative attests. It is nevertheless a fact that this verb cannot form a simple transitive".

2.3 Which verbs alternate

(11), alternating (H&P's "sample", from Young, Morgan & Midgette 1992, YMM page numbers in brackets):

ʼ-(ł-)ʼeeł ‘float away’ [177-83]; ii-(ł-)gááh ‘whiten’ [195]; (ł-)gan ‘dry up’ [199]; ʼ-(ł-)geeh ‘fall away’ [as person, animal; 214,6]; ii-(ł-)kíísh ‘become spotted, put spots on’ [329]; ii-(ł-)k’is ‘crack’ [351]; (ł-)lą́ąh ‘increase’ [in number or quantity; 369]; ʼ-(ł-)lį́ ‘flow away’ [376,7]; di-(ł-)lid ‘be burning’ [371]; ʼ-(ł-)máás ‘roll away’ [397,8]; ii-(ł-)táás ‘bend over, double’ [493]; (ł-)tʼees ‘cook, roast, etc.’ [536]; ʼ-(ł-)tʼééh ‘extend away’ [line, fence; 546,7]; ii-(ł-)tłíísh ‘darken, turn brown’ [571]; (ł-)tłʼis ‘harden’ [as mud, dough; 580]; ii-(ł-)tsóóh ‘yellow’ [614]; di-(ł-)tsʼǫǫd ‘stretch’ [643,4]; di-(ł-)zháásh ‘begin to wear away, down’ [767]; (ł-)zhǫǫh ‘become gentle, make gentle’ [796].

(12), non-alternating ("selected from among non-alternating intransitives having the zero-classifier"):

na-bé ‘swim, bathe’ [69]; -cha ‘cry’ [70]; di-lish ‘spurt urine’ [as of dog; 375]; na-né ‘play’ [423]; ho-taał ‘sing’ [490]; di-zheeh ‘spit’ [771]; ʼ-zhííł ‘gasp, inhale sharply’ [773]; di-yih ‘pant, puff’ [702]; ʼ-yóół ‘inhale’ [723].

Cross-linguistic lists (14)–(15). In English, Miskitu and Navajo alike, these verbs alternate: boil, break, shatter, dry (up), fill, float, melt. These do not: cry, cough, laugh, play, shout, sing, sleep, snore. H&P give Navajo stems for both lists, but the text copy scrambles the column order, so they are not reprinted here. H&P: "the verb classes distinguished by these three languages are identical, for all intents and purposes. It is virtually impossible for this to be an accident."

Their analysis. SOURCED Alternating roots are predicates that need a subject, so their lexical structure contains an internal subject DP (structure (17)). The transitive puts (17) under a higher verbal head that carries ł (structure (18)). The internal subject becomes the object, and "The subject of the transitive, of course, is an external argument, not present in the lexical structure". Non-alternating roots are non-predicates. They have no internal subject, so there is "no source for an object argument", and the ł-structure (20) "is illegitimate". Some non-alternating stems are "cognate" with nouns (ex. (21): ‘laugh’, ‘sleep’, ‘breathe’, ‘inhale’, ‘belch’, ‘spit’; the copy breaks off inside this example). H&P also say that intransitives with an overt classifier (ł, l, d) exist, and "It is not known whether this, in and of itself, inhibits those verbs from participating in the transitivity alternation."

Causatives on other bases. SOURCED The complex causative "can be built upon a verb theme, even a basically transitive one, which contain an inherent classifier". Their examples: OBJ-ʼ-y-ł-dlą́ ‘make OBJ drink something’ [YMM 154], whose base takes the d-classifier; ha-OBJ-y-ł-yeed ‘run OBJ up out’ [657], on the l-classifier theme ha-l-yeed; OBJ-ʼ-y-ł-haazh ‘make OBJ sleep’ [Y&M 87:215], on the ł-classifier theme ʼ-ii-ł-haazh ‘go to sleep’.

VPISH, in their words. SOURCED For the intransitive alternant: "the lexically defined “internal” subject, DP, functions as sentential subject as well—we assume it raises in sentential syntax, out of the verbal projection to a position (presumably, the specifier of I(nfl)) in which it is appropriately licensed (assigned structural case and construed with subject agreement)." They add that tree (18) "does not represent the linear ordering … Navajo being a head-final language". Both points differ from Hale's later work, which licenses arguments in situ (Hale 2003:29) and treats the verb's extended projection as head-initial (Hale 2001:689). The derivations in §7 show both versions.

3. What Dowty says about intransitive classes SOURCED

Dowty 1991, §12 "Proto-roles and intransitives: the unaccusative hypothesis", pp. 605–611. Page images were read at 110 and 170 dpi on 2026-10-07. The printed page equals the PDF page + 545.

  • No single criterion (p. 606, reporting Rosen 1981, 1984): "it is apparently not possible to predict easily (if at all) from a given intransitive verb's meaning whether it will turn out to be unaccusative or unergative in a given language, because no single semantic criterion—volition, agentivity, presentational meaning—or combination of criteria seems to determine this correctly for all verbs." His (73) lists die, sweat, bleed, suffer, be hungry, sneeze, each unergative in one language and unaccusative in another.
  • Two key entailments (p. 607): "the most important for the unergative/unaccusative contrast seems to be volition … Volition necessarily involves sentience, and verbs with both these entailments are always unergative, it seems. The slightly broader presupposition of ‘being predicatable of a human being’ sometimes but not always puts a verb in this class, as does movement." He calls sneeze, bleed, vomit, snore and blush "marginal" for volition and says they vary across languages. On the patient side: "incremental themehood—or, slightly more generally, whether the argument is an incremental or a holistic theme, i.e. whether or not it is telic—seems to be highly significant".
  • Table 1 (p. 607) crosses agentive/non-agentive with atelic/telic. Cell 1 (agentive, atelic) is "definitely unergative". Cell 4 (non-agentive, telic) is "definitely unaccusative". Cells 2 and 3 are "?". Cell 2 includes "stand up and retire", and cell 3 includes "statives like exist and be in the room" (p. 608).
  • Prediction (p. 608): "predicates that are ‘high’ in agentivity and ‘low’ in patient properties are invariably unergative, while those low in agent properties and high in patient properties are invariably unaccusative; only those high in both kinds of entailments, or low in both, should be unstable."
  • Two ways a class can enter a grammar (p. 608). (i) "The distinction is a grammatical one between two classes of intransitive verbs, having a correspondence with some partitioning of the continuum from Proto-Agent to Proto-Patient (though possibly only a rough correspondence). But each individual verb is assigned once and for all to one of the two grammatical classes." (ii) "Certain grammatical constructions have certain meanings associated with them … hence a given intransitive verb is appropriate in such a construction only if it has the right kind of meaning."
  • How to tell them apart (p. 609): under (ii), "the class of predicates permitted to appear in constructions specific to one class can be extended beyond the normal class in certain contexts, for some fanciful, metaphorical, humorous, or otherwise nonliteral effect". Footnote 41 (p. 608): if a verb occurs in both frames and "the occurrence of the verb in the two syntactically diagnostic frames should turn out to correlate with a difference in the verb's interpretation of an agent-vs.-patient sort", the analysis is of type (ii).

HYPOTHESIS For our purposes, the Navajo simple ł-alternation is a diagnostic of the same kind as Dowty's unaccusativity tests: a construction open to one class of intransitives only. H&P say so themselves, calling the non-alternating verbs "sometimes called “unergatives”". So T4 is Dowty's (i)-vs-(ii) question asked of one Navajo construction.

4. The entailments of each verb HYPOTHESIS (ours, from English glosses)

Four properties of the sole argument, taken from Dowty. vol = (27a) volitional involvement. sent = (27b) sentience; here it means the verb requires a living, sentient participant. mov = (27d) movement. tel = Dowty's p. 607 patient-side factor: the participant is an incremental or holistic theme, i.e. the predicate is telic. "±" means the value depends on context, or that Dowty calls the verb "marginal" for that property.

Warning. Every value below is assigned by us from the English gloss that H&P print. Nobody has given a Navajo speaker's judgment on these entailments. Several glosses are underspecified. For instance, ‘float away’ could be telic (leave) or atelic (drift).

Verb (H&P gloss)volsentmovtelAlternates? (source)
shatter (4), melt (5), go out [fire] (6), break [rope] (7), crack, harden, dry up−−−+yes
whiten, darken, yellow, become spotted, bend over, stretch, cook/roast, increase, begin to wear away−−−±yes
float away, flow away, roll away−−+±yes
be burning−−−− (ongoing process)yes
extend away [line, fence]−−− (spatial extent, no motion)−yes
fall away [as person, animal]−+++yes
become gentle−? (gloss silent)−+yes
swim/bathe, play, sing, spit, walk (8), shout (15)++±−no
cry, laugh (10), cough (15), inhale, gasp±+−−no
belch (9), snore (15), sleep (15), pant/puff, spurt urine [as of dog]− or ± (Dowty's "marginal" class)+−−no

5. Which rule fits the sample? HYPOTHESIS

Candidate rule (non-alternating iff …)Fails onVerdict
R-vol: vol ∈ Ebelch, snore, sleep, pant: no volition, yet non-alternating. This works only if Dowty's "marginal" verbs are counted as volitional. That decision is made verb by verb, which amounts to a diacritic.✗ as stated
R-tel: tel ∉ E‘be burning’ and ‘extend away’: atelic, yet alternating. The ± verbs are undecided.✗
R-mov: mov ∈ E (the schema page's "vol/mov-dominant")float/flow/roll away and fall away: motion, yet alternating✗ (mov is dropped from ADD-CAUSER)
R-PA>PP: more Proto-Agent than Proto-Patient entailments (a Dowty-style count)‘fall away [as person, animal]’ has two P-A entailments (sent, mov) against one P-P (tel), so it is predicted non-alternating, but it alternates. ‘float away’ (mov against ± tel) is a tie.✗
Dowty Table 1, agentivity primary: cells 1–2 unergativeThe same as R-vol for belch/snore (cell 3 if non-agentive, predicted "most … unaccusative")✗ unless the marginal verbs count as agentive
R2: sent ∈ E and tel ∉ ENo failure in the sample. ‘fall away [person, animal]’ alternates because it is telic. ‘be burning’ and ‘extend away’ alternate because nothing sentient is involved. belch, snore and sleep do not alternate because they are sentient and atelic.✓ on the sample (but see the posture case below)

The posture case. HYPOTHESIS Dowty's own example for cell 2 (agentive, telic) is stand up. R2 classifies ‘he/she stood up’ (sent, tel) as alternating, which predicts a simple ł-causative. The causative of the ‘stand’ stem printed by Goertz et al. 2006 (O b), biiłzį́ ‘I stood him/her/it up’, begins with bi… even though its subject is first person (keep vs bring about §2.1). So do (M b) binéłdá and (N b) binéłtéézh. In H&P's simple transitives a third-person object with a 1s subject is not overt (4b) sé-ł-tsʼil; in their complex causatives it is overt bi-. Also, H&P segment the first letters of (8b) na-b-ii-sh-ł-á as 3-y, and biiłzį́ begins with the same letters bii. That is a resemblance between strings, not a segmentation: Goertz et al. give none (K1). If the posture causatives are complex, cell 2 behaves like cell 1 and the rule has to put vol back in:

R3   non-alternating  iff   vol ∈ E   ∨   (sent ∈ E  ∧  tel ∉ E)

  cell 1  (vol, atelic)          swim, play, sing, walk          → non-alternating   ✓
  cell 2  (vol, telic)           stand up / sit down / lie down → non-alternating   (IF bi… is the causee bi-; K1, K4)
  cell 3a (sentient, atelic)     belch, snore, sleep, pant       → non-alternating   ✓
  cell 3b (non-sentient, atelic) be burning, extend away         → alternating       ✓
  cell 4  (non-agentive, telic)  shatter, melt, fall away        → alternating       ✓

HYPOTHESIS R2 and R3 agree on every H&P item and differ only on the posture verbs. Both need two of Dowty's entailments plus telicity. Neither needs a list of exceptions. What both add to Dowty is a cut inside his cell 3. Sentient atelic verbs fall on the unergative side, which is where his "marginal" class goes in English and Navajo alike (H&P (15): snore, sleep, cough). Non-sentient atelic verbs fall on the unaccusative side. Dowty's "being predicatable of a human being" (p. 607) is the closest thing he says, and he says it only "sometimes" decides.

6. Tightened rules HYPOTHESIS

The change: ADD-CAUSER stays keyed to position, as on the derivations page. The entailment claim moves into a separate linking rule, so the two can be tested apart.

LINK-1 (first-Merge site of a sole slot)       Σ = {ρ₁}
  (E)  pos_ρ1 = Spec,v   iff   vol ∈ E_ρ1  ∨  (sent ∈ E_ρ1 ∧ tel ∉ E_ρ1)          [= R3; R2 drops the vol disjunct]
       pos_ρ1 = Spec,V   otherwise
  (D)  rival: pos_ρ1 is listed per root (H&P: predicate root → internal subject;
       non-predicate root → no internal subject)

ADD-CAUSER(Σ)   Σ' = Σ ∪ {ρ_C},  E_ρC ⊇ {vol, caus_Δ},  ρ_C external (outer Spec,v = ł)
  (a) pos_ρ1 = Spec,V                    → ρ₁ becomes the object; spell-out ł only           (simple)
  (b) pos_ρ1 = Spec,v (ρ₁ external)       → ρ₁ relinked to Obl; spell-out y + ł + oblique bi-  (complex)
      the base may already have a classifier (ł, l, d); H&P's 'drink', 'run up out', 'go to sleep'
  (c) |Σ| = 2 (transitive base)          → (b) is available at least when the base's object is the
                                             ’- of ‘make OBJ drink something’ (H&P); otherwise analytic
                                             (Kibrik 1993:51 via Gessner; not opened)
  never: (a) on a base with pos_ρ1 = Spec,v                                                  (H&P (13), (20))

C12  Classifier neutrality.  The ł added by ADD-CAUSER(a) contributes no entailment of its own;
     every entailment of ρ_C comes from the operation, and ρ₁ keeps E_ρ1 (plus aff).
     Basis: H&P "no “meaning” attached to the ł-classifier"  (SOURCED); the restatement is ours.

Changes from the schema page, §6: (1) "mov" is dropped as a trigger for case (b), because motion verbs alternate. (2) Case (b) no longer requires a zero-classifier base. (3) Case (c) is no longer simply "rare": H&P print a complex causative on a transitive theme. That the ’- in ‘make OBJ drink something’ is the unspecified object is our reading of the translation (HYPOTHESIS). H&P do not segment it further. (4) MAINTAIN and C10–C11 on the keep vs bring about page are unaffected. LINK-1 sends their stative ‘it’ bases (no vol, no sent, no tel) to Spec,V, and that fits the ł-only stative causatives sełka and séłʼą́, which have no initial bi….

7. Derivations C1–C4, step by step

Conventions are as on the derivations page. Tags: Σ schema step, A word-internal linearization, B clause level. V-H = in-situ licensing (Hale 2003:29). V-P = VPISH raising, which here is H&P 1996's own assumption (raising to Spec,I). Hyphenation follows H&P's segmentation. Nothing is split further.

C1. (4a) Tóshjeeh si-tsʼil ‘The barrel shattered, broke to pieces.’

  1. Σ Σ(√ts’il ‘shatter’, Ø) = {ρ₁}. Eρ1 = {tel, chg}, no vol, no sent (ours). LINK-1 (E): neither disjunct holds, so posρ1 = Spec,V. LINK-1 (D): ‘shatter’ is a predicate root (H&P (17)), so Spec,V. Both agree.
  2. Merge R = -tsʼil with an empty V. ρ₁ is merged as the subject that R needs, inside V's projection. This is H&P (17): V "hosts" both R and its subject.
  3. Incorporate R into V's empty matrix (H&P: "V and R form a single terminal node"). A word-internal.
  4. Merge mode/subject head (SPF). Agree with ρ₁: [3]. The gloss SPF:3 shows no separate 3 exponent. Σ C1: one slot, one (null or fused) exponent.
  5. Move. V-P (H&P 1996, SOURCED): ρ₁ "raises in sentential syntax, out of the verbal projection to … the specifier of I(nfl)", leaving ⟨ρ₁⟩ in Spec,V. V-H (Hale 2003): no movement. The surface string is the same either way.
  6. Spell-out A si ⌢ tsʼil → si-tsʼil. The functional prefix precedes the stem.
  7. B Tóshjeeh is linked to ρ₁ as a clause-level adjunct (C3). Order: DP ⌢ V.

C2. (16b)/(4b) Łeetsʼaaʼ sé-ł-tsʼil ‘I shattered the dish.’ ADD-CAUSER (a)

  1. Σ Input ΣC1 with posρ1 = Spec,V. ADD-CAUSER(a): Σ' = {ρC, ρ₁}. EρC ⊇ {vol, caus_Δ}; ρ₁ gains aff. C10 ✓ (caus_Δ with chg). C12: ł adds nothing.
  2. Merge the C1 structure (V + R + ρ₁) as the complement of a higher verbal head. That head is "supplied with the ł-classifier" (H&P (18)). In site notation, v = ł.
  3. Incorporate successive-cyclically: R → V → v(ł), "yielding ultimately the transitive verb theme -ł-tsʼił" (H&P). A
  4. Merge ρC above the lexical structure, because it is "an external argument, not present in the lexical structure" (H&P). In site notation, Spec,v/Voice.
  5. Σ ρ₁ is now in object position: "This argument will, of course, function now as the sentential syntactic object" (H&P).
  6. Merge mode/subject head (SPF). Agree with ρC: [1s]. ρ₁ is valued [3]. The gloss is "3:SPF:1s": the 3rd-person object has no overt exponent. C4: ρC is 1st person, so VOICE is undefined. C1 ✓.
  7. Move. V-P: ρC to Spec,I. ρ₁ stays as object. V-H: none.
  8. Spell-out A sé ⌢ ł ⌢ tsʼil → sé-ł-tsʼil. Mode/subject precedes the classifier, and the classifier precedes the stem.
  9. B Łeetsʼaaʼ is linked to ρ₁. There is no subject DP. Order: DP ⌢ V.

C3. (10b) ʼAwééʼ bi-y-eesh-dloh ‘I am making the baby laugh.’ ADD-CAUSER (b)

  1. Σ Base Σ(√dloh ‘laugh’) = {ρ₁}. Eρ1 = {sent}, vol ±, no tel (ours). LINK-1 (E): the sent ∧ ¬tel disjunct holds, so posρ1 = Spec,v (external). LINK-1 (D): ‘laugh’ is a non-predicate root with a cognate noun (H&P (19), (21)), so there is no internal subject. Both agree.
  2. Σ ADD-CAUSER(b): Σ' = {ρC, ρ₁}. ρ₁ is relinked to an oblique.
  3. Merge inner verb: V + R -dloh (H&P (19): V with R as complement, no internal subject). Merge ρ₁ as the inner external argument (inner Spec,v₂).
  4. Merge the position-VI prefix y (glossed "y").
  5. Merge outer v = ł, and ρC in its specifier. The gloss shows ł and d together on the stem ("ł-d:laugh"). The surface string has -dloh. How they fuse is not analysed here.
  6. Merge mode/subject head (PROG). Agree with ρC: [1s] → eesh (PROG:1s).
  7. Σ ρ₁ is valued [3] as an oblique, so it is overt bi- even with a 1st-person subject (compare C2, where the object is null). C1 ✓: two slots, two exponents. C4: VOICE undefined. This bi- is the causee marker, not the inverse.
  8. Move. V-P: ρC to Spec,I. ρ₁ is inside the oblique and does not move. V-H: none.
  9. Spell-out A bi ⌢ y ⌢ eesh ⌢ dloh → bi-y-eesh-dloh. Oblique causee ⌢ Qualifier ⌢ mode/subject ⌢ stem. Each functional head precedes its complement, as rule set A requires.
  10. B ʼAwééʼ is linked to ρ₁, the oblique-valued slot. Order: DP ⌢ V.

C4. Why (13) *ʼAwééʼ d-é-saʼ ‘I burped the baby.’ crashes

  1. Σ Base Σ(√za’ ‘belch’) = {ρ₁}. Eρ1 = {sent}; vol is Dowty's "marginal" case; no tel (ours). LINK-1 (E) and (D) both give posρ1 = Spec,v. H&P (21) list di-za’ ‘belch’ among the verbs with noun-cognate stems.
  2. Σ Try ADD-CAUSER(a). Its input condition is posρ1 = Spec,V, which fails. ✗
  3. Forcing it anyway would merge v = ł above V + R with ρC in Spec,v. Then ρ₁ has nowhere to go. Its first-Merge site, the external position, is the one ρC now fills, and the lexical structure has no internal subject that could become the object. This is H&P's (20): the structure is "illegitimate" because it is "transitive" and has no object.
  4. No exponent check catches it. In C2 a 3rd-person object with a 1s subject is also null. So a missing object prefix in d-é-saʼ would not by itself show that anything is wrong. The crash happens at the Σ step (linking), not at spell-out. HYPOTHESIS
  5. The convergent alternative is ADD-CAUSER(b), which gives (9b) ʼAwééʼ bi-di-y-é-saʼ ‘I burped the baby.’ (Gessner (11); D8). H&P note that the stem "-za’ is perfectly possible with the ł-classifier" there. So the block is not on ł + -za’. It is on the simple linking.

HYPOTHESIS C4 crashes under LINK-1 (E) and under (D) alike, so (13) does not decide between them. What does decide is in §9.

What C1–C4 show HYPOTHESIS

  1. The split is made at the Σ step, before any Merge. C2 and C3 differ in where ρ₁ is first merged, and everything else follows: the extra y, the oblique bi-, and whether ρ₁ ends up as object or oblique. That is all rule set A and Σ. Rule set B links one DP in the same way in C1–C3.
  2. VPISH raising does not tell the classes apart. The raised argument is ρ₁ in C1 and ρC in C2 and C3, and each step leaves no trace in the word, as in D1–D5. H&P 1996 assume the raising; Hale 2003 does without it. The strings are identical under both.
  3. The verb's one participant fixes the outcome. If LINK-1 (E) is right, whether a verb takes the simple ł-causative is decided by the entailments of its sole participant, not by any property of a pronoun. That would be the first generalization on this site that is stated over entailments and has no restatement over pronouns. That is exactly what T7 asks for. A PAH account can still state it over a root diacritic, which is (D).

8. Verdict on T4

  • The schema page's hypothesis as stated is wrong in one part. HYPOTHESIS "vol/mov-dominant → complex" fails on motion: ‘float away’, ‘roll away’, ‘flow away’ and ‘fall away [as person, animal]’ all alternate. Mov is out.
  • No single Dowty entailment predicts the split on H&P's sample (vol fails on belch/snore/sleep; telicity fails on ‘be burning’ and ‘extend away’). That matches Dowty's own report of Rosen (p. 606).
  • A two-entailment rule fits the whole sample with no exceptions: non-alternating iff the sole participant is sentient and the predicate is atelic (R2), plus volitional verbs if the posture causatives prove to be complex (R3). HYPOTHESIS All entailment values are ours, from English glosses, and the sample is H&P's illustrative selection, not a census.
  • The entailment rule is not yet shown to beat a diacritic. H&P's own account is structural: predicate vs non-predicate roots, supported by noun-cognate stems (21). Their (13) is starred despite a coherent meaning, but both accounts predict that star. Deciding needs Dowty's (i)/(ii) diagnostics and items outside the sample (§9). So T4 is now partly answered but still OPEN.

HYPOTHESIS Update 2026-10-19 (12:46pm): the two diagnostics on more verbs. The HMP inchoative manuscript prints ‘start to’ forms for five verbs that also have an alternation status: (ł-)gan ‘dry up’ (H&P (11), alternating) shifts (shiʼniigan); ‘spoil’ (alternating, per HMP citing H&P 1996) shifts (biʼniichxʼǫʼ, stem font restored); -cha ‘cry’ (H&P (12), non-alternating) does not shift and takes an inserted ’ (ʼináʼniichah); ‘sleep’ and ‘snore’ keep the subject. That is 5/5 agreement, and LINK-1 fits eight further HMP verbs (HMP page §6, u10).

HYPOTHESIS Update 2026-10-19 (4:46am): a second diagnostic. Hale's inchoative ʼnii- keeps the subject of ‘sleep’ (handout (11)) and ‘snore’ (PA paper p. 60) but turns the participant of ‘get (very) cold’ into an oblique object (PA paper (51), handout (12)). Read as a test of starting position (IN-6 on the ‘start to’ test), it puts ‘sleep’ and ‘snore’ in Spec,v and ‘get cold’ in Spec,V, which is what LINK-1 (E) predicts (sentient + atelic vs a change of state). Three verbs only, and ‘get cold’ is not in H&P's sample; running both tests over the same verbs is T38d.

9. Open questions this page adds

#QuestionWhat would settle it
T4aDo non-sentient atelic verbs of emission (shine, glitter, smell/stink, bubble) alternate simply? LINK-1 (E) says yes, because no sentient participant is involved. A diacritic allows either answer.Young, Morgan & Midgette 1992 or Young & Morgan 1987 entries, checked for (ł-) themes. OPEN
T4bDo sentient but telic, non-volitional verbs (faint, die, wake up, get tired) alternate simply, like ‘fall away [as person, animal]’?Same dictionaries. OPEN
T4cAre the posture causatives biiłzį́, binéłdá, binéłtéézh complex (with a y and a causee bi-) or simple? This decides R2 vs R3. It is the same question as K1/K4 on the keep vs bring about page.A segmentation in Young 2000:239 or Young & Morgan; or 3rd-person-subject versions, where an ordinary object (cf. (7b) kʼí-i-ní-ł-dláád, object glossed "3") and a causee bi- should look different. OPEN
T4dDowty's (ii) test. Can a cell-1 verb (‘cry’, ‘laugh’) take a simple ł-transitive for a joking or nonliteral effect? Does any root have both a simple and a complex causative, with an agent-vs-patient difference in meaning (Dowty n. 41)? A yes to either means a meaning-based constraint; a flat no means a grammatical class.Native-speaker judgment; texts. OPEN
T4eDoes an overt classifier on an intransitive (ł, l, d) block the simple alternation? H&P: "It is not known".A census of YMM (ł-), (l-), (d-) intransitives. OPEN
T4fPage-image check of H&P 1996 (characters restored from a font-garbled text copy; the copy stops in ex. (21)).A library or publisher copy of Jelinek et al. 1996, pp. 1–14. OPEN

Who stays the subject when you add ‘start to’? The Navajo inchoative as a test of where a one-participant verb's participant starts out (T37b)

2026-11-11 audit pass 3. Local box file paths were removed from this page; files inside the repository are given by repo-relative path, and the source citation next to each removed path stays as the reference. See the pass 3 report.

2026-11-11 audit pass 6. Derivation ID1 said “The predicate is telic.” That is our reading of the English gloss ‘get very cold’; no Navajo telicity test has been run. It now says so, and telicity is marked OPEN. See the pass 6 report.

Working note. Kept as written (URL, data and status labels unchanged). Filed under Merge and Agree inside the word; the consolidated argument is the essay merge-agree. Rule IDs: rules index.

Restated on the unified structure (2026-11-15). Voice, causatives and events are restated with every theta role assigned inside vP: the classifier spells a value of v⁰ (H59, distribution NOT UNDERSTOOD), a causer is an external argument in Spec,vP, and the passive keeps a silent indefinite agent (H10b). This page’s causative analysis has not yet been restated on the unified structure (OPEN). Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. 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. Position-class template statements below describe the order of prefixes; they are not the derivation (§3). Derivations below in which the verb or subject does not raise are labelled contrast; the working default raises V→v→T and the subject to Spec,TP. Status on the spine: see unified structure, §4, causatives 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.

On subject on the head §5, rule SH-8 was left OPEN. A one-place verb whose single participant starts low (in Spec,V, Hale's “inner specifier”) is not in the position the Mode/Subject head (M/S) governs, yet its person and number are pronounced on M/S (yishdlóóh ‘I get very cold’). Either the participant moves up inside the word (U-1, VPISH-type raising) or M/S simply takes the closest participant (U-2). Task T37b asked for a test. This page uses one construction that two Hale texts on the box already discuss but this site had never used for this purpose: the inchoative ‘start to …’ prefix ʼnii- (Young & Morgan 1987:187–188, as cited by Hale). Adding it to an unaccusative verb takes the single participant out of the subject role and turns it into an oblique object. With a transitive or an unergative verb, the subject stays where it is.

Update 2026-10-19 (12:46pm): the Hale, Munro & Platero inchoative manuscript (hale028) has now been read (text only) on ‘start to’ in HMP's own words. Its (10a) shináʼniidlóóh ‘I start to get cold (about dawn)’ puts the unaccusative participant's shi- before the disjunct ná-. That answers T38a and contradicts ID4's version of eccentric agreement; HMP's own version survives by “bumping”. HMP place the ’ of ʼnii- in position IV as a “thematic” object, which answers T38b and leaves IN-7 (I) without support. IN-7 (D), a default non-local Mode with no added element, is now favoured. IN-5 is replaced by IN-5′ (BUMP), and IN-8/IN-9 add the object requirement and the role-less placeholder for verbs with nothing below v.

Ground rules (Grant's). Hale's PA paper, his conjunct-verb handout, Hale & Keyser, Willie 1991 and this site's own rules are each treated as possibly partly right. A template position or a gloss label is an analysis, not data. Only contradicting data count against a hypothesis, and silence counts against nothing. Grant does not speak Navajo. Every Navajo form below is copied from an opened source, read on page images, with page and example number. Forms are segmented only where the source prints a segmentation. No form, gloss or judgment is invented. Where a test needs a form no opened source prints, schematic letters are used. Dowty's proto-roles are the baseline, and “role in the action” is used for his entailments. Theta roles and event-structure roles are treated as compatible descriptions of the same participants.

Notation. This is the site's convention. The Syntax · Semantics · Interface bot's BRIEF.md and VERB-QUEUE.md were re-copied on Oct 18 at 10:05pm PT (new file times), but their content has no notation conventions for this site: the brief is still the 2026-09-29 mandate, and the queue's last tick is 2026-10-01. Σ = role-schema step. A = word level (cycle D1 = the MVS). B = clause level (D2 and up). Spine as on subject on the head §5: V < v < M/S (= T) < Qual < (DETR) inside D1. ρ = a pronoun slot ⟨Eρ, posρ, φρ, (δρ)⟩ (proto-role schema; δ from SH-0). CYC = A/B as spell-out §3: C1 non-heads before heads; C2 non-heads higher-first; C3 heads higher-first. PORT = SH-2 (M/S pronounces the φ of the slot it governs). P∅ = the null postposition that Hale says carries oblique case. Status labels: SOURCED HYPOTHESIS GRANT'S CONJECTURE OPEN.

Bottom line, in plain words

SOURCED The facts. Hale prints three pairs of plain and ‘start to’ forms, all with a first-person-singular participant. ‘I tickle him’ Yishhozh → ‘I start to tickle him’ Biʼniishhóósh: sh ‘I’ stays in the subject slot, and the tickled one becomes an overt oblique bi-. ‘I sleep’ ʼAshhosh → ‘I start to sleep’ ʼIʼniishháásh: sh stays (handout pp. 6, (10)–(11)). The same holds for ‘I start to snore’ ʼiʼniishhą́ą́h (PA paper p. 60). But ‘I get (very) cold’ yishdlóóh → ‘I start to get very cold’ shiʼniidlóóh: the ‘I’ is no longer in the subject slot. It shows up as the oblique object shi-, “preceding the 3i object brought in by the inchoative morphology” (PA paper p. 60, (51); handout (12)). In the future, shidíʼnóodlóół ‘I will start to get very cold’, Hale reads the rounded [óo] as a third-person exponent in Mode/Subject (p. 61).

HYPOTHESIS What this shows. The ‘start to’ prefix is a test for where a single participant starts out. A participant that starts in the agent position (Spec,v) stays the subject. One that starts lower (Spec,V) is turned into an oblique and loses the subject slot. So the low participant's subjecthood in yishdlóóh is derived, not fixed by where it starts, and the ‘start to’ prefix can take it away. On the three verbs printed, the test sorts verbs exactly as the site's earlier causative test does (LINK-1 on plain ł-causer): ‘sleep’ and ‘snore’ (sentient, no endpoint) start high, ‘get cold’ (a change of state) starts low.

HYPOTHESIS Raise or take the closest? (T37b). No printed form rules out either option. Hale's own two texts take opposite sides. The handout draws the low participant moving up (trees (13), (14)) and calls the ‘start to’ case “eccentric agreement”. The PA paper says raising is neither “required, or even possible” in a pronominal-argument language (p. 57) and rejects both a trace and eccentric agreement in this very case (p. 61). The closest-slot option (U-2) handles both the plain and the ‘start to’ forms with one rule and no movement. It also suggests a new reading of Hale's puzzle (ours): the third person that Mode shows in the ‘start to’ future may be the ‘start to’ prefix's own indefinite ’ slot, the nearest one left once the low participant has become an oblique. Then no separate empty “expletive” subject is needed. Raising (U-1) needs a movement inside the word, a reason the movement stops in the ‘start to’ form, and still something to give Mode its third person. So U-2 is favoured on economy and on the sources' own framework statements, not on contradicting data. T37b is partly answered.

OPEN What would settle it. (1) A Young & Morgan ‘start to’ form of an unaccusative verb with a disjunct prefix (náá-, da-, a preverb). Hale's transitive example (23a) Bináádaʼniilghal puts the oblique bi- before náá- and da-. If the unaccusative's shi- does the same, it is an oblique (PA paper). If it stays next to ʼnii-, the handout's eccentric agreement is back in play (T38a). (2) Whether the ’ of ʼnii- is a subject or an object in the unaccusative case (T38b). (3) Running the ‘start to’ test over the whole causative sample (T38d).

On this page: 1 Sources · 2 The problem · 3 Data · 4 What the sources claim · 5 Formal rules · 6 Four analyses tested · 7 Derivations ID1–ID8 · 8 What changes on the site · 9 Gets right, misses · 10 Grant's thesis, Dowty, event structure, STRATA, the changeover · 11 Tasks T38a–h

1. Sources opened this run (2026-10-19, 4:46am PT)

SourceWhat was read, and how
Hale, Ken. “On the significance of Eloise Jelinek's Pronominal Argument Hypothesis” (ms.; the site's “Hale 2003”; box copy [local working copy, not in the repository], born-digital, 72 pp.)Newly read on page images: pp. 57–58, 60–61 (end of §4, the Mohawk/Hopi unaccusatives and the Navajo inchoative, (50), (51a, b) and the future form), PDF = printed page, at 110 dpi; p. 59 (opened 2026-10-17 for (50) on postposition units) re-read on the image; 300 dpi crops of (50) and (51); 600 dpi crops of ʼiʼniishhą́ą́h and shidíʼnóodlóół. p. 6 ((4), the same sentence as (50)) read on the image. The text layer maps accented vowels to digits (“dl00h”, “shid7’n0odl00[”), and was used only to confirm that the first vowel of ʼiʼniishhą́ą́h is unaccented.
Hale, Ken. “The Phonological Aspect of the Navajo Conjunct Verb System.” Handout, MIT, undated (cites McDonough 2000) ([local working copy, not in the repository]; a scan)Newly read on page images: p. 6, (10)–(12) the inchoative of transitive, unergative and unaccusative bases with position lines; p. 7, trees (13) Yishdlóóh and (14) “Eccentric Agreement”; p. 9, (23a, b) and tree (23c) (p. 9 had been read for (21)–(22) only); p. 1 (2) template re-checked. 300 dpi crops of every form quoted.
Hale, Ken 2001. “Navajo verb stem position and the bipartite structure of the Navajo conjunct sector.” LI 32 (hale2001.pdf)Newly read on a page image: p. 683 with n. 4 (PDF p. 6): object inflections in the inchoative are “sometimes … conjunct …, sometimes … disjunct”. Reference list: Hale, Munro & Platero 2000.
Hale & Keyser, “The basic elements of argument structure” (ms., hale012, born-digital)Text layer, PDF p. 21: the unaccusative's specifier “raises from its ‘internal’ specifier position to function as grammatical subject in sentential syntax” (English, Miskitu); Navajo intransitive alternants are zero-derived (the Ø-classifier), with the ł-classifier transitive.
Hale 1997 workshop notes (hale023, born-digital)Text layer, PDF p. 60: (24)-type verbs are structurally “unaccusative” and (26)-type verbs “unergative”, which are “fundamentally transitive”; Pam Munro's point that Navajo is semantically closer to “active” systems is to be investigated through the inchoative. n. 27 (text layer): the causee prefix of a causative “sometimes follows the interative [sic] ná-”, a variability to be “taken up in another part of this essay dealing with the inchoative” (its two forms are in a damaged font, not reproduced).
Used as recorded on the siteWillie 1991 p. 23 (“no movement to and from A positions”, Willie's framework); H&P 1996 lists and LINK-1, Dowty 1991 §12 pp. 605–611 (plain ł-causer); Hale 2003 n. 11, n. 13; handout (21b), (22) (subject on the head, A/B as spell-out).
Not openedYoung & Morgan 1987 (pp. 187–188, 217, 781, cited by Hale for every form here); Hale, Munro & Platero 2000, “The Navajo inchoative verb form”, MITWPL, pp. 97–116 (not on the box; it is the source of “eccentric agreement”). Nothing is attributed to them beyond what Hale reports.

2. The problem

SH-8 on subject on the head said: a labile intransitive's single slot is first-merged in Spec,V (Hale 2003 (22)), not in the specifier of M/S's complement, because v intervenes. Hale says it “must be close enough to Mode”. U-1: it raises to Spec,v inside D1. U-2: PORT's locality is read as “the closest slot in M/S's c-command domain”. “Same strings; different predictions only for movement diagnostics.” This page looks for a construction in which the two come apart, or at least in which the low slot's subjecthood can be removed. The inchoative is such a construction. It also bears on three older questions: T4 (is the unaccusative/unergative split predictable from entailments?), T29f's “oblique half” (why oblique objects come leftmost) and T32g (the “unidentified Hale paper on the inchoative”).

3. Data SOURCED

3.1 Hale's PA paper, pp. 6, 59–61

Hale prints a segment-plus-category line under (50) and (51). It is copied exactly, including two characters of his font: ¬ stands where the ł-classifier belongs (Hale & Keyser's ms. prints the ł-classifier the same way, “¬-classifier”), and the stem in (50) is printed “ghoosh”. Our Leipzig line relabels his categories only: Obj(obl) → OBL with the person from the gloss; INCH → INCH; M/S → M/S; v → CL; Obj(’) → INDF. The stem gloss comes from the translation.

Ex., pageForm (as printed)Hale's segment / category lineLeipzig (ours, relabelled)Translation (Hale's)
(4), p. 6ʼAwéé bi-ʼnii-sh-hóósh.baby 3-INCH-1s-ticklebaby 3.OBL-INCH-1SG.SBJ-tickle‘I start to tickle the baby.’
(50), p. 59 (= (4) repeated)ʼAwééʼ bi-ʼnii-sh-hóósh.baby Obj(obl)-INCH-M/S(sh)-v(¬)-V(ghoosh); baby 3-INCH-1s-ticklebaby 3.OBL-INCH-M/S.1SG-CL-tickle‘I start to tickle the baby.’
p. 60 (in prose)ʼiʼniishhą́ą́h— (prose: “both 3i object forms appear”)— (not segmented by Hale)‘I start to snore’
(51a), p. 60 (Y&M 1987:781)yishdlóóhM/S(sh)-v(d)-V(dlóóh)M/S.1SG-CL-get.cold‘I get very cold.’
(51b), p. 60 (cf. Y&M 1987:217)shiʼniidlóóhObj(obl)-Obj(’)-M/S(?)-v(d)-V(dlóóh)1SG.OBL-INDF-M/S(?)-CL-get.cold‘I start to get very cold.’
p. 61 (in prose)shidíʼnóodlóół— (prose: the rounded long vowel [óo] reflects a third-person element in M/S)— (not segmented by Hale)‘I will start to get very cold.’

Hale's prose, quoted. p. 59: “The mark of the inchoative is the complex prefix -’nii- consisting of the 3i indefinite object prefix ’- followed by the qualifiers n- and -ii-. Since the object position is filled, the object of the basic verb (cf. yishhozh ‘I tickle him/her’) must be assigned oblique case (this being traditionally regarded as a null postposition …). The overt bi- in (50), in the presence of a local person subject …, is explained in this way (a third person direct argument would otherwise be itself null in the presence of a local subject).” p. 60: “In the case of an unergative verb, the inherent indefinite object would appear as such, overtly, since the 3i object is ’- in both direct and oblique object forms.” “Since there is no inherent object, the sole argument being the inner specifier (ordinarily the subject, as in (51a) below), the question is what happens to this sole argument. Does it appear as a subject or as an oblique object?” “The single argument of the unaccusative shows up as an oblique object, preceding the 3i object brought in by the inchoative morphology.” “Is that the trace of movement, is this ‘eccentric agreement’ (as suggested in Hale, Munro, and Platero 2000), or is it an expletive subject?” p. 61: “By hypothesis, it cannot be a trace, since the trace would c-command its antecedent (subject being higher than the object in underlying structure). And it cannot be eccentric agreement, because there is no ‘agreement,’ in any usual sense, in a pronominal argument language.” If the future and progressive reveal a third-person exponent in M/S, “then there must be a third person element, presumably expletive, accounting for the rounded long vowel [óo] (tone irrelevant) in the future forms of such inchoatives”. p. 57 (on Mohawk unaccusatives, applied to the PA Hypothesis): “a direct argument, whether pronominal (the usual case) or a bare nominal, is licenced in situ, hence no raising (for case reasons or to satisfy the EPP) is required, or even possible.”

3.2 Hale's handout, pp. 6, 7, 9

Position numbers are Young & Morgan's as Hale gives them in the handout's (2): I preverbs, II iterative, III distributive plural (disjunct); IV direct-object pronouns, V deictic subject, VI qualifiers, VII mode/aspect, VIII subject pronouns, IX classifier (voice), X stem (conjunct). His (3) labels V–VI as E and VII as M. Rows are copied as printed; “—” in (12b) is his.

Ex.FormPositions (as printed)Leipzig (ours)Translation
(10a)Yishhozh.IV Ø- · VII Ø- · VIII sh- · IX ł- · X -ghozh3.OBJ-M-1SG.SBJ-CL-tickle‘I tickle him.’
(10b)Biʼniishhóósh.IV bi- · V–VI ’nii- · VII Ø- · VIII sh- · IX ł- · X -ghóósh3.OBL-INCH-M-1SG.SBJ-CL-tickle‘I start to tickle him.’
(11a)ʼAshhosh.IV ’- · VII Ø- · VIII sh- · IX ł- · X -ghoshINDF.OBJ-M-1SG.SBJ-CL-sleep‘I sleep.’
(11b)ʼIʼniishháásh.IV ’- · V–VI ’nii- · VII Ø- · VIII sh- · IX ł- · X -ghááshINDF.OBL-INCH-M-1SG.SBJ-CL-sleep‘I start to sleep.’
(12a)Yishdlóóh.VII Ø- · VIII sh- · IX d- · X -dlóóhM-1SG.SBJ-CL-get.cold‘I get cold.’
(12b)Shiʼniidlóóh.IV shi- · V–VI ’nii- · VII Ø- · VIII — · IX d- · X -dlóóh1SG.OBL-INCH-M-(none)-CL-get.cold‘I start to get cold.’
(23a, b) (Y&M87:187)Bináádaʼniilghal.IV/I bi- “it” · Ic náá- “more” · III da- “pl” · VI ’nii- “Inch” · VII (´)- “perf” · VIII iid- “1nsg” · IX d-ł- “voice” · X ghal “eat-meat”3.OBL-more-DISTR-INCH-PFV-1NSG.SBJ-CL-eat.meat‘We started to eat some more if [sic] it (meat).’

(23) heading, as printed: “The Object of a verb in the Inchoative aspect (-’nii-) is assigned an oblique case P):”. Tree (23c) (described): E = ʼnii (VI) on top; M = (´) (VII); D = iid (VIII); v = dł (IX); inside V′ three adjuncts, from highest to lowest: [P b P], náá (Ic), da (III), then V ghal. Tree (13) Yishdlóóh ‘I get cold, freeze to death’ (described): DP (Subj) (shí) in Spec,MP; M = Ø sh; DP (Subj) (shí) [trace] in Spec,vP; v = d; DP [trace] as sister of V dlóóh. That is, the participant starts next to V and moves up twice. Tree (14), titled “Eccentric Agreement”, Shiʼniidlóóh ‘I start to get cold’ (described): DP (Subj) (shí) in Spec,EP; E = sh (IV) + ʼnii (VI); M = Ø —; DP (Subj) [trace] in Spec,vP; DP [trace] next to V. That is, the participant moves up three times and is pronounced on E instead of M.

Two labelling details, copied, not interpreted as data. (i) In (10b)–(12b) ʼnii- is labelled “V–VI”, which puts its ’ in position V, the deictic-subject position. In (23b) it is “VI” alone, and the PA paper calls the ’ an “object prefix”. (ii) The handout writes (12b)'s first prefix as position IV and (23b)'s bi- as “IV/I”: with náá- and da- present, the oblique stands before both disjuncts.

3.3 Supporting statements

  • Hale 2001:683 n. 4 (page image): “the object inflections in the inchoative (see Young and Morgan 1987 and Hale, Munro, and Platero 2000) appear sometimes as conjunct inflections, sometimes as disjunct inflections, depending on various factors.” The same note says position IV objects are “at the left edge” of the conjunct domain, “and the idea that they are in the disjunct sector is not entirely out of the question”.
  • Hale & Keyser hale012, PDF p. 21: unaccusatives are “composite lp-dyadic” verbs whose specifier is the “internal argument”. “Under appropriate circumstances … this argument raises from its ‘internal’ specifier position to function as grammatical subject in sentential syntax.” Navajo intransitive alternants show “zero-derivation”, the transitive adds the ł-classifier.
  • Hale 1997, PDF p. 60: the (24)-type verbs are structurally “unaccusative”, since “their syntactically active DP argument is an internal argument, a specifier, as evidenced by the ability of these verbs to undergo simple transitivization”. Verbs of the (26) type are “unergative”, and “fundamentally transitive”. “Pam Munro has pointed out, correctly, that the Navajo verb system represents, from the semantic perspective, a system more closely akin to the ‘active’ agreement grammars … This is an aspect of Navajo which will be investigated in the context of a study of the Navajo ‘inchoative’ construction.”
  • Willie 1991:23 (as recorded): in a pronominal-argument language there is “no movement to and from A positions”.
  • H&P 1996 (15) (as recorded on plain ł-causer §4): ‘snore’ and ‘sleep’ are among the non-alternating (“unergative”) verbs.

4. What the sources claim, side by side

QuestionHandout (10)–(14), (23)PA paper pp. 57–61Hale & Keyser p. 21Willie 1991This site before today
Where does the single participant of ‘get cold’ start?Next to V (tree (13))“the inner specifier”internal specifier— (no positions)Spec,V (SH-8; LINK-1 Spec,V for chg)
How does it become the subject in yishdlóóh?Moves to Spec,vP, then Spec,MP“licenced in situ”; “no raising … required, or even possible” (p. 57)“raises … to function as grammatical subject” (English, Miskitu)no movement to/from AOPEN (U-1 vs U-2)
What happens under ʼnii-?Moves to Spec,EP and agrees with E (“eccentric agreement”); M/S “—”Oblique object; M/S has “a third person element, presumably expletive”——— (not discussed)
Object of a transitive base under ʼnii-Oblique P object (23); P adjoined above the disjuncts in (23c)Oblique, null P (p. 59)——(50) cited on postposition units as a null-P oblique
Evidence offered—Rounded [óo] in the future = 3rd person in M/S (p. 61)———

5. Formal rules

IN-0 INCH-SHAPE   INCH = ’ + n + ii. The qualifiers n-, -ii- are heads in the Qual layer
                  (Hale's E). The ’ is the exponent of a slot rho_I that INCH adds:
                  rho_I = <E = (none), pos = the direct-object position, phi = 3 (zero),
                  delta = INDF>.
                  (SOURCED: the shape, Hale PA p. 59; E = 'nii, handout (14), (23c).
                  rho_I's empty E and its delta are ours: no translation shows a participant.)
IN-1 OBJ-TAKE     rho_I occupies the verb's direct-object licensing position.
                  (SOURCED wording: "Since the object position is filled", p. 59.)
IN-2 OBL          Under INCH, every OTHER slot first-merged below v (rho_X in Comp,V, a sole
                  slot in Spec,V, an unergative's inherent object) is licensed only inside
                  [PP rho P_null]. The slot in Spec,v is untouched.
                  (SOURCED as Hale's generalization over (50)/(10b), (11b)/'i'niishhą́ą́h,
                  (51b)/(12b); stating it over first-merge positions is ours.)
IN-3 PP-OPAQUE    A slot inside a PP is not a goal for M/S.  (HYPOTHESIS; standard.)
IN-4 PORT-CLOSEST (= U-2; refines SH-1/SH-2, provisionally resolves SH-8)
                  M/S pronounces the phi of the CLOSEST eligible slot it c-commands:
                  the slot in Spec,v if there is one; otherwise the highest non-PP slot.
                  A delta left over goes to the highest nuclear head (SH-4, unchanged).
                  (HYPOTHESIS. With a filled Spec,v this is exactly SH-2.)
IN-5 OBL-HIGH     The INCH oblique PP is merged as the highest non-head of VP, above the
                  disjunct adjuncts; it is spelled out as its own domain (pronoun ⌢ P_null)
                  and then ordered by C2.  (HYPOTHESIS, read off handout tree (23c);
                  WHY it merges there is OPEN; the preverb competition of Hale n. 13 is
                  not derived.)
IN-6 DIAG         A sole slot is first-merged in Spec,v iff it stays the M/S subject under
                  INCH.  (HYPOTHESIS; a theorem of IN-2 + IN-4.)
IN-7 THREE WAYS   In (51b), M/S's third person comes from
                  (E)  an expletive: a role-less phi = 3 in Spec,v (Hale PA p. 61), or
                  (I)  rho_I itself, by IN-4 (ours), or
                  (EA) nothing: the participant raises to Spec,EP and agrees with E
                       (handout (14)); M/S is empty and needs a default for the future.
RuleRests onStatus
IN-0, IN-1Hale PA p. 59; handout (10b)–(12b), (14), (23)SOURCED shape and wording; the content of ρI is HYPOTHESIS
IN-2(50), (10b), (11b), ʼiʼniishhą́ą́h, (51b), (12b), (23a)SOURCED pattern; the positional statement is HYPOTHESIS
IN-3standard; consistent with every oblique hereHYPOTHESIS
IN-4SH-2 + SH-8 (U-2); Hale PA p. 57HYPOTHESIS; new; favoured over U-1, not proved
IN-5handout (23c); Hale 2001 n. 4; Hale 2003 n. 13 (as recorded)HYPOTHESIS; partial answer to T29f's oblique half. Replaced 2026-10-19 (12:46pm) by IN-5′ BUMP, which adds HMP's cause (the object position is full) and the blocking case (HMP page §5)
IN-6derived; three verbsHYPOTHESIS; agrees with LINK-1 on all three
IN-7 (I)IN-0 + IN-4 + SH-4; Hale 2001 (22b) as a modelHYPOTHESIS; new. Demoted 2026-10-19 (12:46pm): HMP put the ’ in IV as a thematic object, and the handout's “V–VI” names E; IN-7 (D) favoured

6. Four analyses tested against the data

Analyses. A1 = U-1 with eccentric agreement (handout (13), (14)). A2 = bare U-1: raising in (51a), no raising in (51b) because the participant is already licensed by P∅, plus (E) or (I) for M/S. A3 = the PA paper: no raising, an expletive in M/S (IN-7 (E)). A4 = this page: IN-4 (U-2) with IN-7 (I).

Tests. t1 in (51b)/(12b) the participant is pronounced as shi-, not in M/S. t2 M/S shows a third person in the ‘start to’ future shidíʼnóodlóół (Hale p. 61; Hale's reading of the vowel, which we cannot check without Young & Morgan). t3 transitive and unergative bases keep the subject in M/S ((50), (10b), (11b), (23a), ʼiʼniishhą́ą́h). t4 one analysis covers the prefix before ʼnii- in (10b)/(50)/(23a) and in (12b)/(51b): both are spelled as object-series pronouns in position IV, and the PA paper glosses both “Obj(obl)”. t5 (23a): the oblique precedes náá- and da-. t6 (framework, not data) no A-movement inside the word (Willie p. 23; Hale PA p. 57). t7 (economy) no element beyond the printed morphemes.

Analysist1t2t3t4t5t6t7
A1 handout (raise + eccentric agreement)✓ (its own label: sh on E)✗ unless M/S gets a default 3rd (an added assumption that is A3's expletive under another name)✓✗ two analyses for one spelled position (P object in (23c), raised subject in (14))— transitive only; for unaccusatives it predicts the reverse (T38a)✗ (three movements)✓
A2 bare U-1✓✓ only with (E) or (I)✓✓✓ with IN-5✗ (one movement in (51a))depends on (E)/(I)
A3 PA paper (in situ + expletive)✓✓✓✓✓ with IN-5✓✗ one role-less element not printed
A4 this page (IN-4 + IN-7 (I))✓✓ (ρI's φ = 3)✓✓✓ with IN-5✓✓ (the ’ is printed)

HYPOTHESIS Verdict. No printed form contradicts any of the four analyses outright: all four produce every string. A1 costs the most. It needs three movements inside the word, a default third person for the future, and two different analyses for the same spelled prefix. Its one clear prediction (the participant's prefix stays next to ʼnii- even when a disjunct intervenes) is untested (T38a). A3 and A4 differ only in where M/S's third person comes from: a separate expletive, or the ‘start to’ prefix's own indefinite slot. A2 needs (E) or (I) anyway, so the inchoative adds nothing in favour of raising. For T37b: the inchoative shows that a low participant's subjecthood is derived and can be bled. Both U-1 and U-2 allow that. U-2 covers (51a) and (51b) with one rule (IN-4) and no movement, and matches the PA paper's own statement (p. 57) and Willie's “no movement to and from A positions”. U-1 matches Hale & Keyser's English/Miskitu statement and the handout's trees. So U-2 is favoured on economy and on the Navajo sources' framework statements, not on contradicting data. T37b is partly answered, and the decisive tests are T38a–c.

Update 2026-10-19 (12:46pm): t5 now has an unaccusative datum. HMP (10a) shináʼniidlóóh has the participant before ná-, so A1 as stated here (ID4) is contradicted. A five-analysis re-run with HMP's own version (A1h) and a new A5 (default Mode) is on ‘start to’ in HMP's own words §6.

7. Derivations ID1–ID8

ID1. (51a) yishdlóóh ‘I get very cold’, U-2 (IN-4)

  1. Σ Σ(√GET.COLD) = {ρ1}. Eρ1 ∋ chg (the participant undergoes a change of state: Dowty's Proto-Patient (28a)). No vol. We take the predicate as telic (our reading of the English gloss; no Navajo telicity test has been run, so telicity is OPEN). LINK-1 (E): neither vol nor (sent ∧ ¬tel) → posρ1 = Spec,V. φρ1 = 1sg.
  2. A Merge V dlóóh; ρ1 in Spec,V (Hale's “inner specifier”). Merge v = d with no specifier. Merge M/S (imperfective, Ø per handout (12a) VII).
  3. A Agree / PORT (IN-4): Spec,v is empty; the closest eligible slot in M/S's c-command domain is ρ1 → M/S pronounces 1sg sh. No movement.
  4. A Spell-out D1: N = ∅; H = {M/S sh, v d, V dlóóh} by C3 → sh ⌢ d ⌢ dlóóh ✓ (Hale's line M/S(sh)-v(d)-V(dlóóh)). Neither of Hale's lines lists the initial y(i); we do not segment it.
  5. B Nothing to link.

ID1′. The same form under U-1 (handout tree (13))

  1. Σ As ID1.
  2. A Merge as ID1. Move ρ1: Spec,V → Spec,v (A-movement inside D1). (The handout moves it once more, to Spec,MP.)
  3. A PORT under strict SH-1 LOC: ρ1 is now in the specifier of M/S's complement → sh on M/S.
  4. A Spell-out D1: the same string. The two derivations differ only in the movement step, which nothing in the string shows.

ID2. (51b) shiʼniidlóóh ‘I start to get very cold’, A4 (IN-4 + IN-7 (I))

  1. Σ Σ as ID1. The participant's role in the action is unchanged: it still undergoes the change.
  2. A Merge V dlóóh, ρ1 in Spec,V. INCH (IN-0, IN-1): ρI = ⟨E none, φ = 3, δ = INDF⟩ in the object position. IN-2: ρ1 is first-merged below v and is not ρI, so it is licensed in [PP ρ1 P∅].
  3. A Merge v d (no specifier), M/S, Qual [n ii].
  4. A PORT (IN-4): Spec,v empty; the PP is opaque (IN-3); the closest eligible slot is ρI → M/S takes φ = 3 (zero in this mode; Hale's rounded [óo] in the future, ID8). DELTA-HOST (SH-4): δ = INDF is left over and left-adjoins to the highest nuclear head, Qual: [’ + n ii].
  5. A Spell-out: the PP is its own domain: shi ⌢ P∅ → shi. D1: N = {PP}; H = {[’ n ii], M/S Ø, v d, V dlóóh}. C1, C3 → shi ⌢ ’ ⌢ nii ⌢ Ø ⌢ d ⌢ dlóóh ✓ (Hale's line Obj(obl)-Obj(’)-M/S(?)-v(d)-V(dlóóh)). Hale's “?” is ρI's φ.
  6. Comparison. The output has the same make-up as Hale 2001 (22b) shiʼdooʼniid ‘someone told me’ (SD4): object, indefinite ’ before the qualifier, third person in M/S. In (22b) Hale calls the ’ “the grammatical subject”. On A4 the same holds in (51b), and the handout's “V–VI” label for ʼnii- fits that reading (T38b).

ID3. (51b) under A3 (the PA paper's expletive)

  1. Σ As ID2.
  2. A Merge as ID2, plus an expletive ρEXP = ⟨no E, φ = 3⟩ in Spec,v.
  3. A PORT (strict SH-1): ρEXP → 3rd person on M/S. ρI is an ordinary object (non-head) → ’.
  4. A Spell-out: N = {PP, ρI}, C2 → shi ⌢ ’; H → nii ⌢ Ø ⌢ d ⌢ dlóóh. Same string ✓. Cost: one role-less element that no morpheme spells.

ID4. (51b) under A1 (handout tree (14), “eccentric agreement”)

  1. Σ As ID2.
  2. A Merge V + ρ1; Move ρ1 → Spec,v → (Spec,MP) → Spec,EP. M = Ø, subject “—”.
  3. A Agree E ⟶ ρ1: 1sg pronounced on E, before ʼnii (handout labels it IV).
  4. A Spell-out: sh + ʼnii ⌢ Ø ⌢ d ⌢ dlóóh → shiʼniidlóóh ✓ (with an epenthetic i).
  5. Costs. Three A-movements inside D1, against Willie p. 23 and the PA paper p. 57 (framework claims, not data). In the future, M/S needs a third-person default (Hale p. 61). A prediction: with a disjunct prefix the 1sg should stay next to ʼnii (shape: X ⌢ shi ⌢ ’nii …), whereas on A3/A4 it should come first (shi ⌢ X ⌢ ’nii …, as bi ⌢ náá ⌢ da in (23a)). Schematic only; T38a. Update 2026-10-19, 12:46pm: tested and contradicted by HMP (10a) shináʼniidlóóh (HD2).

ID5. (50) ʼAwééʼ bi-ʼnii-sh-hóósh ‘I start to tickle the baby’: transitive base, and the noun

  1. Σ Σ(√TICKLE) = {ρA (the tickler: volition, causation; Proto-Agent), ρX (the one tickled: affected; Proto-Patient side)}. ρA = 1sg in Spec,v (VPISH); ρX = 3 in Comp,V.
  2. A Merge V + ρX; INCH: ρI in the object position (IN-1); IN-2: ρX → [PP ρX P∅]. Merge v = ł with ρA in Spec,v; M/S; Qual [n ii].
  3. A PORT: ρA is in Spec,v → 1sg sh on M/S (IN-4 = SH-2 here). ρI's φ is not taken, so its δ is not stranded (SH-5): ρI is an ordinary non-head, pronounced ’.
  4. A Spell-out: PP domain: pronoun ⌢ P∅ → bi. The third person is overt because it is an oblique. Handout (22)(ii) and Hale p. 59: a direct 3rd-person object would be Ø under a local subject, as in (10a) Yishhozh with IV Ø (SH-6 ✓). D1: N = {PP, ρI} by C2 (PP higher, IN-5) → bi ⌢ ’; H = {nii, sh, ł, hóósh} → bi ⌢ ’ ⌢ nii ⌢ sh ⌢ ł ⌢ hóósh ✓ (Hale's line Obj(obl)-INCH-M/S(sh)-v(¬)-V(ghoosh)).
  5. B Merge ʼAwééʼ in D2 and link it to ρX (φ 3), the slot inside the PP. Hale p. 6: the noun “is an adjunct, not an argument of the verb”. The noun links to the slot wherever INCH has put it. Nothing in the noun changes.

ID6. ʼiʼniishhą́ą́h ‘I start to snore’ / (11b) ʼIʼniishháásh ‘I start to sleep’: unergative base

  1. Σ Σ(√SLEEP) = {ρA (sentient, no endpoint: Dowty's “marginal” class, as on plain ł-causer §4), ρO (the inherent indefinite object: Hale 1997 “fundamentally transitive”; handout (11a) IV ’-)}. LINK-1 (E): sent ∧ ¬tel → ρA in Spec,v.
  2. A INCH: ρI in the object position; IN-2: ρO → [PP ρO P∅], spelled ’ (“the 3i object is ’- in both direct and oblique object forms”). ρA stays in Spec,v.
  3. A PORT: ρA → sh on M/S.
  4. A Spell-out: N = {PP ’, ρI ’}; H = {nii, Ø, sh, ł, stem} → ’ ⌢ ’ ⌢ nii ⌢ sh ⌢ ł ⌢ gháásh ✓ (handout (11b): IV ’- · V–VI ’nii- · VII Ø- · VIII sh- · IX ł- · X -gháásh). Hale gives no segment line for ʼiʼniishhą́ą́h; it is cited for the two ’ only.

ID7. (23a) Bináádaʼniilghal ‘We started to eat some more if [sic] it (meat)’: the oblique before the disjuncts

  1. Σ Σ(√EAT.MEAT) = {ρA = 1nsg (Spec,v), ρX = 3 (the meat; Comp,V)}. Adjuncts: náá- ‘more’ (Ic), da- distributive plural (III).
  2. A Merge V + ρX; da, then náá as V-level non-heads; INCH: ρI; IN-2: ρX → PP; IN-5: the PP is merged above náá (handout tree (23c) has [P b P] > náá > da). v = d-ł, ρA in Spec,v; M/S (´) perfective + iid; Qual [n ii].
  3. A PORT: ρA → M/S (iid, with the perfective tone).
  4. A Spell-out D1: N = {PP, náá, da, ρI} by C2 (higher first) → bi ⌢ náá ⌢ da ⌢ ’; H by C3 → nii ⌢ (´) ⌢ iid ⌢ d-ł ⌢ ghal. Output order = the handout's position line ✓.
  5. Result for T29f. Hale 2003 n. 13 called “oblique objects … in the highest prefixal position” “not a matter entirely devoid of mystery”. If the oblique PP is merged highest among the V-level non-heads, as Hale's own handout tree draws it, then C2 derives the order with no exception. Why it is merged there is still OPEN (T38f). So is the preverb competition: with a Ib preverb present, the pronoun shifts to IV (n. 13). Hale 2001 n. 4 (“sometimes … conjunct …, sometimes … disjunct”) fits a PP whose position depends on what else is merged.

ID8. The future shidíʼnóodlóół ‘I will start to get very cold’: what can and cannot be derived

  1. Hale prints no segmentation, so none is given here. Only his claim is used: the rounded long vowel [óo] is the trace of a third-person exponent in M/S, as in other future and progressive forms (p. 61; compare Hale 2001 (24a) shidoołhosh, w = “3rd:Perf”, SD1).
  2. A A4: M/S's φ = φ(ρI) = 3 → rounding ✓. A3: φ(ρEXP) = 3 ✓. A1: M/S has no φ; rounding needs a default (✗ as stated). A2: needs A3's or A4's source.
  3. Consequence. Every analysis that gets t2 must supply a third-person value with no participant behind it. A4 is the only one that uses an element the word already pronounces (’).

8. What changes on the site

  • Subject on the head: SH-8 gets a status note. U-2 is now stated as IN-4 (PORT-CLOSEST) and favoured, with SH-2 as its special case when Spec,v is filled. T37b row → partly answered.
  • Plain ł-causer: T4 gets a second, independent diagnostic. IN-6 sorts ‘sleep’, ‘snore’ (high) and ‘get cold’ (low) exactly as LINK-1 does.
  • A/B as spell-out: the T29f oblique half gets a printed Y&M 1987:187 form (via the handout) and a C2 derivation (IN-5), with its cause left open.
  • Postposition units: T32g's “unidentified Hale paper on the inchoative” is Hale, Munro & Platero 2000 (identified from Hale 2001's references and the PA paper p. 60; not on the box).
  • Open questions: new row T38a–h; the T37 and T4 rows get notes; T2 gets a datum (the inchoative oblique straddles B1, Hale 2001 n. 4).

9. What each source gets right, and what it misses

SourceGets rightMisses or leaves open
Hale handout (10)–(14), (23)The data, laid out position by position for all three verb types; the oblique of a transitive base as a P object placed above the disjuncts (23c), which derives the order of (23a); the subject slot of (12b) left empty, “—”.Treats the unaccusative's shi- differently from the transitive's bi- (raised subject agreeing with E vs P object), though both are spelled alike. The raising chain uses movement inside the word, and the future's third person in M/S is not accounted for.
Hale PA paper pp. 57–61One oblique analysis for all bases; the data point that M/S has a third person (rounding in the future); the argument against a lowering trace; a consistent in-situ position (p. 57).The expletive is posited, not shown; the indefinite ’ of ʼnii- is already a 3rd-person slot that could do the job (A4). The argument against eccentric agreement is theory-internal (“there is no ‘agreement’ … in a pronominal argument language”), so it is not contradicting data.
Hale & Keyser hale012Unaccusatives have an internal specifier; the ł-classifier transitive.Raising to subject is stated for English and Miskitu; nothing in it shows that Navajo's portmanteau needs raising.
Hale 1997The structural unaccusative/unergative split; unergatives' inherent object (visible as ’- in (11a)); Munro's “active” point, with the inchoative named as the place to test it.The inchoative study is only announced.
Willie 1991No movement to and from A positions: consistent with A3/A4.Silent on the inchoative in the chapters read.
This site before todaySH-8 posed the right question; LINK-1 predicted the three verbs' positions before this test was run.SH-1/SH-2 as written gave no M/S value for a low-only verb without a movement step. IN-4 now supplies one.

10. Grant's thesis, Dowty, event structure, STRATA, the changeover

  • GRANT'S CONJECTURE → HYPOTHESIS Weak thesis (role-bearing pronoun slots are the arguments; nouns are optional add-ons linked to them): supported. INCH re-licenses slots by their starting position, and the noun of (50) simply links to the re-licensed slot (ID5). No form here needs a noun in an argument position. The strong thesis (nouns are never arguments) is not tested here; it has already failed elsewhere.
  • HYPOTHESIS Feature bundles. On A4 the indefinite slot of ʼnii- is a bundle with no role at all (E empty, φ = 3, δ = INDF). It is pronounced in two ways depending on the company it keeps: as an ordinary object (ID5, ID6), or with its φ in M/S and its δ before the qualifier (ID2), like the indefinite subject of (22b). That is a second case, after the fourth person, of one bundle with two pronunciation sites.
  • HYPOTHESIS Dowty (baseline). The test sorts by role in the action: the one who undergoes a change of state (‘get cold’, Proto-Patient) starts low; sentient participants of predicates with no endpoint (‘sleep’, ‘snore’) start high. That is Dowty's §12 picture of unaccusativity, with the site's sentient + no-endpoint refinement for his “marginal” class (LINK-1). This is an extension of Dowty, not a departure. Three verbs is a small sample (T38d).
  • HYPOTHESIS Theta roles and event structure, compatible. INCH is an event-structure operation: it adds an onset (‘start to’). It does not change anyone's role in the action: the cold one still undergoes the change, the tickler still acts. It changes only where a slot is licensed and pronounced. The two descriptions apply to the same participants at different points.
  • GRANT'S CONJECTURE → HYPOTHESIS STRATA. The derivations need this order: layer 1, slots and their starting positions (Σ, LINK-1); layer 2, the event-structure operation INCH, which reads those positions (Spec,v vs lower) and re-licenses the lower slots; then spell-out of the word (PORT, DELTA-HOST); and only then layer 3, nouns (ʼAwééʼ), which link to whatever slot results. INCH cannot apply before layer 1 (it needs the positions), and nothing in it looks at nouns. This is consistent with STRATA. It does not separate STRATA from the adjunct view or the inside-the-clause view of where the nouns go, because Navajo nouns carry no case that INCH could change.
  • HYPOTHESIS The changeover (rule set A vs B). INCH makes a small phrase inside the word: a postpositional phrase, with P after its object (the “lexical heads take their argument on the left” pattern), spelled out as its own domain and then placed as one non-head in the word. Hale 2001 n. 4 says these inchoative objects are sometimes conjunct and sometimes disjunct, and (23a) shows a disjunct one. So the oblique straddles the disjunct/conjunct line B1, as the reflexive does (reflexive vs reciprocal). That is one more datum that B1 is not the cycle boundary, consistent with D1 = the MVS (T2).
  • HYPOTHESIS Phrase-level packaging. ʼnii- is listed as a unit made of a slot exponent plus two heads (Hale's “complex prefix”; Young & Morgan's “inchoative”). It is lexicalized above the single morpheme: a stored piece of structure, like an English ‘start V-ing’ frame, not a stored word.

11. Tasks T38a–h

#TaskNeeds
T38aAnswered 2026-10-19 (12:46pm): the participant's prefix is first, HMP (10a) (HD2). OPEN A1 vs A3/A4: an unaccusative ‘start to’ form with a disjunct prefix (náá-, da-, iterative, a preverb). Is the participant's prefix first, like (23a)'s bi- (oblique), or next to ʼnii- (E agreement)? (ID4)Young & Morgan 1987:187–188, 217 (not opened); Hale, Munro & Platero 2000
T38bAnswered as far as the analyses go, 2026-10-19 (12:46pm): IV, thematic (HMP). OPEN A3 vs A4: in an unaccusative ‘start to’ form, is the ’ of ʼnii- a subject (position V, as the handout's “V–VI” label has it) or an object (IV, as the PA paper's “Obj(’)” has it)? Look for a form in which something can stand between IV and V.Young & Morgan paradigms
T38cOPEN Is M/S's third person constant for every person of the oblique participant (2sg, 1pl, 4th) in the future and progressive? A3/A4 say yes; A1 says M/S has no person at all.Young & Morgan paradigms
T38dPartly answered 2026-10-19 (12:46pm): 5/5 shared verbs agree. OPEN Run the ‘start to’ test over the H&P/LINK-1 sample. Do all plain-ł-alternating verbs lose their subject under ʼnii-, and all non-alternating ones keep it? Two diagnostics, one starting position (joins T4).Young & Morgan; Young, Morgan & Midgette 1992
T38ePartly answered 2026-10-19 (12:46pm): “thematic” (HMP); role-less placeholder for ‘cry’ and weather verbs (IN-9). OPEN What is the indefinite ’ of ʼnii-: a role-less slot (like pleonastic ho-, T17), or an onset/event argument? Can a noun ever link to it?Hale, Munro & Platero 2000; texts
T38fPartly answered 2026-10-19 (12:46pm): the object position is full (HMP n. 2); blocking by so-. OPEN Why is the oblique PP merged above the disjunct adjuncts (IN-5)? Compare the causative oblique (Hale 1997 n. 27 records two orders of the causee prefix and iterative ná- for ‘make him drink something (iterative)’; the forms are in a damaged font and are not reproduced) with the inchoative oblique, and the preverb competition (Hale 2003 n. 13). Joins T29f.Young & Morgan 1987D:215; Rice 2000
T38gPartly answered 2026-10-19 (12:46pm): the HMP manuscript read (text only); the 2000 printed version not opened. OPEN Read Hale, Munro & Platero 2000 (“The Navajo inchoative verb form”, MITWPL 3, pp. 97–116): their eccentric agreement in their own words, and Hale 2001 n. 4's “various factors”.A copy (not on the box)
T38hOPEN A fourth-person subject with ʼnii-. SH-4 + C1 predict OBL ⌢ ’ ⌢ j ⌢ n ⌢ ii … (the inchoative ’ as an object before the deictic on the qualifier), not OBL ⌢ j ⌢ ’ ⌢ n ⌢ ii …. Schematic only; find a printed form.Young & Morgan

Sources: Hale PA paper pp. 6, 57–61 ([local working copy, not in the repository], page images; crops under [local working copy, not in the repository], not committed). Hale handout pp. 1, 6, 7, 9 (hale062.pdf, page images). Hale 2001 p. 683 n. 4 (page image). Hale & Keyser hale012 PDF p. 21 and Hale 1997 hale023 PDF p. 60 (text layers of born-digital PDFs). As recorded on the site: Willie 1991 p. 23; H&P 1996 (15); Dowty 1991 §12; Hale 2003 nn. 11, 13. Cross-links: subject on the head (SH-8), plain ł-causer (T4, LINK-1), A/B as spell-out (T29f), postposition units (T32g), reflexive vs reciprocal, open questions (T38), sources.

‘Start to’ in Hale, Munro & Platero's own words: the participant's prefix goes first, the ’ is an object, and verbs with nobody low get a placeholder (T38a, T38b, T38d–g)

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. Kept as written (URL, data and status labels unchanged). Filed under Merge and Agree inside the word; the consolidated argument is the essay merge-agree. Rule IDs: rules index.

Verbs on this page → dictionary: (ł-)gan ‘dry up’ · hoołtį́į́ł ‘rain (progressing)’

Restated on the unified structure (2026-11-15). Voice, causatives and events are restated with every theta role assigned inside vP: the classifier spells a value of v⁰ (H59, distribution NOT UNDERSTOOD), a causer is an external argument in Spec,vP, and the passive keeps a silent indefinite agent (H10b). This page’s causative analysis has not yet been restated on the unified structure (OPEN). Light d, used below, is withdrawn (Grant, 2026-11-14); light n is an OPEN structural question. 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. Position-class template statements below describe the order of prefixes; they are not the derivation (§3). Flat-S and other pre-VPISH notation below is labelled contrast. Status on the spine: see unified structure, §4, causatives 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.

2026-11-11 audit pass 3. The A/B headedness changeover is a hypothesis and is not established; sentences that stated it as fact now say so. See the pass 3 report.

Local box file paths were removed from this page; files inside the repository are given by repo-relative path, and the source citation next to each removed path stays as the reference.

2026-11-11 audit pass 6. The derivation step that called ‘get cold’ telic now says this is our reading of the English gloss, and marks telicity OPEN. See the pass 6 report.

The last run (the ‘start to’ test) used two Hale texts on the inchoative ‘start to’ prefix ʼnii- and ended with eight open tasks (T38a–h). Most of them needed Young & Morgan or the paper Hale cites for “eccentric agreement”, Hale, Munro & Platero 2000, and neither was on the box. This run found an open copy of a Hale, Munro & Platero manuscript on the inchoative (“Aspects of Navajo Verb Morphology and Syntax: The Inchoative”, a Navajo Language Academy workshop paper, hale028). It is not the 2000 printed paper. It is a draft that stops at “(To be continued …)”. It has sixteen numbered examples, all from Young & Morgan 1980, and covers all three verb types plus weather verbs. On those data it answers T38a and T38b, and it partly answers T38d, T38e, T38f and T38g.

Ground rules (Grant's). The HMP manuscript, Hale's handout, his PA paper and this site's rules are each treated as possibly partly right. A template position or a gloss label is an analysis, not data. Only contradicting data count against a hypothesis; silence counts against nothing. Grant does not speak Navajo. Every Navajo form below is copied from the HMP manuscript or from a page already on the site, with its example number. Forms are segmented only where the source prints a segmentation. Where we align an unsegmented form with a segmented one, the alignment is marked as ours. No form, gloss or judgment is invented. Dowty's proto-roles are the baseline, and “role in the action” is used for his entailments. Theta roles and event-structure roles are treated as compatible descriptions of the same participants. HMP's trees put the subject in a VP-adjoined position (VPISH, after Koopman & Sportiche 1991). They are transitional trees, not flat-S trees. They are restated and tested below, and are not used as evidence about where nouns attach.

Notation. This is the site's own convention. The Syntax · Semantics · Interface bot's BRIEF.md and VERB-QUEUE.md have not changed since Oct 18, 10:05pm PT, and they still contain no notation conventions for this site. Σ = role-schema step. A = word level (cycle D1 = the MVS). B = clause level. The spine, ρ-slots ⟨Eρ, posρ, φρ, (δρ)⟩, CYC (C1 non-heads before heads; C2 non-heads higher-first; C3 heads higher-first), PORT and IN-0 to IN-7 are as on the ‘start to’ test §5. ρI = the slot spelled by the ’ of ʼnii-. ρC = the slot that satisfies INCH's object requirement (new, IN-8). HMP's categories: E = “position VI and associated object, deictic elements prefixed to that position”; M = mode; AGRs/AGRo = subject/object agreement (their n. 1).

Bottom line, in plain words

SOURCED The facts. (1) In ‘I start to get cold about dawn’, shináʼniidlóóh (HMP (10a), Y&M80D:217), the ‘I’ prefix shi- comes first, before the repetitive ná-. Ordinary objects do the same: bi-ná-ʼnii-sh-… ‘start to drink it’ (repetitive mode, (2b)). HMP call this “bumping”: the ’ of ʼnii- already sits in the object position (IV), so the object prefix is pushed out to the far left, “into the disjunct prefix Position I” (their n. 2). (2) That ’ is “the Position IV indefinite object prefix … functioning here as a thematic element”, citing Young & Morgan 1980G:320–321. (3) The inchoative “must always be preceded by overt object agreement morphology”. In this it “behaves like an ‘incorporated’ postposition”. (4) A verb with no participant low enough to fill that need gets an indefinite ’ instead. Examples are ‘the baby starts to cry’ ʼináʼniichah (7b), ‘it started to snow’ ʼiʼniichííl (16a) and ‘winter started’ ʼiʼniihai (16c). With ‘spoil’, by contrast, the participant itself appears in that spot: ‘this meat has started to spoil’ biʼniichxʼǫʼ (10d; font restored, see §1). (5) Five unaccusative verbs show the subject-to-object shift: get cold, get dry/skinny, get fat, get old, die/get sick ((11)).

HYPOTHESIS What this shows. T38a is answered. An unaccusative participant under ‘start to’ goes where every other object under ‘start to’ goes, the far left. That contradicts the handout's eccentric-agreement version as this site had stated it (ID4 predicted the participant would stay next to ʼnii-). HMP's own version survives because it adds the bumping rule. The slot analyses need the same rule, now with a reason: the object position is taken. T38b is answered as far as the sources go: the ’ is an object-position element in all three HMP descriptions that state a position. The handout's “V–VI” label is the E domain, which by HMP's definition includes the object position. So this site's 4:46am idea that this ’ is the hidden third-person subject loses its two supports. The cheaper account is the one HMP state: in the unaccusative ‘start to’ form the subject position is simply empty, “as it is regularly” for every non-local person. The main new point is this: ‘start to’ needs a filled object slot, not a particular word or a particular participant. When the verb supplies none, a role-less placeholder fills it. That is Grant's feature-bundle picture in a single construction.

OPEN Still open: whether the third person in the future is the same for every person of the participant (T38c); a fourth-person subject with ʼnii- (T38h); the 2000 printed version; and page images of this manuscript, whose Navajo font did not survive text extraction (T39a).

On this page: 1 Sources · 2 The questions · 3 Data · 4 Claims side by side · 5 Formal rules · 6 Five analyses tested · 7 Derivations HD1–HD9 · 8 What changes on the site · 9 Gets right, misses · 10 Grant's thesis, Dowty, event structure, STRATA, the changeover · 11 Tasks

1. Sources opened this run (2026-10-19, 12:46pm PT)

SourceWhat was read, and how
Hale, Ken, Pamela Munro & Paul Platero. “Aspects of Navajo Verb Morphology and Syntax: The Inchoative.” MIT / UCLA / Window Rock, Navajo Language Academy Linguistics Workshop, ms., undated (cites Hale 1997 workshop and Smith 1996; ends “(To be continued. …)”). Hale's paper list, hale028.pdf; found through its CiteSeerX record (10.1.1.695.3081). We call it HMP ms.Newly opened: the full text (pp. unnumbered in the extraction), examples (1)–(16), notes 1–6, references. Access: the box could not reach the MIT host (curl timed out over http; https, the Wayback Machine and CiteSeerX failed TLS or timed out). The text was read through the web-fetch tool's text extraction, without page images. Font: the Navajo font comes through damaged in two consistent ways. A nasal hook shows as “²” before its vowel (“-dl²á ‘drink’”, which is -dlą́, the stem H&P print for ‘make OBJ drink something’ on plain ł-causer §2.3). ł shows as “¬”, as in Hale & Keyser's ms. (‘start to’ test §3.1). Forms that contain neither mark are quoted exactly. Forms that contain them are given as extracted, with our restoration in square brackets, and no argument rests on a restored character. The trees' layout cannot be seen, so only HMP's prose about the trees is used.
Used as recorded on the siteThe ‘start to’ test (Hale PA paper (50), (51), pp. 57–61; handout (10)–(14), (23)); Hale 2003 n. 13 (A/B as spell-out f2); H&P 1996 lists (11), (12), (15) and LINK-1 (plain ł-causer); Dowty 1991 §12 and (73); INV-0 to INV-5 (inverse); SH-2 to SH-4 (subject on the head); the T32g search snippet (postposition units).
Not openedHale, Munro & Platero 2000, “The Navajo inchoative verb form” (MITWPL 3, pp. 97–116): the printed version may differ from the ms. Young & Morgan 1980, 1987: every HMP example cites them, but nothing is attributed to them beyond what HMP report. Smith 1996, Speas 1988 (as HMP cite it), Hale & Platero 1996.

2. The questions

From ‘start to’ test §11. T38a: in an unaccusative ‘start to’ form with a disjunct prefix, is the participant's prefix first (like the oblique bi- of handout (23a)) or next to ʼnii- (handout tree (14))? T38b: is the ’ of ʼnii- a subject (V) or an object (IV)? T38c: is Mode's third person constant? T38d: does the ‘start to’ test sort verbs as the plain ł-causer test does? T38e: is the inchoative ’ role-less? T38f: why is the oblique leftmost? T38g: HMP in their own words. T38h: a fourth-person subject with ʼnii-.

3. Data and HMP's statements SOURCED

3.1 What ʼnii- is, and what it requires

  • Shape: “The constant element in the morphology of the inchoative is a prefix complex of the form -’nii- … it consists of three parts—namely, (i) the Position IV indefinite object prefix ’- (glossed 3i in Young and Morgan), functioning here as a thematic element; (ii) the Position VIc prefix n-, glossed ‘inceptive’ by Young and Morgan; and (iii) the Position VId prefix -ii-, glossed ‘punctual’ (Young and Morgan, 1980G:320-321). This composite prefix, like Position VI aspectual prefixes generally, belongs to the functional category E”.
  • Stem: the inchoative depends on “the verb stem, which must be drawn from the momentaneous aspect set”.
  • Requirement: “The inchoative must always be preceded by overt object agreement morphology.” The forms in (2) “show overt object agreement (third person there), in contrast to the simple non-inchoatives (2a′–e′) which follow the usual pattern according to which third person object agreement in Position IV is non-overt (null) when the subject is first or second person. In this respect, the inchoative prefix complex behaves like an ‘incorporated’ postposition—by definition, a postposition requires overt agreement.”
  • n. 3: “a postposition necessarily combines with morphology representing its complement (object), be this agreement or the nominal head incorporated. See Speas (1988) for a very suggestive proposal … analyzing the third person objective morphology b- as an incorporated pronoun.” This is the passage the T32g search snippet quoted (postposition units).
  • n. 2, “bumping”: “Normally, objective morphology is in Position IV. However, the first element in the inchoative complex, i.e., the indefinite (3i), occupies that position. The conflict is resolved in part by ‘bumping’ the objective prefix (corresponding to an argument of the verb) leftward into the disjunct prefix Position I (see (2b) above). Certain Position I prefixes block this, in which case the objective prefixes appear in their standard position.”
  • n. 1, E: “nucleus defining the forward edge of the conjunct prefix system, i.e., position VI and associated object, deictic elements prefixed to that position”.

3.2 (2) ‘drink’ across the modes, with HMP's gloss lines

HMP print segment and gloss lines. They give no translation per form: the forms are “inchoative forms of the verb -dl²á [-dlą́] ‘drink’”, with the non-inchoative “to the right for comparison”. Only the stems are damaged.

Ex.Inchoative (as extracted) [stem restored]HMP glossPlain counterpart [stem restored]HMP gloss
(2a) [I]bi-ʼnii-sh-dl²í²íh [dlį́į́h]3-INCH-Ø.1SG-dCL.drink.MIyish-dl²á [dlą́] [CI]Ø.1SG-dCL.drink.CI
(2b) [R]bi-ná-ʼnii-sh-dl²í²íh [dlį́į́h]3-R-INCH-Ø.1SG-dCL.drink.MUná-sh-dl²í²íh [dlį́į́h]R-Ø.1SG-dCL.drink.MU
(2c) [P]bi-ʼnii-sh-dl²á²áʼ [dlą́ą́ʼ]3-INCH.PERF-1SG-dCL.drink.MPyi-sh-dl²á²áʼ [dlą́ą́ʼ]PERF-1SG-dCL.drink.MP
(2d) [F]bi-dí-ʼnée-sh-dl²í²í¬ [dlį́į́ł]3-F-INCH-PROG.1SG-dCL.drink.MFdee-sh-dl²í²í¬ [dlį́į́ł]F.PROG-1SG-dCL.drink.MF
(2e) [O]bi-ʼnoo-sh-dl²²á²áʼ [dlą́ą́ʼ]3-INCH.OPT-1SG-dCL.drink.MOwó-sh-dl²á²áʼ [dlą́ą́ʼ]OPT-1SG-dCL.drink.MO

Our Leipzig relabelling (categories only): 3 → 3.OBJ (HMP call it “object agreement”); INCH → INCH; Ø.1SG / 1SG → M-1SG.SBJ; dCL → CL; R → REP. So (2b) = 3.OBJ-REP-INCH-M-1SG.SBJ-CL-drink.USIT. The doubled “²²” in (2e) is a font artefact, copied as extracted.

3.3 Transitive bases, (1)

Ex.Verb word (as extracted)Translation (HMP's)Note
(1a) (Y&M80D:218)yiʼniidiz‘This morning my grandmother started to spin the black wool.’3rd-person subject and object; the word begins yi-. The rest of the sentence has damaged characters and is not reproduced.
(1b) (Y&M80D:218)biʼniisol‘I started to blow on the fire but my mother told me not to … .’1sg subject; bi-. The noun is “K²o’” [kǫʼ] in the extraction.
(1c)/(3a) (Y&M80D:219)biʼniishdl²²á²áʼ [biʼniishdlą́ą́ʼ]; (3a) Shí gohwééh biʼniishdl²á²áʼ‘It wasnt [sic] until last year that I started to drink coffee.’; (3a) ‘I started to drink coffee.’HMP derive (3a) in (3)–(4) (§3.7)

None of these is segmented by HMP. That (1a)'s yi- is the 3rd-person object prefix used under a 3rd-person subject is our reading (HYPOTHESIS), set against (1b)'s bi-.

3.4 Unergative bases, (5)–(8), n. 4

“Many so-called ‘unergative verbs’ in Navajo are explicitely [sic] transitive, supporting an argument structure in which the inner argument is represented by the indefinite object … ’-¬-h²á²áʼ [ʼ-ł-hą́ą́ʼ] ‘snore’, ʼ-l-zhish ‘dance’, ’-¬-hosh [ʼ-ł-hosh] ‘sleep’.” The indefinite ’- “nonetheless qualifies as the object agreement required by the inchoative, just as it would in the case of an ordinary transitive verb”.

Ex.Verb wordTranslation (HMP's)
(6a) (Y&M80D:470)ʼIʼnii¬h²á²áʼ [ʼIʼniiłhą́ą́ʼ]‘My wife awakened me when I started to snore.’
(6b) (Y&M80D:471)ʼidíʼníílzhish‘When I say ‘now’ … you'll start to dance.’
(6c) (Y&M80G:320)ʼIʼniishháásh.‘I am in the act of starting to sleep.’ (= handout (11b), there ‘I start to sleep.’)
(7a) (Y&M80D:779)yícha (in … baa yícha)‘I cried over my kitten when it was run over by a car.’
(7b) (Y&M80D:470)ʼawééʼ ʼináʼniichah‘When it gets dark the baby starts to cry.’
n. 4sodiʼniiszin (theme so=d-l-zin ‘pray, say a prayer’)‘I started to pray’

HMP on ‘cry’: it is “not explicitely transitive … under ordinary conditions, no overt indefinite object morphology appears. Interestingly, however, in the inchoative of these verbs, the indefinite object morphology ’- appears”. Following Hale & Platero 1996, ‘cry’ is denominal, a bare noun cha ‘weeping, crying’ conflated with a null verb, structure (8) [V N cha V]. “Unlike the unaccusatives …, unergative verbs do not project a specifier, and hence do not have an argument in that position. Given that their complement is conflated with the verb, there is no argument around which can enter into the agreement relation in the normal way. Use of indefinite agreement is a reasonable alternative.” But “The explanation is not straightforward, however, and we will postpone our implementation of it to another occasion.” n. 4: the Position I prefix so- “prevents the object prefix from moving from position IV into Position I, as would normally happen under ‘pressure’ from the thematic ’ inherent to the inchoative (cf. (7b)). As a result, the objective ’- remains in Position IV, giving rise, hypothetically at least, to the sequence ’-’- which reduces to a single glottal stop”.

3.5 Unaccusative bases, (9)–(13), n. 5

Ex.Verb wordTranslation (HMP's)
(10a) (Y&M80D:217)shináʼniidlóóh‘I start to get cold about dawn, so I rekindled the fire.’
(10b) (Y&M80D:217)shiʼniigan‘When I was taken prisoner in the war I started to become skin and bones in two months.’
(10c) (Y&M80D:220)shílaʼ biʼniitin‘While I was fixing a tire my hands started to get cold (lit. freeze) … .’
(10d) (Y&M80D:217)Díí ʼats²iʼ biʼniichx²oʼ. [ʼatsįʼ biʼniichxʼǫʼ]‘This meat has started to spoil.’
(10e) (Y&M80D:217)biʼniilkʼaii‘I fed my horse hay every day and he started to get fat.’

(11), “Compare the following first person forms”:

Non-inchoativeInchoativeGloss (HMP's)
yishdlóóhshiʼniidlóóh‘become cold’
yishganshiʼniigan‘become dry, skinny’
nishkʼahshiʼniilkʼááh‘become fat’
haashtihshiʼniitih‘become old’
daastsaahshiʼniitsaah‘die, get sick’

HMP: “the single argument of these unaccusative verbs is represented by object agreement, not subject agreement. Consequently, Position VIII, the standard locus of subject agreement, is entirely empty in the verbs of (10)”. “In the inchoative, the subject is construed ‘eccentrically’ with the morphology normally construed with an object. We maintain, however, that there is no diathesis change here—the subject does not ‘become’ the object.” On (13): “The inchoative, so to speak, demands object agreement, and this requirement is evidently paramount, forcing the subject to be construed with object agreement in E, AGRo, overriding the ‘normal’ construal with subject agreement in M. Subject agreement can be overridden, presumably, because that position can be empty, as it is regularly in persons other than first or second (i.e., in the third, fourth, spatial, and indefinite persons).” On locality: E “is not local in relation to the subject; there is an intervening head, M. Whether M projects a barrier to agreement in this situation is an empirical issue. If it does not, then … the simplest scenario is sufficient—the subject checks object agreement in situ. If M projects a barrier …, the subject must raise and adjoin to MP”. n. 5: unaccusative subjects “are assumed to be raised from a specifier position with in [sic] the VP”. Their unaccusative class is Hale & Platero 1996's: verbs that take part in the simple transitivity alternation, “-gan/-¬-gan [-ł-gan] ‘dry (intransitive/transitive)’, -chx²o²oh/-¬-chx²o²oh [-chxʼǫǫh/-ł-chxʼǫǫh] ‘ruin, spoil’”.

3.6 Areal and weather verbs, (14)–(16), n. 6

Ex.Verb word (as extracted) [restored]Translation (HMP's)
(14a) (Y&M80D:463)hweeshdla¬ [hweeshdlał]‘I was in the process of cutting a furrow to my field [for water] … .’
(14b) (Y&M80D:461)hoo¬t²í²í¬ [hoołtį́į́ł]‘Here comes the rain.’
(15a) (Y&M80D:72)ʼahoʼnii¬dláád [ʼahoʼniiłdláád]‘This morning, I started plowing in the field.’
(15b) (Y&M80D:72)ʼanáhoʼnii¬t²í²íh [ʼanáhoʼniiłtį́į́h]‘It starts to rain every afternoon.’
(16a) (Y&M80D:469)ʼiʼniichííl‘Last night at ten o'clock it started to snow.’
(16b) (Y&M80D:470)ʼiʼniish²í [ʼiʼniishį́]‘Summer has begun with us.’
(16c) (Y&M80D:469)ʼiʼniihai‘Winter started before I got all my corn put away.’

HMP on (15): “We are not sure of the proper analysis of these forms … Is it the areal morphology ho- (< h w-)? Or is it the initial element ʼa- (< ’-)? Young and Morgan assume that this is the Position Ia adverbial prefix ’- ‘away’. But we suspect that it is the indefinite object prefix”. “The areal agreement morphology, evidently, does not itself accomplish this.” n. 6 gives Young & Morgan's alternative and adds that the adverbial's appearance “would, however, be a mystery”. ‘Snow’ “lacks the areal agreement morphology. But like the latter, and like unergatives generally, it employs the indefinite object morphology in the inchoative. The verbs of season likewise use the indefinite object”. Their to-do list includes “weather, year, and seasonal unaccusatives”.

3.7 HMP's derivation of (3a), in their words

“The subject DP, shí, being adjoined to VP, is in the position required to ‘check’ subject agreement in M. The object DP, gohwééh, on the other hand, must raise and adjoin to MP, from which position it can check object agreement in E. The Extended Projection Principle is satisfied by raising the subject and adjoining it to EP. The verb word itself is assembled by means of Head-Movement, adjoining V to the right of M and the latter to the right of E. This derivation is standard”. Tree (3b) has EP over MP over VP, with E = AGRo b + ʼnii and M = Ø + AGRs sh (labels only; the layout cannot be seen). For the unaccusative, (12b)/(13b) keep the subject DP adjoined to VP, with AGRo shi on E and “(AGRs)” on M.

4. What the sources claim, side by side

QuestionHMP ms.Handout (Hale)PA paper (Hale)This site before today
Where is the ’ of ʼnii-?Position IV, object, “thematic”“V–VI” in (10b)–(12b); “VI” in (23b)“Obj(’)”, “the 3i indefinite object prefix”ρI in the object position (IN-0, IN-1), but read as the subject in the unaccusative (IN-7 (I))
What does INCH require?Overt object agreement before it, like a postpositionOblique P for the object (23)“Since the object position is filled”, an oblique (p. 59)IN-2 (lower slots → PP)
Unaccusative participantStill the subject, “construed eccentrically” with AGRo in E; no diathesis change; checks in situ or from MPRaised to Spec,EP (tree (14))An oblique object; no trace, no eccentric agreementOblique (A3/A4), or raised (A1)
Subject slot of the unaccusative inchoativeVIII “entirely empty”, as for all non-local persons“—”A third-person element, “presumably expletive” (p. 61)ρI's φ (A4) or an expletive (A3)
Why the object is leftmost“Bumping” from IV to Position I; some Position I prefixes block it[P b P] adjoined above the disjuncts (23c)n. 13: “when not in competition with a preverb”IN-5 OBL-HIGH, cause OPEN
Verbs with nothing lowIndefinite ’ inserted (cry, snow, seasons, rain); mechanism postponed——— (not covered)
NounsDPs that check agreement (subject VP-adjoined, object raised to MP)DPs in specifiersAdjuncts; the pronouns are the argumentsLinked to slots in rule set B

5. Formal rules

IN-8  PRECEDE-OBJ  INCH (= rho_I's ' + Qual n + Qual ii, inside E) has a P-like requirement:
                   it needs a complement slot rho_C with an overt exponent, pronounced to
                   its left.  P_null of IN-2 is this requirement.
                   (SOURCED: HMP "must always be preceded by overt object agreement
                   morphology"; "behaves like an 'incorporated' postposition"; n. 3.
                   The slot statement is ours.)
IN-9  FILL         rho_C = the slot first-merged below v (Comp,V or Spec,V), if there is one;
                   otherwise a fresh rho_D = <E = none, phi = 3, delta = INDF>, spelled '.
                   A slot in Spec,v is never rho_C.  Areal ho- is never rho_C.
                   (SOURCED pattern: (2), (6), (10), (11) vs (7b), (16), (15) and n. 4;
                   HMP: "use of indefinite agreement is a reasonable alternative".
                   rho_D as a role-less slot is ours.)
IN-5' BUMP         (replaces IN-5)  rho_C cannot be pronounced in the object position,
                   which rho_I fills (IN-1).  Its PP is merged as the highest V-level
                   non-head, so C2 spells it at the left edge, before every disjunct
                   (Position I), UNLESS a thematic Position I prefix (so-) or a Ib preverb
                   holds the left edge.  Then rho_C stays in the object position and fuses
                   with rho_I (' + ' -> ').
                   (SOURCED: HMP n. 2, n. 4, citing Y&M 1980G:320-321; Hale 2003 n. 13
                   "when not in competition with a preverb".  The CYC restatement is ours.)
IN-7  (D) DEFAULT  (new fourth option)  If PORT finds no eligible slot (Spec,v empty; every
                   lower slot inside a PP; rho_I a thematic object), M/S is spelled in its
                   non-local (third-person) form.  No element is added.
                   (SOURCED reason: HMP "that position can be empty, as it is regularly in
                   persons other than first or second".  The future vowel [óo] is that form
                   (Hale PA p. 61's datum).  The rule statement is ours.)
IN-10 HM-RIGHT     The verb word is built by head movement that adjoins V to the RIGHT of M
                   and M to the RIGHT of E, inside head-final phrases (VP, MP, EP).  This
                   gives head order E > M > V = C3 (heads higher-first).  Object-series
                   exponents precede E (C1, non-heads before heads).
                   (SOURCED: HMP's prose on (3)-(4).  The link to C1/C3 is ours.)
IN-6' DIAG         (extends IN-6)  Under INCH a verb shows one of three patterns:
                   (a) its sole slot is first-merged in Spec,V -> that slot = rho_C (object
                       prefix before INCH; M/S default);
                   (b) it has a slot in Comp,V (transitive, or an inherent ') -> that slot =
                       rho_C; the Spec,v slot stays the M/S subject;
                   (c) it has no slot below v (conflated noun; weather, season) -> rho_D.
                   (HYPOTHESIS; a theorem of IN-8, IN-9 and PORT.)
RuleRests onStatus
IN-8HMP §2 and n. 3; (2a–e) vs (2a′–e′)SOURCED requirement; the slot form is HYPOTHESIS
IN-9(1)–(2), (6), (10)–(11) vs (7b), (15), (16), n. 4SOURCED pattern; ρD is HYPOTHESIS. It extends IN-2, which said nothing about verbs with no lower slot.
IN-5′HMP n. 2, n. 4; (2b), (10a), (7b), (15b); Hale 2003 n. 13; handout (23a, c)SOURCED generalization (two Hale texts agree); the CYC statement is HYPOTHESIS. It gives IN-5 a cause (the object position is full); why the escape site is the far left remains OPEN (T39c).
IN-7 (D)HMP on (13); Hale PA p. 61 (the vowel)HYPOTHESIS, with a SOURCED reason; now favoured over (I) and (E) (§6)
IN-10HMP prose on (3)–(4)SOURCED mechanism; the C1/C3 link is HYPOTHESIS
IN-6′derived; 13 one-participant verbs plus four weather and season verbs (§6, u10)HYPOTHESIS; no counterexample in the printed set

Update 2026-10-20, 4:46am (T41e). Chomsky 2000/2001 let an expletive fill only a T-like, non-phase head's extra position (MI (5a), (6), (45b)). This supports IN-8's location of the requirement on PINCH over the qualifier-layer location used on the originals page. Read as an obligatory EPP-feature of PINCH (EP-2, EP-4), it lets IN-9's two stipulations follow (EP-3). The placeholder appears only when nothing is below v: “Merge preempts Move” (MI p. 106), and a low participant would otherwise have no position, since ρI holds the object position. And the Spec,v participant never fills the slot, because it is merged above PINCH. IN-7 (D) gains MI's defective-intervention default ([3person], p. 128) as one of two mechanisms (EP-5). The rules above are kept as written. Derivations re-run: EP1–EP5.

6. Five analyses tested against the data

Analyses. A1s = the handout's eccentric agreement as this site stated it in ID4: the participant raises to Spec,EP and is pronounced on E next to ʼnii-. A1h = HMP's own version: AGRo in E is checked by whichever DP is available (in situ, or adjoined to MP), the object prefix is bumped to Position I, and VIII is empty. A3 = the PA paper: an oblique, plus an expletive in Spec,v. A4 = this site at 4:46am: IN-4 + IN-7 (I), with ρI's φ on M/S. A5 = this page: IN-8, IN-9, IN-5′, IN-4 + IN-7 (D).

Tests. u1 (10a) shináʼniidlóóh: the participant's prefix precedes the disjunct ná- (T38a). u2 one statement covers the first prefix of all three base types before ná-: (2b) bi-ná-ʼnii-, (10a) shináʼnii-, (7b) ʼináʼnii-, (15b) ʼanáhoʼnii-. u3 the ’ of ʼnii- is described as a IV object in every source that states a position (HMP, PA paper). The handout's “V–VI” is E, which by HMP n. 1 includes the object position. This is a label test, not a data test. u4 verbs with nothing below v get an inserted ’ ((7b), (16)), and areal ho- does not satisfy INCH ((15)). u5 a Position I so- blocks the leftward shift (n. 4), as does a Ib preverb (Hale 2003 n. 13). u6 in the future, the vowel after ʼn- tracks the subject: (2d) ʼnée with 1sg sh, (6b) ʼníí with ‘you’, and Hale's shidíʼnóodlóół ʼnóo with the unaccusative. This alignment of unsegmented forms is ours. u7 yi- under a 3rd-person agent (1a), bi- for the unaccusative's 3rd-person participant ((10c–e)) and under a 1sg subject ((1b)). u8 (framework, not data) no A-movement inside the word (Willie 1991:23; PA p. 57). u9 (economy) no element that nothing pronounces. u10 T38d: the two diagnostics agree.

Analysisu1u2u3u4u5u6u7u8u9
A1s handout, as on ID4✗ contradicted: predicts X ⌢ shi ⌢ ʼnii✗ two analyses for one spelled prefix✓— silent— silent✓ with a default✓✗ three movements✓
A1h HMP ms.✓ by bumping✓ every prefix is “objective” AGRo, every one bumped✓ (their own)✓ as description; mechanism “postponed”✓ (their n. 2, n. 4)✓ (VIII “empty”)✓✓ in the in-situ scenario (nouns move as DPs in the clause: framework)✓
A3 PA paper✓ with IN-5′✓✓✓ only with IN-9 (the PA paper is silent)✓ with IN-5′✓✓✓✗ expletive
A4 site 4:46am✓ with IN-5′✓~ strained: a thematic IV object would be the grammatical subject; both of its supports are gone (the “V–VI” label is E; the (22b) indefinite ’ is a position-V subject, this one is IV)✓ with IN-9✓✓✓✓✓
A5 this page✓✓✓✓✓✓✓✓✓

u7 for every analysis. INV-0 defines yi-/bi- voice only when Σ has ρA and some ρX, all third person. In (1a) both are third person and the grandmother outranks the wool, so INV-3 gives yi-. In (10c–e) Σ has no ρA, so voice is undefined and the overt oblique bi- appears, as with the 1sg subject in (1b) (PA p. 59: a third-person oblique is overt). An expletive (A3) or ρI (A4) is not ρA, so they make the same prediction. INCH changes where the object is licensed, but it does not take the object out of the voice competition. That fits INV-1 (roles fixed) and HMP's “no diathesis change” for roles.

u10, T38d. Verbs printed by both HMP and the causative sources: (ł-)gan ‘dry up’ (H&P (11), alternating) → shiʼniigan, shifts ✓. -chxʼǫǫh ‘spoil’ (alternating, per HMP citing H&P 1996) → biʼniichxʼǫʼ, shifts ✓. -cha ‘cry’ (H&P (12), non-alternating) → ʼináʼniichah, no shift, ’ inserted ✓. ‘Sleep’ and ‘snore’ (H&P (15), non-alternating) → keep the subject ✓. That is 5 of 5 agreeing. LINK-1's entailment prediction (ours, from English glosses) also fits the eight other one-participant HMP verbs. Change of state, telic, no volition → Spec,V: get cold, get fat (two stems), get old, die/get sick, freeze; all shift ✓. Sentient, no endpoint → Spec,v: dance, pray (plus the inherent or inserted ’); neither shifts ✓. Weather and season verbs have no participant slot, so LINK-1 does not apply, and IN-9 (c) gives ’ ✓. No counterexample among these 13 verbs, and the four weather and season verbs fall under IN-9 (c). T38d is partly answered: H&P's 19-verb alternating sample has not been run.

HYPOTHESIS Verdict. T38a: answered. In (10a) the unaccusative's shi- comes before ná-, like the transitive's bi- in (2b) and handout (23a), and like the inserted ’ in (7b). Eccentric agreement as this site stated it (A1s) is contradicted by a printed form. HMP's own eccentric agreement (A1h) is not, because it adds bumping. So eccentric agreement and the oblique analysis no longer differ on word order. They differ on whether the participant stays the subject (HMP: “no diathesis change”, a claim about grammatical function, not role) and on whether nouns are the DPs that check the agreement. No printed form decides either point. T38b: answered as far as the sources' analyses go. The ’ is a IV object, and the handout's “V–VI” names the E domain. A4's reading is not contradicted by any datum, but it has lost its support. Among the slot analyses, A5 is favoured: it uses no expletive (vs A3), it does not make a thematic object a subject (vs A4), and its default is HMP's own stated reason. A1h and A5 make the same predictions about the printed strings. They differ in framework: what checks or links, a DP or a slot. Under the site's ground rules, that is not data. T38e, T38f, T38g: partly answered (§11).

7. Derivations HD1–HD9

HD1. (2b) bi-ná-ʼnii-sh-dl[į́į́]h ‘start to drink it’ (repetitive mode): transitive base with a disjunct, A5

  1. Σ Σ(√DRINK) = {ρA (the drinker: volition, causation; Proto-Agent) = 1sg, Spec,v; ρX (what is drunk: incremental theme; Proto-Patient) = 3, Comp,V}.
  2. A Merge V dlį́į́h + ρX; Merge REP ná as a V-level non-head (disjunct). INCH: ρI takes the object position (IN-1). IN-8/IN-9: ρC = ρX (Comp,V) → [PP ρX PINCH]. IN-5′: the PP is merged as the highest V-level non-head, above ná. Merge v = d, ρA in Spec,v; M/S; Qual [n ii].
  3. A Agree / PORT (IN-4): ρA in Spec,v → 1sg sh on M/S.
  4. A Spell-out: the PP is its own domain → bi (third person overt because oblique; under a 1sg subject a direct object would be Ø, as in (2b′) ná-sh-…). D1: N = {PP, ná, ρI} by C2 → bi ⌢ ná ⌢ ’; H by C3 (= IN-10's right-adjunction order) → nii ⌢ sh ⌢ d ⌢ dlį́į́h. Output bi-ná-ʼnii-sh-dl[į́į́]h ✓ (HMP's segmentation, 3-R-INCH-Ø.1SG-dCL.drink.MU).

HD2. (10a) shináʼniidlóóh ‘I start to get cold (about dawn)’: unaccusative base with a disjunct, T38a

  1. Σ Σ(√GET.COLD) = {ρ1}; Eρ1 ∋ chg (undergoes a change of state, Dowty (28a)); no vol; telic by our reading of the English gloss (telicity OPEN). LINK-1 → Spec,V. φ = 1sg. The role in the action is unchanged by INCH.
  2. A Merge V dlóóh, ρ1 in Spec,V; REP ná as a V-level non-head. INCH: ρI in the object position. IN-9: ρC = ρ1, the only slot below v → [PP ρ1 PINCH]; IN-5′: merged above ná. v = d, no specifier; M/S; Qual [n ii].
  3. A PORT (IN-4): Spec,v is empty; ρ1 is inside a PP (IN-3); ρI is a thematic object (HMP), not eligible. No goal → IN-7 (D): M/S takes its non-local form (VIII “entirely empty”, HMP).
  4. A Spell-out: PP → shi. N by C2 → shi ⌢ ná ⌢ ’; H by C3 → nii ⌢ (M Ø) ⌢ d ⌢ dlóóh. Output string shi ná ʼnii dlóóh ✓ = shináʼniidlóóh. HMP do not segment (10a). The alignment with (2b) is ours.
  5. Comparison, A1s (ID4): ρ1 raises to Spec,EP and is pronounced on E next to ʼnii. The order is then ná ⌢ shi ⌢ ʼnii …, which is ✗ against (10a). A1h: the DP checks AGRo shi on E, and n. 2 then bumps it to Position I, giving shi ⌢ ná ⌢ ʼnii ✓. The bump is a reordering step after the syntax, in template terms. In A5 the same order comes from merge height plus C2.

HD3. (7b) ʼawééʼ ʼináʼniichah ‘the baby starts to cry’: nothing below v

  1. Σ Σ(√CRY) = {ρA} (sentient, no endpoint; LINK-1 → Spec,v). H&P/HMP: the complement is a conflated noun cha, not a slot. φ = 3.
  2. A Merge [V N cha V] (structure (8)); REP ná. INCH: ρI. IN-9: no slot below v, so ρC = a fresh ρD ⟨E none, φ 3, δ INDF⟩ → [PP ρD PINCH]; IN-5′: above ná. v; ρA in Spec,v; M/S; Qual.
  3. A PORT: ρA → 3rd on M/S (Ø).
  4. A Spell-out: PP → ’ (“the 3i object is ’- in both direct and oblique object forms”, PA p. 60); N by C2 → ’ ⌢ ná ⌢ ’; H → nii ⌢ … ⌢ chah. String ʼi ná ʼnii chah ✓ = ʼináʼniichah (unsegmented by HMP; the alignment and the i after the first ’ are not segmented by us either).
  5. B Link ʼawééʼ ‘baby’ to ρA (φ 3). It cannot link to ρD, which has no role. The noun is never part of satisfying INCH.

HD4. (10d) Díí ʼatsįʼ biʼniichxʼǫʼ ‘this meat has started to spoil’ (font restored): the contrast with HD3

  1. Σ Σ(√SPOIL) = {ρ1}: change of state, no sentience → Spec,V. φ = 3.
  2. A IN-9: ρC = ρ1 → PP. PORT: no goal → IN-7 (D).
  3. A Spell-out: PP → bi (oblique third person overt; INV-0 undefined, no ρA). String bi ʼnii chxʼǫʼ ✓. The crucial prefix biʼnii- is undamaged in the extraction. Only the stem vowel is restored.
  4. B Link díí ʼatsįʼ ‘this meat’ to ρ1, inside the PP.
  5. Result: two third-person, one-participant verbs. ‘Cry’ pronounces a placeholder before INCH; ‘spoil’ pronounces its participant there. This is the ‘start to’ test in a single minimal contrast, and IN-6′ (a) vs (c) derives it.

HD5. (1a) yiʼniidiz ‘(my grandmother) started to spin (the black wool)’: voice survives INCH

  1. Σ {ρA = grandmother (3), Spec,v; ρX = wool (3), Comp,V}. INV-0 holds; INV-3: human > wool → EXT = ρA → VOICE = yi-.
  2. A INCH; ρC = ρX → PP; PORT: ρA → 3rd (Ø).
  3. A Spell-out: the PP's pronoun is spelled with the voice exponent yi → yi ⌢ ’ ⌢ nii ⌢ … ✓. That the voice prefix here is the PP's pronoun is our reading (HYPOTHESIS; postposition objects take yi-/bi- as well, one-place verb + P).
  4. B Link shimá sání to EXT = ρA, and the wool noun (next to the verb, INV-6) to ρX.

HD6. n. 4 sodiʼniiszin ‘I started to pray’: blocking

  1. Σ ‘pray’ = theme so=d-l-zin; ρA = 1sg in Spec,v. Following HMP's “cf. (7b)”, nothing is below v, so ρC = ρD (IN-9 (c)).
  2. A IN-5′: the thematic Position I prefix so holds the left edge, so ρD stays in the object position beside ρI. ’ + ’ → ’ (HMP: “hypothetically at least”).
  3. A Spell-out: so ⌢ di ⌢ ’ ⌢ nii ⌢ … ⌢ zin ✓. We do not segment di or the material before zin; HMP give no segmentation. Under A1h the same blocking is HMP's own statement. Under A5 it is IN-5′'s exception clause. That clause is not derived (T39c).

HD7. (16a) ʼiʼniichííl ‘it started to snow’: no participant at all

  1. Σ Σ(√SNOW) = ∅ (no participant; Dowty's entailments do not apply).
  2. A INCH; IN-9 (c): ρC = ρD. PORT: no goal → IN-7 (D).
  3. A Spell-out: ’ ⌢ ’ ⌢ nii ⌢ chííl ✓ (two glottal stops, as in ‘snore’). HMP call these verbs “unaccusatives”, but in the inchoative they behave like unergatives. On IN-6′ this is expected: there is no low slot to shift.

HD8. The future vowel: (2d), (6b) and Hale's shidíʼnóodlóół (u6, T38c)

  1. Printed: (2d) bi-dí-ʼnée-sh-dl[į́į́]ł, HMP gloss 3-F-INCH-PROG.1SG-…; (6b) ʼidíʼníílzhish ‘you'll start to dance’ (unsegmented); PA p. 61 shidíʼnóodlóół ‘I will start to get very cold’ (unsegmented; Hale: [óo] reflects a third person in M/S).
  2. A A5: in (2d), ρA = 1sg → M/S 1sg (vowel ée + sh). In (6b), ρA = 2sg → íí (our alignment). In shidíʼnóodlóół, no goal → IN-7 (D), the non-local form → óo. The vowel after ʼn- is in M/S and tracks the subject. When the participant is in the PP, M/S shows the form it has for every non-local person.
  3. What is not shown: whether óo stays the same when the oblique participant is 2sg or 1pl (T38c). A3, A4 and A5 all predict it does; A1h predicts it too (VIII empty). So T38c can no longer separate the analyses, only confirm the default.

HD9. (3a) Shí gohwééh biʼniishdl[ą́ą́]ʼ ‘I started to drink coffee’: HMP's derivation, restated step by step, then the slot version

  1. HMP (A1h), as they state it. B Merge [VP gohwééh V] (head-final); adjoin the subject DP shí to VP (VPISH); Merge M (Ø + AGRs sh), then E (AGRo b + ʼnii); the clause is head-final at each level. Move the object DP: adjoin to MP, then check AGRo in E. Move the subject DP: adjoin to EP (EPP). A Head movement (IN-10): V right-adjoins to M, and [M V] right-adjoins to E → E ⌢ M ⌢ V = bʼnii ⌢ sh ⌢ d-dlą́ą́ʼ. Two DP movements in the clause, two head movements, and agreement checked by DPs.
  2. A5 (STRATA order). Σ layer 1: ρA (1sg, Spec,v), ρX (3, Comp,V). A layer 2: INCH (event structure: onset) reads the positions; ρC = ρX → PP; PORT → sh; spell-out by C1–C3 → bi ⌢ ’ ⌢ nii ⌢ sh ⌢ d ⌢ dlą́ą́ʼ. B layer 3: link gohwééh to ρX and shí to ρA. No movement.
  3. Same string. The difference is where the nouns are and what they do: in A1h they are DPs that check agreement, and in A5 they are linked to slots. The word-level part is shared. HMP's right-adjoining head movement gives the head order that CYC C3 states, and AGRo before E gives C1. That is how the changeover hypothesis (not established) would look in HMP's terms: head-final phrases, with a head-initial order of heads inside the word.

8. What changes on the site

  • The ‘start to’ test: update notes at the top and in §6. IN-5 is replaced by IN-5′. A1 as in ID4 is contradicted by (10a). A4 (IN-7 (I)) is demoted and IN-7 (D) added. The T38 rows get their statuses.
  • Postposition units: the T32g snippet (“a postposition requires overt agreement”; Speas on b- as an incorporated pronoun) is HMP ms. §2 and n. 3. INCH “behaves like an ‘incorporated’ postposition”.
  • Plain ł-causer: T4 note, the two diagnostics agree on the five shared verbs.
  • A/B as spell-out: T29f oblique half. The oblique's leftward shift has a cause (the object position is full), and a printed blocking case (n. 4). HMP's head movement gives C3.
  • Subject on the head: HMP's “Position VIII … empty, as it is regularly” for non-local persons is noted as a reason for IN-7 (D). Areal ho- does not satisfy INCH, which fits SH's treatment of areal ho- as a deictic subject, not an object.
  • Open questions: new row T39a–h; T38 row statuses.

9. What each source gets right, and what it misses

SourceGets rightMisses or leaves open
HMP ms.The widest data set: all modes of one transitive (2); all three base types; weather and season verbs. The object requirement stated as a postposition-like need. Bumping stated once for every base, with a blocking case. The empty subject position tied to the regular emptiness of non-local persons. No diathesis change (roles fixed). Locality stated as an empirical question.The inserted ’ of ‘cry’ is “not straightforward”, and the mechanism is postponed. “Eccentric” construal keeps the participant a subject although it never shows subject properties in the printed forms. Bumping is a template repair. The areal forms are unresolved (their own words). The nouns are treated as DPs that check agreement, which is a framework choice the data do not force.
Hale handout (10)–(14), (23)The position-by-position layout; the oblique as a P object above the disjuncts (23c).Tree (14) puts the unaccusative participant next to ʼnii on E. (10a) shows it at the far left.
Hale PA paper pp. 57–61One oblique analysis for all bases; the future vowel as evidence of a third person in M/S; in-situ licensing.The expletive is not needed once the empty non-local subject is taken as the default (HMP). Silent on verbs with nothing below v.
Hale 2003 n. 13“when not in competition with a preverb”: the same blocking HMP state.“not a matter entirely devoid of mystery”. HMP's reason (the object position is full) answers part of that.
This site, 4:46amIN-2 (lower slots → oblique), IN-6 (the diagnostic), the prediction that tested A1s.IN-7 (I) leaned on a label (“V–VI”) that names E. IN-5 had no cause. IN-2 said nothing about verbs with no lower slot.

10. Grant's thesis, Dowty, event structure, STRATA, the changeover

  • GRANT'S CONJECTURE → HYPOTHESIS Weak thesis (the arguments are role-bearing pronoun slots; nouns are optional add-ons linked to them): supported again, more sharply. INCH's need is met by a slot in a certain position. It may be a patient ((2), (1)), the undergoer of a change ((10), (11)), an inherent indefinite (‘snore’, ‘dance’, ‘sleep’), or a slot with no role at all (‘cry’, ‘snow’, the seasons). It is never met by a noun, and never by the Spec,v participant. In ‘it started to snow’ no noun and no participant exist, and the need is still met. The strong thesis is not tested.
  • HYPOTHESIS Feature bundles. ρD = ⟨no role, 3rd person, indefinite⟩ is a feature bundle that exists only to satisfy the morphology. It is the clearest case on the site of a verb's argument position being filled by a bundle that is “not quite a word” and not a participant either. HMP describe it as “indefinite agreement” that has nothing to agree with. That is the same object described in agreement terms. Only contradicting data count, and none contradicts the bundle reading. It explains HMP's postponed puzzle without extra machinery. That is an advantage in coverage, not a refutation of HMP.
  • HYPOTHESIS Dowty (baseline). The five verbs in (11) are all change-of-state, non-volitional predicates (Proto-Patient (28a)), and all start low. ‘Die, get sick’ is in Dowty's (73) list of verbs that are unergative in some languages and unaccusative in others. In Navajo the ‘start to’ test puts it with the unaccusatives, as his p. 608 prediction expects for a non-agentive, telic predicate. Sentient participants of activities with no endpoint (cry, dance, pray, sleep, snore) start high. All of this extends Dowty's §12; nothing departs from it. Weather verbs lie outside it, because they have no participant to have entailments.
  • HYPOTHESIS Theta roles and event structure, compatible. INCH is an event-structure operator: an onset, with a momentaneous stem (HMP), built from “inceptive” n- and “punctual” -ii-. It changes no one's role in the action (HMP: “no diathesis change”; INV still sees the patient in (1a)). It reads the positions the roles were linked to. The two descriptions apply to the same participants at different layers.
  • GRANT'S CONJECTURE → HYPOTHESIS STRATA (theta roles first, event structure next, nouns later), tested as a third option beside the adjunct and inside-the-clause views. INCH must apply after layer 1 (it reads Spec,v vs lower) and before nouns (it is satisfied where there is no noun, (16), and is never satisfied by one, (7b)). STRATA fits. The adjunct view also fits. The inside-the-clause view (HMP's DPs) has to describe (7b) and (16) as agreement with no DP, which HMP leave unexplained. That is the first point on this site where the three views differ in coverage: STRATA and the adjunct view cover these forms, and the inside-the-clause view has a gap. They do not yet differ on a contradicting datum. A sharper test would be a noun incorporated into INCH's complement, which n. 3 says some postpositions allow. STRATA predicts none (T39f).
  • HYPOTHESIS The changeover (rule set A vs B). HMP's own derivation (HD9) makes it explicit. The phrases are head-final (VP before M, MP before E), and the word is built by head movement that right-adjoins V to M and M to E, so inside the word the heads come out in head-initial order E ⌢ M ⌢ V. That is Grant's picture (GRANT'S CONJECTURE, not established): right-headed clause syntax and left-headed prefix order, with the word as the changeover. IN-10 makes CYC C3 a consequence of head movement rather than a separate rule. The bumped object shows that a word-internal oblique can land in the disjunct zone, so the disjunct/conjunct line is not the cycle boundary (T2).
  • HYPOTHESIS Phrase-level packaging. ʼnii- is stored as a three-piece unit (a thematic object prefix plus two qualifiers) that also brings a postposition-like requirement. It is a lexicalized piece of structure above the word level. In this it is like the verb + postposition units on postposition units: HMP's own comparison is to an “incorporated” postposition.

11. Tasks: status of T38, new T39a–h

#Status / taskNeeds
T38aSOURCED Answered: (10a) shináʼniidlóóh: the participant's prefix is first. A1s contradicted; A1h survives by bumping.—
T38bSOURCED Answered as far as the analyses go: IV object, thematic (HMP; PA); “V–VI” = E (HMP n. 1). No datum separates A4 from A5.—
T38cOPEN Narrowed: every analysis now predicts a constant non-local form (HD8)Y&M paradigms (2sg, 1pl unaccusative futures)
T38dHYPOTHESIS Partly answered: 5/5 shared verbs agree; 13 verbs, no counterexampleH&P's alternating sample under INCH (Y&M)
T38eHYPOTHESIS Partly answered: ʼnii-'s own ’ is “thematic” (no argument); the inserted ’ of ‘cry’ and the weather verbs is role-less (IN-9)A noun linked to either (none printed)
T38fHYPOTHESIS Partly answered: cause = the object position is full (HMP n. 2); blocking by so- and Ib preverbsT39c
T38gHYPOTHESIS Partly answered: the ms. read (text only); the 2000 printed paper not openedT39a, T39g
T38hOPEN no fourth-person form in HMPY&M
T39aOPEN Page images of hale028: check the restored stems and the tree layouts ((3b), (4), (12b), (13b)), especially whether AGRs follows M and AGRo precedes E.A reachable copy (the box could not reach the MIT host)
T39bOPEN A4 vs A5: does the unaccusative ‘start to’ form show any sign of a third-person subject (a =go clause chain with the same or a different subject, the fourth person in the next clause)?Grant's texts; speaker judgments
T39cOPEN Which Position I prefixes block the shift (“certain”, n. 2)? Thematic only (so-), or also the Ib preverbs of Hale 2003 n. 13? Why is the escape site the far left and not just above IV?Y&M 1980G:320–321; Y&M 1987:69–71
T39dOPEN (15): is the ʼa- of ʼahoʼnii- the indefinite object (HMP) or the Ia adverbial ‘away’ (Y&M)? IN-9 predicts the indefinite object.Y&M 1980D:72; other areal inchoatives
T39eOPEN Weather and season verbs: zero-place (IN-9 (c)), or a low slot that INCH skips? HMP's “unaccusatives” label vs their inchoative behaviour.HMP 2000 printed version; Y&M
T39fOPEN Can an incorporated bare noun satisfy INCH, as n. 3 says it can for some postpositions? STRATA predicts no; HMP's DP view allows it.Y&M; texts
T39gOPEN Hale, Munro & Platero 2000: does the printed version keep eccentric agreement, bumping, and the empty VIII?A copy
T39hOPEN Why momentaneous stems only (HMP §1)? An event-structure restriction (onset needs a punctual stem) or a listed theme property?Smith 1996 (not on the box)

Sources: Hale, Munro & Platero ms. hale028 (text extraction only; access notes in [local working copy, not in the repository], not committed). As recorded on the site: Hale PA paper pp. 57–61, handout (10)–(14), (23), Hale 2003 n. 13, H&P 1996 (11), (12), (15), Dowty 1991 §12, (28a), (73), Willie 1991:23. Cross-links: the ‘start to’ test, postposition units (T32g), plain ł-causer (T4), A/B as spell-out (T29f), subject on the head, inverse, open questions (T39), sources.

Who likes it? Pleonastic ho-, psych verbs, and the experiencer inside a postposition

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.

2026-11-11 audit pass 4. Interlinear rows whose form and gloss do not line up one-to-one were re-checked against the source page. Each is marked † with a note under its table: SOURCED where the source prints it that way, OPEN where the source cannot confirm it. See the pass 4 report.

Working note. Kept as written (URL, data and status labels unchanged). Filed under Event structure and argument realization; the consolidated argument is the essay events. Rule IDs: rules index.

Verbs on this page → dictionary: yiztał ‘kick’ · yóóʼeelwod ‘run away’ · yiʼash ‘walk (two)’ · ditą́ʼ ‘be deep; (with areal ho-) the area is deep’ · nizhóní ‘be nice; with shił ‘with me’: I like it’ · hózin ‘be known; with shił béé: I know it’ · hodiʼil ‘(the area) is dense’ · ʼádílzhééh ‘shave (reflexive: shave oneself; agentive passive: be shaved)’ · ʼádíshdleesh ‘paint (reflexive: paint oneself; agentive passive: be painted)’ · yíchxǫʼ ‘be ruined; (classifier) ruin’ · yishááł ‘I am walking (1sg, progressive)’ · déyá ‘I am going; with noun=góó: I am going (to a place)’

Restated on the unified structure (2026-11-15). The fourth person is restated as a person value on an argument merged in Spec,vP or VP, with nothing extra moving: the optative fourth bound by the wisher (H51), the respect/report/story ji- as an Agree mark (H57a, OPEN against H57b), the areal ho- as a personless area argument (H54a, H60). 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. Position-class template statements below describe the order of prefixes; they are not the derivation (§3). Status on the spine: see unified structure, §4, fourth-person rows. The data, judgments, translations and citations below are unchanged; the analysis below is kept as the record of what was argued then.

This page reads Willie's (1991) two passes over Navajo “pleonastic” subjects and “psych” verbs: §2.5 (pp. 45–49) and its near-copy §4.1 (pp. 102–108). Both are short, so it also reads the material next to §2.5 that bears on the same two phenomena: §2.3 on indefinite agents and patients (pp. 35–39), §2.4 on grammatical vs lexical postpositions (pp. 39–44), and the §2.6 summary with Willie's own definition of the Maximal Verb-Sentence (pp. 49–51). For the proto-role test it sets these beside Dowty's (1991) account of psych verbs (pp. 579–580). The test case is shił nizhóní ‘I like it’, literally “with me, it is nice” (p. 47). The person who likes is not the verb's subject. It is the object of a postposition ‘with’. The subject is the liked thing, or else ho-, a prefix for “an abstract space, area, or entity” (p. 45). The run gives four results. (1) Dowty's account allows this. A psych verb's two participants tie on agent properties, so either one may be the subject (§8). (2) In ‘I know it’, shił béé hózin, both participants are carried by postpositions, and the verb word itself carries only ho- (§5.2). So the roles a verb needs need not be slots in the verb word. Willie places them in the larger verb-plus-postpositions unit (p. 50, (44c)), and that forces a revision of this site's WORD-EDGE rule (§9). (3) Whether a noun may be added depends on the kind of value in a slot, not on any noun. A plain third person takes a noun (‘the water is deep’). The areal ho- takes only a place phrase (‘deep there in the water’). The indefinite prefixes take nothing (§7). (4) Willie calls ho- both “pleonastic (expletive)” and something that “refers to an abstract space”. That leaves open whether every prefix slot is a role, which is a direct question for Grant's thesis (§8.3, T17a).

On this page: 1 Question · 2 Sources · 3 One section printed twice · 4 Areal ho- · 5 Psych verbs · 6 Indefinite agents · 7 Postpositions and the MVS · 8 Proto-role test · 9 Word edge revised · 10 Rules · 11 Derivations H1–H7 · 12 VPISH · 13 Open

1. The question

This site writes a verb's arguments as a set of role slots, Σ(√) = {ρA, ρP, …}, each with Dowty entailments (proto-role schema). Pronominal prefixes value the slots (rule set A), and nouns are only linked to them at clause level (rule set B). Psych verbs and pleonastic subjects test this in three ways:

  • Proto-roles (phenomenon 2). An experiencer is sentient, which is a Proto-Agent property. Where does Navajo put it, and does that fit Dowty's selection principle?
  • Empty slots (phenomenon 2). If a subject is “pleonastic”, it is a slot with no role. Does Navajo have slots that are not roles?
  • Rule set A vs B (phenomenon 1). If a role required by the verb sits on a postposition outside the verb word, where does the verb's schema close?

2. Sources opened this run (2026-10-11, ~12:50pm PT)

  • Willie, MaryAnn 1991. Navajo Pronouns and Obviation. PhD dissertation, University of Arizona (the scan on the box). Fresh 150 dpi page images of printed pp. 35–51 (§2.3–2.6) and pp. 102–108 (§4.0–4.1). Page offset: the page headers show PDF = printed + 5 for Chapter 2 (printed 45 = PDF 50) and PDF = printed + 6 for Chapter 4 (printed 102 = PDF 108). Every Navajo line of (20)–(43) and Ch. 4 (1)–(4) was checked on 300 dpi crops, including the (27), (28), (30) and (37) tables. (17)–(19) on p. 35 were read at 150 dpi only and are not reprinted. No OCR used for any form. The OCR layer was used only to find the table of contents (it prints the §2.5.2 heading as “"Pysch" Verbs”).
  • Dowty, David 1991. “Thematic proto-roles and argument selection.” Language 67(3):547–619. Page images of printed pp. 579–580 (PDF 34–35, at 80, 150 and 300 dpi): §8.3 “Argument selection indeterminacy”, the psych predicates (38), Croft's stative/inchoative observation, n. 21–23.

Transcription and glosses. Navajo as printed, with the glottal stop as ’ and the nasal hook as ą/į/ǫ. Willie's gloss lines are copied exactly, misprints included and marked [sic]. Her gloss order for postpositions is object.P (“1sg.with”). Where she gives no gloss line ((25), (26), (32) in part, (34), (35), (38), (39)), we add none.

Cited by Willie in these pages, not opened: Belletti & Rizzi 1988 (experiencers as internal arguments; Dowty n. 23 cites Belletti & Rizzi 1986); Young & Morgan 1987 (“agentive passive”). Cited by Dowty, not opened: Croft 1986a; Postal 1970; Rogers 1974; Talmy 1985b; Dowty 1982a; Rozwadowska 1988.

3. One section printed twice SOURCED

Willie says why §4.1 repeats §2.5: “Before dealing deal [sic] with the specifics of the fourth person, I will provide a discussion of pleonastic subjects, in order to set them aside as distinct from the fourth person forms. This is crucial since there is some homophony between pleonastic subjects and fourth person forms” (p. 102). The two copies agree in substance and differ in small ways. Every difference we found is listed here, because a later reader should not pick one copy silently.

Item§2.5 (pp. 45–49)§4.1 (pp. 103–108)
‘deep’ditą́ʼ, hoditą́ʼ (final ’) (40a–e)ditą́, hoditą́ (no final ’) (1a–f)
*‘locative PP + plain 3’not given(1f) *táyiʼ ditą́
Second translation of ‘deep there in the water’not given(1e) ‘The water is deep there.’
(c) ‘I like the town’*kintah shil nizhóní, plain l [sic] (41c)*kintah shił nizhóní (2c)
(d) gloss of shił“ls.with” [sic] (41d)none (2d)
(e) mark and translation#kintah-di shił nizhóní ‘I like him/it [when I'm] in town.’ (41e)*Kintah-di shił nizhóní ‘I like it in town [being in town].’ (2e) [sic?]: the text calls it “a good output” in both copies
‘know’: bracketed paraphrase“[about X, knowing is with me.]” (42)“It is known to me by means of it.)” [sic] (3)
‘know’ (43)/(4): gloss of béé“3.about” (43)“3.means of” (4)
Who bił refers to“bil "with me"” [sic] (p. 49)“bił "with him"” (p. 108)
Psych PPs“a postpositional phrase whose object has an Experiencer theta role” (p. 47)“a second postpositional phrase with a first person object prefix” (p. 105), though (2a)–(2c) have only one PP
Status of ho-“an incorporated pleonastic subject pronoun” that “refers to an abstract space, area, or entity” (p. 45)“a verbal prefix ho- that marks a locative, or abstract pleonastic (expletive) subject” (p. 103)

4. Areal ho- in locative sentences (§2.5.1, §4.1.1) SOURCED

P. 45: “It has been claimed that Navajo lacks pleonastic subjects, since Navajo has no non-referential "dummy" lexical items such as the English expletives "it" and "there" which fill subject positions that are not assigned a theta role. Navajo has no expletive nominals, but since nominals are not arguments in Navajo, there is no need for one. Instead, there is an incorporated pleonastic subject pronoun, ho-, which has the same shape as the fourth person object pronoun, to be described in Chap. 4. While the fourth person refers exclusively to human beings, pleonastic ho- refers to an abstract space, area, or entity.” P. 103: “Pleonastic subjects, like other items that occupy argument positions in Navajo, are pronominal affixes, not nominals.” Pp. 102–103: “there is only one pleonastic argument, the subject pronoun ho-, which has the same shape as the fourth person object pronoun. This by itself is evidence against treating these constructions as examples of the fourth person.”

Willie 1991:45–46, (40); 104, (1f)
(40a)ditą́ʼ
3-is deep
‘It is deep.’
(40b)tóditą́ʼ
water,3-is deep
‘The water is deep.’
(40c)hoditą́ʼ
ho-is deep
‘It (the area) is deep.’
(40d)*tóhoditą́ʼ
waterho-is deep
(40e)táyiʼhoditą́ʼ
water:inho-is deep
‘It is deep there in the water.’
(1f)*táyiʼditą́
water:in3-is deep

P. 46: “The presence of ho- in (38) and (40d) excludes an adjoined nominal; it is necessary to enclitize a postposition to the nominal, and build a locative expression”. P. 105: “while the areal prefix ho- can appear both with and without a locative postpositional phrase, a bare nominal adjunct, as in (1d), can not be adjoined when pleonastic ho- is present, and a locative postpositional phrase cannot be adjoined when ho- is not present.”

Willie 1991:44, (37)–(39)
(37a)-tah‘among’(37b)-taʼ‘between’
(38)Tsintah hodiʼil.
‘Among the trees/forest, it is dense.’
(39)Nihitah deeyá
‘Among us he will come.’

P. 44: (37) are “one or two postpositions that can function either as grammatical or lexical postpositions.” P. 46: “Example (38) above also has a pleonastic "areal" subject.”

HYPOTHESIS What each subject value licenses. The three sentences have the same verb and differ in one prefix. The prefix decides what can be added at clause level:

Subject value (A)Bare noun linked (B)Locative PP (B)Example
3 (definite)yes: tó ditą́ʼno: *táyiʼ ditą́(40b), (1f)
ho- (areal)no: *tó hoditą́ʼyes: táyiʼ hoditą́ʼ(40d), (40e)

On this site's terms, ‘water’ fills no slot in any of these. In (40b) it is linked to the 3 subject. In (40e) the locative phrase ‘in the water’ is linked to the areal subject, as the place the area is. The B-level phrase must match the type of the A-level value: an individual for 3, a place for ho-. Willie states the distribution; the matching rule (SLOT-TYPE-LINK, §10) is ours.

5. Psych verbs (§2.5.2, §4.1.2) SOURCED

P. 47: these “include a postpositional phrase whose object has an Experiencer theta role. These are examples of "psych" verbs, where the argument with an Experiencer theta role is comparable to a "Dative Subject" (see Belletti and Rizzi, 1988 on experiencers as internal arguments in "psych" verb constructions).” P. 105 adds: “the argument with an Experiencer theta role is the object of a postposition”.

5.1 ‘I like it’ and ‘I like it there’

Willie 1991:47, (41); 106, (2e)
(41a)shiłnizhóní
1sg.with3.nice
‘With me, it is nice’ ‘I like it.’
(41b)shiłhózhóní
1sg.withho-nice
‘I like it there (the area).’
(41c)*kintahshil [sic]nizhóní
town1sg.with3.is nice
‘I like the town.’
(41d)kintah-dishiłhózhóní
town -atls.with [sic]ho-is nice
‘I like it there in town (the area).’
(41e)#kintah-dishiłnizhóní
town-at1sg.with3 .is nice
‘I like him/it [when I'm] in town.’
(2e)*Kintah-dishiłnizhóní
town-at1sg.with3 .is nice
‘I like it in town [being in town].’ [sic?: see §3]

P. 48: “The postpositional phrase and its pronominal object, which has an Experiencer theta role, is obligatory with these "psych" verbs. We see that the pleonastic subject pronoun can occur without a locative expression but the reverse is not the case. When a bare nominal and the pleonastic subject cooccur, there must be a locative expression such as in (41d). In (41e), we have a good output but the meaning is not consistent with the construction that we are exploring --instead it means "I like him/it in town, [not here where I am now]." The generalization is that a bare nominal cannot appear in a sentence without a personal pronoun to which it is adjoined.”

OPEN A puzzle in (41c). Willie's generalization (a bare noun needs a personal pronoun to adjoin to) does not by itself exclude *(41c). The gloss gives nizhóní a 3 subject, “3.is nice”, and a 3 subject takes a bare noun in (40b) tó ditą́ʼ. So why can't ‘town’ link to the 3 subject of ‘nice’ and mean ‘I like the town’? Willie does not say. Possible answers, none tested: the 3 of nizhóní is not a “personal pronoun” in her sense; or ‘town’ is a place word that must be linked through a locative; or (41c) is out for a reason specific to ‘like’. We do not choose (T17b).

5.2 ‘I know it’: both participants on postpositions

Willie 1991:48–49, (42)–(43); 107, (3)–(4)
(42)Shiłbééhózin
1.with3.aboutho-known
‘[about X, knowing is with me.]’ ‘I know it.’
(43)Kee yóóʼeelwod=ígííbiłbééhózin
3.ran away=COMP3.with3.aboutho.known
‘He/she knows that Kee ran away.’
(4)Kee yóóʼeelwod=ígííbiłbééhózin
3.ran away=COMP3.with3.means ofho.known
‘He/she knows that Kee ran away.’

P. 48: “The pleonastic ho- and Dative experiencer can also appear with another class of "psych" verbs, those referring to knowledge. In the English gloss, the first postpositional object is the subject.” P. 49: “This example is not a Passive; an enlightening gloss is not easy to find. These constructions permit adjoined subordinate clauses.” “for verbs like 'know' the Experiencer is the object of the postpostitional [sic] phrase bil [sic] "with me" and is comparable to a "Dative Subject". I analyze the adjoined subordinate clause as an adjunct to the object of the postpositional phrase béé, "about it".” P. 108: “In this section I have identified the locative and pleonastic uses of the 'areal' subject pronominal prefix ho- which contrasts with fourth person subject ji- and fourth person object ho- prefixes.”

HYPOTHESIS Where the roles of ‘know’ are. ‘Know’ has two participants: the knower (sentient) and what is known. In (42)–(43) the knower is the object of -ł ‘with’ and the thing known is the object of béé. The verb word hózin carries only ho-, which Willie calls pleonastic. So neither participant is a prefix on the verb word. Both roles are present, and both are filled by pronouns, as this site's schema requires. But they sit on two grammatical postpositions, not in the verb word. In (43) the clause ‘that Kee ran away’ is linked (B) to the object of béé, exactly as a noun is linked to a slot. That is Willie's analysis (p. 49), restated.

Cross-link: the same string appears as a relative, shił béé hózin=ígíí ‘that which is known to me’, Ch. 6 (1g), p. 151, glossed there “1.with 3.about 3.known=REL” (relatives). Ch. 6 has “3.known”, Chs. 2 and 4 have “ho-known”. We keep both as printed (T17g).

6. Indefinite agents and patients (§2.3, pp. 35–39) SOURCED

Pp. 35–36: “a group of pronominal prefixes that mark indefinite and non-referential arguments. A crucial feature of these pronominals is that, like the anaphors, they exclude adjoined coreferent nominals … Like anaphors, they show a classifier that marks the construction as intransitive.” P. 36 (§2.3.1, the “Agentive Passive”, Young & Morgan's term): “the indefinite agent must be understood as human, but no noun that describes this agent may be adjoined. A very interesting property of the Agentive Passive is that the pronominal prefix that marks the unspecified agent is identical in phonological shape to the reflexive anaphor … ʼádi-. But note that the form 'adi- in the Agentive Passive in Navajo occurs to the right of (internal to) the object pronoun, proving that it marks the unspecified agent, and is not an object.”

Willie 1991:37, (20)–(23)
(20)ʼádílzhééh(ʼádi-ni-l-zheeh in underlying form)
REF.2s.shaving
‘You are shaving yourself.’
(21)niʼdilzhééh(ni-ʼadi-l-zheeh in underlying form)
2s.PASS.shaving
‘You are being shaved.’
(22)ʼádíshdleesh
REF.1s.painting
‘I am painting myself.’
(23)shiʼdidleesh
1s.PASS.painting
‘I am being painted.’

Pp. 37–38: “In both these construction types, the patient precedes the agent in the prefix array. In the reflexive forms, the anaphor is to the left of the subject, and in the agentive passive forms, the object pronoun is to the left of the unspecified agent.”

Willie 1991:38–39, (24)–(26)
(24)ʼííchxǫʼ
unspec. subj. ruined
‘Ruining took place.’
(25)(hatáál)yíchxǫʼ
‘(The ceremony) it is ruined.’
(26)ʼasdzą́ą́ʼííłchxǫʼ
‘The woman did the ruining (of it).’

P. 38 (§2.3.2.1, the -’a prefix): “This pronoun marks an indefinite inanimate argument. When it occurs with an intransitive verb, it marks an indefinite animate subject; when it occurs with a transitive verb, it marks an indefinite inanimate patient. In [sic] never permits an adjoined coreferent nominal.” (24) “might be used for example when a mistake made in a ceremony required it to be terminated or corrected.” P. 39: “The contrast between (25) and (26) is marked in the classifier -_ [sic: the classifier symbol is not printed], which marks the verb transitive.” “Example (24) is an intransitive construction, with an indefinite inanimate subject marker 'a-, which will not allow a nominal adjunct. In contrast, (26) is a transitive construction with an indefinite inanimate object argument, and permits an adjunct to the third person subject pronoun.”

SOURCED + HYPOTHESIS Position in the word identifies the role. Willie uses prefix order as evidence of role. ʼadi- in (21) is the agent because it is to the right of the object pronoun (p. 36). Her summary is general: “the relative order of the prefixes that mark the verbal arguments is consistent, with the patient always to the left of the agent” (p. 50). On this site's terms that is a rule-set-A linearization stated over roles, ρP ≺ ρA. It holds whatever the English subject is: in the “passive” (21)/(23) the patient is the English subject, and it still precedes the agent. Willie's underlying forms show it directly: ni-’adi-l-zheeh, patient 2s first, unspecified agent second (p. 37).

Animacy note, as printed. P. 38 calls -’a “an indefinite inanimate argument” and then says that with an intransitive verb “it marks an indefinite animate subject”. P. 39 calls the same subject in (24) “an indefinite inanimate subject marker”. We copy all three and do not decide (T17h).

7. Grammatical vs lexical postpositions, and Willie's Maximal Verb-Sentence (§2.4, §2.6) SOURCED

Pp. 39–40: “Postpositions resemble verbs in Navajo in being inflected for their object arguments. The postpositional object pronouns and the possessive pronouns are identical in form. The postpositional phrase corresponds to an oblique argument of the verb. Like the oblique phrases seen in other languages, some are case-marked arguments that the verb is subcategorized for, and some are more like adjuncts. I will call the former the grammatical postpositions, and the latter the lexical postpositions. The grammatical postpositions occur only with pronominal objects and are a component of the MVS. The lexical postpositions occur only with nominal (lexical) objects.”

Willie 1991:40–42, (27), (28), (30)
(27)Sing.DualPlural
1stshi-nihi-da-nihi
2ni-nihi-da-nihi
3yi/bi-yi/bida-bi
4hahada-ha
(28)Grammatical Postpositions(30)Lexical Postpositions
a.-aa‘to’ Dativea.-gi ‘at’
b.-chʼįʼ‘to’ Goal; Allativeb.-di ‘at’
c.-tsʼą́ą́ʼ‘from’ (take away from; Malefactive)c.-dóó ‘from’
d.-á‘for’ Benefactived.-dę́ę́ʼ ‘from’
f. [sic: no e.]-ee‘with’ Instrumentale.-góó ‘to’
g.-ł‘with, together’ Comitativef.-jįʼ ‘to’
h.-ghá‘through’ (penetrate through)
i.-lááh‘beyond’ Comparative, Locative

P. 41: “Of these postpositions, (28 a,c,d,g) occur only with animates, (28 f) occurs only with inanimates, and others occur with both.” P. 42: “Note that these postpositions are all directionals” (of (30), which include two glossed ‘at’).

Willie 1991:41–44, (29), (31)–(36)
(29)shimábil [sic]déʼáázh
1.mother3.withdl.walk
‘I am going with my mother.’
(31)Kinłání=góódéyá
Flagstaff=toI will go
‘I will go to Flagstaff.’
(32)Kinłání=jįʼdéyá
to†
‘I will go as far as Flagstaff.’
(33)ʼawééʼbichʼįʼyishááł
baby3.to1.walking
‘I am walking to the baby.’
(34)*ʼawééʼ-góóyishááł
‘(I am walking to the baby.)’
(35)hooghan=góódéyá
‘I am going home.’ ‘*I am going to the hoghan.’
(36)dibé baghanbitsʼą́ą́jįʼyishááł
sheep 3s.home3s.from.at1s.walking
‘I am walking away from the corral.’

SOURCED † Segmentation note (2026-11-11 audit pass 4): Willie 1991 (32), p. 42, glosses only the postposition =jįʼ (‘to’) and leaves Kinłání and déyá unglossed. Kept exactly as printed.

P. 43: “It is not possible to use -goo [sic: -góó], a lexical postposition, with an animate object.” “Lexical postpositions and their nominal objects can become lexicalized, that is, have a narrow interpretation.” P. 44: “While it is not possible to stack two grammatical or two lexical postpositions, it is possible to have the kind of combination shown in (36), which produces a complex grammatical postpositional phrase.”

Pp. 49–50 (§2.6): “The verb and these postpostional [sic] phrases form a constituent -- the MVS (Maximal Verb-Sentence), that has the following properties: (44) a. it cannot be interrupted; b. it may constitute a single phonological unit; c. it is the domain of the expression of the subcategorizational features of the verb, and d. it behaves as a unit under certain syntactic processes (to be discussed in Chap. 3).” P. 50: “the relative order of the prefixes that mark the verbal arguments is consistent, with the patient always to the left of the agent, and the postpositional phrases leftwards of the direct arguments.” Pp. 50–51: two classes of argument prefixes: “the prefixes that mark definite referential arguments: first, second, third and fourth person; also included here are the anaphors (reflexive and reciprocal)”, and “the prefixes that mark indefinite arguments: the 'adi that appears in the "Agentive Passive", the Impersonal/indefinite -'a, and the pleonastic ho-.” “the definite pronouns permit nominal adjuncts (including free pronouns), while the indefinite pronouns exclude adjoined nominals.”

SOURCED This page-checks the B3 boundary on the headedness page, which had been sourced only through Willie & Jelinek 2000:270. Willie 1991 calls it the Maximal Verb-Sentence (MVS), and (44c) makes it the domain of the verb's subcategorization.

8. Phenomenon 2: the proto-role test

8.1 What Dowty says about psych verbs SOURCED

Dowty 1991:579 (§8.3), (38): Experiencer-subject x likes y, x fears y, x supposes (that) S … beside Stimulus-subject y pleases x, y frightens x, (it) seems (to) x (that) S …. “What I believe sets this class of predicates off from all other natural-language verbs is that (i) the predicate entails that the Experiencer has some perception of the Stimulus—thus the Experiencer is entailed to be sentient/perceiving, though the Stimulus is not—and (ii) the Stimulus causes some emotional reaction or cognitive judgment in the Experiencer. The first of these is a P-Agent entailment for the Experiencer, while the second is a P-Agent entailment for the Stimulus argument. Moreover, these predicates have no OTHER entailments for either argument that are relevant to argument selection (with one possible exception to be discussed directly), which leaves a situation in which each argument has a weak but apparently equal claim to subjecthood.” P. 580: Croft (1986a) observed that “the Experiencer-Subject verbs of this class (lefthand column in 38) are always stative, while the Stimulus-Subject verbs can be either stative or inchoative”. Dowty: “the inchoative interpretation entails a Proto-Patient property in the Experiencer that is not present in the stative”, so in an inchoative the Experiencer “must be the direct object according to the selection principle in 31.” N. 23: “some languages (Atsugewi) have only Experiencer-Subject verbs as basic … This may show that there is some sense in which sentience (Experiencer) outranks causation”.

8.2 Navajo against Dowty HYPOTHESIS

Experiencer (sent)Stimulus (caus of the reaction)Subject prefix of the verb wordFits Dowty?
(41a) shił nizhóní ‘I like it’object of -ł3 subjectStimulusYes. A tie may go either way (p. 579), and Stimulus-subject verbs may be stative (p. 580).
(41b) shił hózhóní ‘I like it there’object of -łho- “(the area)”ho-Yes, if ho- is the Stimulus (the area). Not covered by Dowty if ho- has no role (§8.3).
(42) shił béé hózin ‘I know it’object of -łobject of béého- (pleonastic)Outside the ASP. Neither participant is subject or direct object.

So nothing here contradicts Dowty. His Argument Selection Principle (1991:576, (31); theta schema §1) ranks candidates for subject and direct object. In Navajo psych verbs the Experiencer is never a candidate: it is always on a grammatical postposition. Three honest consequences:

  1. Navajo is the reverse of Atsugewi. In the pages opened, every Navajo psych verb has the Stimulus or ho- as subject, and never the Experiencer. Dowty's n. 23 speculates that sentience may outrank causation. Navajo, in these data, ranks the other way, or does not put the Experiencer in the ranking at all. Within Dowty's tie account this is allowed, not predicted.
  2. Croft's generalization cannot be tested yet. The opened forms are all stative (‘is nice’, ‘is known’). Dowty's account predicts that a Navajo inchoative ‘come to like / come to know’ would keep the Experiencer out of subject position. We have no such form (T17e).
  3. The test that matters most for Grant is not Dowty's. Dowty is about which participant becomes subject. Grant's thesis is that the verb's arguments are a schema of roles, not lexical items. In (42) the two roles are both there, both filled by pronouns, and both obligatory. They are just not in the verb word (§9).

8.3 Is a pleonastic slot a role? OPEN (cuts against Grant's thesis on one reading)

Willie describes ho- in two ways that pull apart:

  • (a) A slot with no role. She calls it “pleonastic (expletive)” (p. 103), in contrast to English “it” and “there”, which “fill subject positions that are not assigned a theta role” (p. 45). Read that way, the verb word has a subject slot that carries no proto-role entailments at all. This cuts against the strong form of Grant's thesis (“only a schema of the theta roles”): a slot in the verb would then not be a role.
  • (b) A role with an abstract value. She says ho- “refers to an abstract space, area, or entity” (p. 45), and the translations say “It (the area)” (40c), “I like it there (the area)” (41b). Read that way, ho- is a slot with a role (the thing that is deep or nice), valued by a person-like feature “area” instead of 1, 2, 3 or 4. Then Σ has no empty slots, and ho- differs from 3 only in the type of value, which is what §4 found (SLOT-TYPE-LINK).

The data opened do not separate (a) from (b). One split looks possible from the translations alone. ‘Deep’ and ‘nice’ have an area reading (b), while ‘know’ (42) has no area in its translation (“knowing is with me”), which suggests (a). If that is right, there are two ho-s, or one ho- whose role depends on the verb. We do not decide (T17a).

9. Phenomenon 1: the roles reach past the verb word; WORD-EDGE revised

SOURCED Three statements from Willie, all on pages opened this run. The Experiencer PP “is obligatory with these "psych" verbs” (p. 48). Grammatical postpositions are “a component of the MVS” (p. 40). The MVS “is the domain of the expression of the subcategorizational features of the verb” and “may constitute a single phonological unit” (p. 50, (44b–c)).

HYPOTHESIS WORD-EDGE needs revising. The possessor raising page proposed WORD-EDGE: the A→B changeover is where the verb's schema Σ closes, and that boundary is structural, not phonological. Psych verbs keep the second half and break the first, if “closes” means the end of the verb word. In shił béé hózin two of the verb's required participants are on postpositions outside the word hózin. So there are two boundaries, and they are different:

BoundaryWhat it closesEvidence
B2, the verb wordprefix linearization (rule set A: prefixes left of the stem, ρP ≺ ρA)p. 36, 37–38, 50
B3, the MVSthe verb's role schema Σ (subcategorization), including the roles valued on grammatical postpositionsp. 40, 48, 50 (44c)

WORD-EDGE′ (replaces WORD-EDGE): Σ closes at the MVS (B3), not at the verb word (B2). The linearization switch Grant describes is at B2 or at B3. Between B2 and B3, the order is already head-final: PPs come “leftwards of the direct arguments” (p. 50), and the object pronoun comes before the postposition (shi- + -ł, from Willie's tables (27) + (28g); the segmentation is ours). So the stretch between B2 and B3 belongs to Σ, as rule set A does, but is ordered like rule set B. This is a concrete datum for T2 (where is the switch?). It favors putting the ordering switch at B2 and the schema boundary at B3. Grant's single changeover point would then be two points. (Later the same day, the nominal adjunction page added two more properties that pivot at B3: the nested default links of nouns, and adjunction to the MVS as a whole.) WORD-EDGE's claim that the boundary is not phonological survives: (44b) says the whole MVS “may constitute a single phonological unit”.

What this does not show. We have not shown that the Experiencer slot belongs to Σ of the verb rather than to the postposition's own one-slot schema, with the verb selecting the PP as a whole. “Obligatory” and “subcategorized” favor the first reading but do not prove it (T17d).

10. Rules

Notation as on the possessor raising page. π(ρ) is a slot's value. We add the value ar for areal ho- and ind for the indefinite prefixes (ʼadi-, -’a). Neither is a person in Willie's sense. OblP(ρ) means ρ is valued by the object prefix of the grammatical postposition P.

HO-PLEO    ho- is the only pleonastic argument: a subject prefix, the same shape
           as the 4th-person object, referring to "an abstract space, area, or
           entity", never to humans.                        (pp. 45, 102–103)
HO-NONOUN  With ho-, a bare nominal cannot be adjoined.      (pp. 46, 105; *(40d), *(41c)?)
HO-LOC     ho- allows (does not need) a locative PP; without ho- a locative PP
           cannot be adjoined.                     (pp. 46, 105; (38), (40e), *(1f))
PSY-EXP    In psych verbs the Experiencer is the object of a grammatical
           postposition (-ł 'with'), "comparable to a Dative Subject"; the PP
           is obligatory.                                (pp. 47–48, 105–106; (41))
PSY-HO     With ho- and a psych verb, a bare nominal needs a locative ((41d));
           "a bare nominal cannot appear in a sentence without a personal
           pronoun to which it is adjoined."                   (pp. 48, 106–107)
PSY-KNOW   'know': Experiencer = object of -ł, content = object of béé, subject
           ho-; "not a Passive"; an adjoined clause is "an adjunct to the
           object of the postpositional phrase béé".     (pp. 48–49, 107–108; (42)–(43))
IND-NONOUN Indefinite prefixes ('adi-, -'a, ho-) exclude adjoined coreferent
           nominals; definite ones (1, 2, 3, 4, anaphors) permit them. (pp. 35–36, 50–51)
AG-PASS    'adi- = unspecified human agent, "to the right of (internal to) the
           object pronoun"; no agent noun.                      (p. 36; (21), (23))
P-LEFT     "the patient always to the left of the agent" in the prefix array;
           PPs "leftwards of the direct arguments".             (pp. 37–38, 50)
IMP-A      -'a: indefinite argument; intransitive subject or transitive patient;
           never an adjoined coreferent nominal; ł-type classifier contrast
           (25)/(26).                                      (pp. 38–39; (24)–(26))
GPP/LPP    Grammatical postpositions: only pronominal objects, part of the MVS,
           subcategorized; lexical postpositions: only nominal objects, more
           like adjuncts; -góó not with an animate object.  (pp. 39–43; (28)–(35))
MVS        Verb + grammatical PPs = MVS: uninterruptible; may be one phonological
           unit; domain of the verb's subcategorization.      (pp. 49–50, (44))
RuleStatusBasis
HO-PLEO, HO-NONOUN, HO-LOC, PSY-EXP, PSY-HO, PSY-KNOW, IND-NONOUN, AG-PASS, P-LEFT, IMP-A, GPP/LPP, MVSSOURCEDWillie's statements and examples as cited. The rule names are ours. The “?” on *(41c) under HO-NONOUN marks that (41c) has 3, not ho- (§5.1).
SLOT-TYPE-LINK. A clause-level (B) phrase X can be linked to slot ρ only if X's type matches π(ρ). Bare noun ↔ π ∈ {1,2,3,4} (individual). Locative PP ↔ π = ar (place). Nothing ↔ π = ind.HYPOTHESISHO-NONOUN + HO-LOC + IND-NONOUN. It generalizes the fourth-person noun ban (NOM-4) to all value types. It does not yet explain *(41c) (T17b).
EXP-OBL. Σ(√ψ) for a psych root contains ρE (sent) valued as Obl-ł(ρE), never as the subject prefix.HYPOTHESISPSY-EXP, from (41)–(43) and (2)–(4) only. Consistent with Dowty 1991:579 (tie). Untested on inchoatives (T17e).
WORD-EDGE′. Σ closes at the MVS (B3). Prefix linearization closes at the verb word (B2). Revises WORD-EDGE.HYPOTHESISMVS (44c) + PSY-EXP + PSY-KNOW (§9).
ROLE-ORDER. Rule set A orders argument prefixes by role, ρP ≺ ρA, not by which one is the English subject.HYPOTHESISP-LEFT + AG-PASS; Willie's underlying forms (20)–(21).
Is π = ar a role, or a slot without one?OPEN§8.3; T17a.

11. Derivations H1–H7, step by step

Conventions as on the derivations page: Σ schema step, A word-internal linearization (prefixes left of the stem), B clause level (complements and adjuncts left of their head). Verbs are not segmented beyond Willie's glosses and her own underlying forms. V-P is the VPISH variant, in which the subject slot raises to Spec,M/SP inside the word. V-H keeps it in situ (Hale). Postposition objects are segmented only as Willie's tables (27)–(28) give them.

H1. (40a) ditą́ʼ and (40b) tó ditą́ʼ ‘The water is deep’

  1. Σ Σ(√DEEP) = {ρTh}, one slot (what is deep). Entailments: stative, no P-A properties (ours).
  2. Merge ρTh in the verb's domain. Agree: the subject prefix is valued π = 3 (gloss “3-is deep”). (V-P) Move ρTh to Spec,M/SP. (V-H) no move. Same string.
  3. A Spell-out: ditą́ʼ. Not segmented. The word closes: A→B.
  4. B (40b): adjoin tó on the left. Link tó → ρTh: allowed, since a bare noun is an individual and π = 3 (SLOT-TYPE-LINK; IND-NONOUN, definite).

H2. (40c) hoditą́ʼ, *(40d) tó hoditą́ʼ, (40e) táyiʼ hoditą́ʼ, *(1f) táyiʼ ditą́

  1. Σ Σ(√DEEP) = {ρTh} as in H1. Here π(ρTh) = ar. Under §8.3 (b) the slot is still ρTh, “the area”. Under (a) it is a non-role slot, and Σ has zero roles. OPEN
  2. Agree: the subject is valued ar, spelled ho- (Willie's gloss “ho-is deep” segments it). A ho- left of ditą́ʼ, giving hoditą́ʼ. A→B.
  3. B *(40d): link tó → ρTh. Blocked: a noun is an individual and π = ar (HO-NONOUN).
  4. B (40e): first build a lexical PP [tá-yiʼ] ‘water:in’, then adjoin it. Link PP → ρTh as the place. Allowed (HO-LOC). The noun ‘water’ is the object of a lexical postposition (GPP/LPP), so it links to no slot of the verb.
  5. B *(1f): the same PP with π = 3. There is no place-type slot to link to, and (1f) shows no other reading (HO-LOC). Compare (41e), where such a PP does get another reading (T17f).

H3. (41a) shił nizhóní ‘I like it’ and (41b) shił hózhóní ‘I like it there’

  1. Σ Σ(√NICEψ) = {ρSt, ρE}. ρE = the one who likes (sent). ρSt = what is liked (its niceness causes the reaction). Entailments are ours, after Dowty 1991:579.
  2. Merge P = -ł ‘with’ (28g, animates only) with ρE as its object. Agree (inside PP): π(ρE) = 1, spelled shi- (27). A object prefix left of P: shił. (Segmentation ours, from Willie's tables.)
  3. Merge ρSt in the verb's domain. Agree: (41a) π = 3; (41b) π = ar, spelled ho-. (V-P) Move ρSt to Spec,M/SP. (V-H) no move. A nizhóní / hózhóní. Not segmented further. B2 closes.
  4. B-ordered, inside Σ: Merge [shił] to the left of the verb word (p. 50, “leftwards of the direct arguments”). The MVS closes here, and so does Σ (WORD-EDGE′). B3 closes.
  5. The ASP is not consulted for ρE, since it is not a subject or object candidate (EXP-OBL). Reading: ‘With me, it is nice’ = ‘I like it.’

H4. *(41c) kintah shil [sic] nizhóní, (41d) kintah-di shił hózhóní, #(41e) kintah-di shił nizhóní

  1. Σ + A as in H3: the MVS is [shił nizhóní] (π(ρSt) = 3) or [shił hózhóní] (π = ar).
  2. B (41d): build the lexical PP [kintah-di] ‘town-at’ and adjoin it outside the MVS. Link it → ρSt (π = ar) as the place. Allowed (HO-LOC). ‘I like it there in town (the area).’
  3. B *(41c): adjoin kintah and try to link it → ρSt (π = 3). SLOT-TYPE-LINK allows this, but Willie stars it. OPEN (T17b). We do not patch the rule to fit.
  4. B #(41e): [kintah-di] cannot link to ρSt, since π = 3 is not a place. It survives as a frame adjunct on the whole clause: ‘I like him/it [when I'm] in town’ (p. 48). HYPOTHESIS: why the same move fails in *(1f) is open (T17f).

H5. (42) Shił béé hózin ‘I know it’ and (43) Kee yóóʼeelwod=ígíí bił béé hózin

  1. Σ Σ(√KNOW) = {ρE, ρC, (ρho)}. ρE = the knower (sent). ρC = what is known. ρho = the subject slot, a role or not (§8.3).
  2. Merge P1 = -ł with ρE: (42) π = 1, giving shił; (43) π = 3, giving bił. Merge P2 = béé with ρC: π = 3 (gloss “3.about” / “3.means of”). We do not segment béé. It might be 3 + (28f) -ee ‘with’ (instrumental), which would fit “by means of it” (3), but Willie does not say so (T17g).
  3. Agree: the subject slot is valued ar, spelled ho-. A hózin. B2 closes, with no participant inside the verb word.
  4. Inside Σ, ordered as in B: shił/bił ≺ béé ≺ hózin. Willie: “In the English gloss, the first postpositional object is the subject” (p. 48). B3 closes.
  5. B (43): adjoin [Kee yóóʼeelwod=ígíí] on the left. Link the clause → ρC, the object of béé (Willie p. 49: “an adjunct to the object of the postpositional phrase béé”). A clause, not a noun, fills the role, so the linked phrase is not a lexical item.
  6. HYPOTHESIS Result for Grant's thesis: both roles of ‘know’ are present as pronominal slots, both are obligatory, and neither is a lexical item. Their slots are on postpositions, inside the MVS but outside the verb word (WORD-EDGE′).

H6. (20) ʼádílzhééh ‘You are shaving yourself’ vs (21) niʼdilzhééh ‘You are being shaved’

  1. Σ Σ(√SHAVE) = {ρA, ρP}.
  2. (20) Agree: π(ρA) = 2. ρP is valued as an anaphor bound by ρA, spelled ʼádi-. (21) Agree: π(ρP) = 2, spelled ni-. π(ρA) = ind, spelled ʼadi- (AG-PASS: human, unspecified).
  3. A ROLE-ORDER, ρP ≺ ρA. Willie's underlying forms: (20) ʼádi-ni-l-zheeh (anaphor = ρP, then 2s = ρA); (21) ni-’adi-l-zheeh (2s = ρP, then ’adi = ρA). The same order of roles holds in both, though different slots hold the English subject.
  4. Both have the l classifier, which marks “the construction as intransitive” (p. 36). HYPOTHESIS: the classifier is computed from how many slots are definite and free (one in each), not from how many roles Σ has (two in each).
  5. B No agent noun may be linked in (21) (IND-NONOUN). (V-P) In (21), does ρP raise to subject? ROLE-ORDER keeps it left of ρA either way, so the prefix order cannot tell (§12).

H7. (24) ʼííchxǫʼ, (25) (hatáál) yíchxǫʼ, (26) ʼasdzą́ą́ ʼííłchxǫʼ

  1. Σ (24)/(25): Σ(√RUINintr) = {ρTh}. (26): Σ(√RUINtr) = {ρA, ρTh}. Willie: the two differ “in the classifier” (p. 39), and (26) is printed with ł. This is the ł-causer pattern of the plain ł-causer page.
  2. (24): π(ρTh) = ind, spelled -’a (gloss “unspec. subj.”). (25): π(ρTh) = 3. (26): π(ρA) = 3, π(ρTh) = ind.
  3. A ʼííchxǫʼ / yíchxǫʼ / ʼííłchxǫʼ. Not segmented beyond Willie's statements.
  4. B (24): no noun can link (IND-NONOUN). (25): hatáál ‘ceremony’ links to ρTh (π = 3). (26): ʼasdzą́ą́ links to ρA (π = 3), and nothing can link to the indefinite ρTh. In every case the slot's value type, not the noun, decides (SLOT-TYPE-LINK).

12. VPISH

HYPOTHESIS Three points.

  1. Where is a pleonastic Merged? In standard VPISH accounts, a thematic subject is first-Merged inside the verb phrase and raises, while an expletive has no role and is Merged directly high. If ho- is pleonastic (§8.3 a), VPISH predicts it is Merged outside the verb's role domain. If it is areal (b), it should behave like any subject and start low. The prefix template could tell, if the areal subject ho- and the fourth-person object ho- occupy different positions. Willie says only that they have “the same shape” (p. 45). Their template positions are not given in the pages opened (T17c).
  2. Nothing raises to subject in psych verbs. Belletti & Rizzi (cited by Willie, not opened) treat the experiencer as an internal argument, and in Italian it can surface high, as a “Dative Subject”. In Navajo the Experiencer stays in its PP, and the English subject is a Navajo oblique. On V-P, only ρSt or ho- raises, inside the word. Psych verbs therefore give no sign of experiencer raising. That is not evidence against VPISH, but it is not support either.
  3. The agentive passive does not reorder. In (21)/(23) the patient is the English subject, but the prefix order is still patient ≺ agent (P-LEFT). If the patient raised to a subject position inside the word, the raising would not change linear order, so prefix order cannot detect it. This matches what the inverse page found for bi-: the decisive test is scope or binding, not order.

13. Open questions

IDQuestionWhat would settle it
T17aIs pleonastic ho- a slot with no role (“expletive”, p. 103), or a role with an abstract value (“refers to an abstract space, area, or entity”, p. 45)? Are there two ho-s, area (‘deep’, ‘nice’) vs none (‘know’)? Reading (a) cuts against Grant's “only a schema of the theta roles”.Diagnostics that separate an expletive from a referential subject: can the area be questioned, counted with da-, or picked up by a later pronoun? Young & Morgan 1987. OPEN
T17bWhy is *(41c)/(2c) kintah shił nizhóní ‘I like the town’ out, when nizhóní has a 3 subject and (40b) tó ditą́ʼ links a noun to a 3 subject?Other nouns with shił nizhóní (non-place nouns); a speaker. OPEN
T17cDo areal subject ho- and fourth-person object ho- occupy the same template position? That is the VPISH test of where a pleonastic is Merged.Young & Morgan 1987 paradigms; a verb with both an areal subject and an object prefix. OPEN
T17dIs the Experiencer a slot of the verb's Σ, closing at the MVS (WORD-EDGE′), or the postposition's own argument, with the verb selecting the whole PP?Whether yi-/bi- or da- on the verb can see the Experiencer; whether the Experiencer can be fourth person and still obey verb-level rules. OPEN
T17eCroft/Dowty (p. 580): Stimulus-subject verbs may be inchoative, Experiencer-subject verbs may not. Does Navajo have an inchoative ‘come to like / come to know’, and is its Experiencer still on -ł?A source with inchoative psych forms. OPEN
T17f*(1f) táyiʼ ditą́ has no rescue reading, but #(41e) kintah-di shił nizhóní has a frame reading (‘when I'm in town’). And §4.1 marks it * (2e) while calling it “a good output”. Why does the frame reading exist for ‘like’ but not ‘deep’? Is * in (2e) a slip for #?More locative PPs with plain 3 verbs. OPEN
T17gbéé is glossed “3.about” ((42), (43), (3)) and “3.means of” ((4); “by means of it”, (3)). Is it 3 + -ee ‘with’ (instrumental, inanimates only, (28f))? And is hózin “ho-known” (Chs. 2, 4) or “3.known” (Ch. 6 (1g))?Y&M 1987; another source for ‘know’. OPEN
T17h-’a: “indefinite inanimate argument”, yet with intransitives “an indefinite animate subject” (p. 38), and (24)'s subject is “inanimate” again on p. 39. Which? And which classifier is the missing “-_” (p. 39)?Another source on -’a (Y&M). OPEN
T17iHow does the agentive passive relate to the inverse? Both make the patient the English subject, but ʼadi- takes the agent out of ranking (indefinite), while bi- ranks two thirds. Is ʼadi- another “exit”, like the fourth person (fourth person)? Compare biʼdistał=yę́ę (relatives (1f)).Agentive passives with two third persons; Willie's later chapters. OPEN

Fidelity notes. Forms are as printed, including the differences between §2.5 and §4.1 (§3). (41c) shil and (29) bil have plain l, beside shił/bił elsewhere [sic]. (41d) glosses shił “ls.with”. (28) has no item e. P. 43 writes -goo without tone marks for (30e) -góó. The p. 39 classifier is printed “-_”. (2e) has * where (41e) has #. (3)'s paraphrase opens no parenthesis. P. 105 has an unbalanced parenthesis after “Rizzi, 1988)”. Willie's “Pysch” (TOC), “Before dealing deal”, “In never permits”, “postpostional”, “postpostitional” and “bil "with me"” (p. 49) are all as printed. In (36) the gloss “3s.home” is for baghan. Dowty's citation of Belletti & Rizzi is 1986 (p. 580, n. 23); Willie's is 1988.

Is event structure built inside the Navajo verb word, alongside the role slots?

Working note. Kept as written (URL, data and status labels unchanged). Filed under Event structure and argument realization; the consolidated argument is the essay events. Rule IDs: rules index.

Verbs on this page → dictionary: neezdá ‘sat down; with yikáá’ ‘on it’: sat down on it’

Restated on the unified structure (2026-11-15). Voice, causatives and events are restated with every theta role assigned inside vP: the classifier spells a value of v⁰ (H59, distribution NOT UNDERSTOOD), a causer is an external argument in Spec,vP, and the passive keeps a silent indefinite agent (H10b). This page’s causative analysis has not yet been restated on the unified structure (OPEN). Status on the spine: see unified structure, §4, causatives 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.

2026-11-11 audit pass 2. Two status badges read “NEW” and “NEW (HYPOTHESIS)”, which are not in the site’s label set. Both are now HYPOTHESIS; the claims are unchanged.

2026-11-11 audit pass 3. Status badges now use only the label set SOURCED, HYPOTHESIS, SPECULATIVE, OPEN, NONE and GRANT'S CONJECTURE; verdict or task words that had been styled as badges (for example Opened, PRE-VPISH, DONE, SUPPORTS, PARTLY ANSWERED) are now plain text or a label with a qualifier. See the pass 3 report.

Grant's question: does the Navajo verb word encode telicity, causation and change of state grammatically, in addition to its pronominal role slots, so that the verb puts together both the event and its participants before any noun comes in? This page sorts what is already in the literature from what would be a new link. Written 2026-10-14 (PT).

Bottom line. Partly yes, and most of that part is already known. Every verb word carries a mode (viewpoint) and an aspectual stem category, and the mode is fused with the subject pronoun in one portmanteau (Hale 2003 ms.:10). Valence and causation are marked in the word too: the ł-classifier, and the causative y- with a causee object (Hale & Platero 1996). The perfective presents an event as bounded (Smith, Perkins & Fernald 2007). Telicity, though, is not a grammatical category in Navajo on Smith's analysis. Morphemes that indicate telicity occur, but telicity has “no consistent distributional correlates” (SPF 2007 §4.1). The one sourced link between event type and argument structure is the Hale & Platero split, and it is about change of state, not aspect prefixes. A direct link between aspect prefixes and theta roles is OPEN. We found no source that states one.

On this page: 1 Sources · 2 Which morphemes carry what · 3 Does the perfective entail completion? · 4 Does aspect constrain roles? · 5 Aspect and argument structure · 6 Rules · 7 Derivations · 8 Known vs new · 9 Open

1. Sources

  • SOURCED (opened) Smith, Carlota S., Ellavina T. Perkins & Theodore B. Fernald 2007. “Time in Navajo: Direct and indirect interpretation.” IJAL 73(1):40–71 (Swarthmore Works copy). Cited as SPF 2007 by section and example number. Page numbers are given only where the copy's running heads show them (pp. 41, 49, 51, 52, 65, 67).
  • SOURCED (opened) Hale, Ken 2003 (ms.). “On the significance of Eloise Jelinek's Pronominal Argument Hypothesis.” Manuscript pp. 10, 29 (and n. on Y&M 1987:167–199).
  • SOURCED (opened) Hale, Ken & Paul Platero 1996. “Navajo reflections of a general theory of lexical argument structure” (see causative split).
  • OPEN (cited, not opened) Smith 1996, “Aspectual categories in Navajo,” IJAL 62:227–63; Smith 1997, The Parameter of Aspect (2nd ed.), Navajo chapter; Midgette 1996, “Lexical aspect in Navajo: the telic property”; Young & Morgan 1987:144–189 (modes and aspectual categories); Kari 1979; Axelrod 1993; Rice 2000:283–303. Everything we say about these comes through SPF 2007 or Hale and is marked as secondhand.

2. Which morphemes carry what (question 1)

SOURCED Smith's two-component model: situation type (states, activities, accomplishments, achievements, semelfactives) and viewpoint (SPF 2007 §2.1, n. 2). In Navajo, “aspectual viewpoint is expressed by certain mode morphemes” (§2.1). Hale places mode in the head M, “the primary functional head above the verbal projection … It consists of two aspectual elements, Situation Aspect and Viewpoint Aspect,” assembled with the subject “into a portmanteau (Mode/Subj)” (Hale 2003 ms.:10).

CategoryWhat it contributes (source)Telicity?Navajo example (source)
Perfective modeViewpoint: “makes events visible … with endpoints (atelic, durative events have arbitrary final endpoints)” (§4.1)Bounded, not necessarily telic(19d) Shimá chʼiyáán laʼ bá nahálniiʼ ‘I bought some groceries for my mother’; (23a) Hooghan binishishnish ‘I did some work on a hogan’ (§4.2)
Imperfective modeViewpoint without endpoints. Can focus an internal or a preliminary interval (§4.1)No(18a) ʼAwééʼ tsáskʼeh yikááʼdóó ʼadah ʼiigeeh ‘The baby is about to fall off the bed’ (Midgette 1995:61, via §4.1)
Progressive modeImperfective type, “available only for motion verbs”, internal intervals only. Needs a Cursive stem (§4.1, n. 33)No(18b) … ’adah yigoh ‘The baby is falling off the bed’
Customary (Usitative)“Derived atelic events” or derived states (§3.1, (25))Makes atelic(25a) ʼAbínígo gohwééh yidlą́ą́h ‘S/he usually drinks coffee in the morning’
IterativeDerived activities (§4.2.2)Makes atelic(26b) ʼAwééʼ náchah ‘The baby cries over and over again’
Future, OptativeNo viewpoint (neutral)—(6a) Deeshchah ‘I will cry’ (§3.2.1)
Aspectual stem categories (“Verb Lexeme Categories”)Smith 1996 via SPF n. 36: “Transitional bases, and Continuatives under certain circumstances, express events with the feature of telicity”. Conclusive and Durative are durative. Cursive and Continuative are “super-lexical”. Momentaneous “does not have a consistent meaning”. The Semelfactive's consistency “is in doubt”.Transitional: telic(19b) Jáan nídiiʼnééh ‘John is getting up’ (Momentaneous, n. 38)
Cessative / completive prefixesMark the final endpoint (§4.2.2)Yes(28b) … kintahjįʼ bichidí nidoolbąs ‘S/he will drive his/her car to town in an hour’. Needs “the morpheme ni expressing the final endpoint” (p. 65). (16b) Shįįdą́ą́ʼ diyogí ninítłʼó ‘Last summer I finished weaving the rug’ (cmpl)

Navajo forms are as printed in the Swarthmore copy of SPF 2007, whose text layer garbles some nasal hooks. Where we could not resolve a character confidently we give the gloss, not the form. (28) and (29) are on p. 65.

3. Does the perfective entail completion? (question 2)

SOURCED No, it entails boundedness. The Perfective presents events “with endpoints”, but “atelic, durative events have arbitrary final endpoints” (SPF 2007 §4.1). So an atelic perfective stops without being completed: (23a) ‘I did some work on a hogan’. Telic events are “intrinsically bounded”, and perfective boundedness is “semantic: it is coded by the linguistic forms”, while zero-marked boundedness is pragmatic and can be overridden (§4.1). The perfective is not a past tense: with kʼad ‘now’ it is either out or read as recent past (§3.2.3, (17)), and it can be future (32a–d), e.g. (32c) ‘Tomorrow when you come to see us we will have already eaten’ (p. 67).

SOURCED How it interacts with telicity. Completion comes from telicity, not from the perfective. Telicity is shown by an adverbial test: tʼáá łáʼí ʼahwééʼílkidjįʼ ‘up to an hour’ means ‘for an hour’ with atelic verbs ((27a) herd sheep, (27b) sleep) and ‘in an hour’ with telic ones. The telic reading needs ʼałtso ‘completely’ in (28a) ‘I will write the whole paper in an hour’ or the cessative ni- in (28b) (pp. 64–65). Smith's caveat is that telicity “does not” have distributional correlates. “Navajo does not have distinct adverbials conveying duration vs. completion … Nor does Navajo have verbal pairs such as stop vs. finish (Smith 1996). The situation types of Navajo, then, are States, Punctual events, and Durative events” (§4.1).

4. Does aspect type constrain which theta roles a verb can take? (question 3)

OPEN No source we opened says that a mode or aspectual category selects agent, theme or incremental theme. What is sourced comes close but is not that claim:

  • SOURCED Situation type “is built up compositionally from the verb or verb base, its arguments, and relevant other forms” (SPF 2007:41). So arguments feed aspect. The reverse (aspect restricting arguments) is not claimed.
  • SOURCED In (28a), naaltsoos ʼałtso ʼadeeshłííł ‘I will write the whole paper’, the telic reading needs ʼałtso ‘all of it’. HYPOTHESIS That is an incremental-theme effect in Dowty's (1991) sense: the event ends when the whole object is done. But ʼałtso is a free adverb or quantifier outside the verb word, so here telicity is not built before the noun comes in. This cuts against the strongest form of Grant's question.
  • SOURCED The Progressive is restricted to motion verbs with Cursive stems (n. 33). That is a restriction to a verb class, not to a role.
  • SOURCED (secondhand) Young & Morgan 1987:144–189 list “constraints on the co-occurrence of Modes with different classes of verb base” (SPF n. 18). Whether any of those classes are defined by roles has not been checked (ES-c).

5. Is any source linking aspect or event structure to argument structure? (question 4)

  1. SOURCED Change of state ↔ the transitivity alternation. Verbs like ‘boil’ alternate with just the ł-classifier. Verbs like ‘cry’, ‘swim’ and ‘play’ cannot, and need the causative y- with a causee object (Hale & Platero 1996, (4)–(13)). H&P explain this with Hale–Keyser lexical structures: the alternating verbs have a theme “internal to the lexical structure”, and non-alternating verbs have only an external argument. This is a sourced link between event type (change of state vs. activity) and argument structure. It is marked inside the word (ł vs. y- + ł). H&P also say there is “no ‘meaning’ attached to the ł-classifier”: it marks the alternation and does not itself encode causation. Our earlier causative split page tests this against Dowty's entailments.
  2. SOURCED Aspect and the subject are fused. Mode (situation + viewpoint aspect) and the subject pronoun form one portmanteau in M (Hale 2003 ms.:10). That is a morphological link between aspect and a role slot. It is not a semantic one.
  3. SOURCED Maintained-state causatives (keep vs bring about) already show the word distinguishing causing a change from keeping a state.
  4. OPEN Midgette 1996 (“the telic property”) and Smith 1996 are the most likely places for a statement tying telicity markers to the object or theme. Neither has been opened.

6. Rules

ES-1  Every event verb word has exactly one mode; Perfective/Imperfective/
      Progressive = viewpoint; others = neutral viewpoint.      SOURCED (SPF §3.1)
ES-2  Mode + subject = one portmanteau in M, above vP.          SOURCED (Hale ms.:10)
ES-3  Perfective ⇒ bounded; telic ⇒ completed; atelic perfective ⇒
      arbitrary endpoint.                                        SOURCED (SPF §4.1)
ES-4  Telicity is semantic (event structure), not a grammaticalized
      situation type in Navajo.                                  SOURCED (SPF §4.1; Smith 1996)
ES-5  ł alone = transitivity alternation, only for verbs with an internal
      theme (change of state); else y- + ł causative.            SOURCED (H&P 1996)
ES-6  The verb word assembles Σ (role slots) AND viewpoint + situation
      category before any noun is adjoined.                      HYPOTHESIS (from ES-1, ES-2,
                                                                 Willie 1991:51–52)
ES-7  Telicity can depend on material outside the word ('ałtso, incremental
      object).                                                    SOURCED fact (28a);
                                                                 incremental-theme reading HYPOTHESIS
ES-8  Aspect category restricts the role frame.                   GRANT'S CONJECTURE / OPEN

7. Derivations (VPISH-aware)

E1. (23a) Hooghan binishishnish ‘I did some work on a hogan’ (SPF 2007 §4.2)

Σ     {ρA (vol), ρGoal/P via bi- 'on it'}                    HYPOTHESIS (labels)
Merge ρA Spec,v; root √WORK; P 'on' with 3 object            HYPOTHESIS
M     Perfective viewpoint + 1s subject portmanteau            SOURCED (gloss 1subj-perf; ES-2)
Move  V-P: ρA → Spec,M/S                                       HYPOTHESIS
A     spell-out binishishnish — bounded, atelic ⇒ "some work"  SOURCED (ES-3)
B     adjoin hooghan, linked to the P object                   SOURCED pattern (Willie ADJ-PP)
Result: viewpoint and role slots are set before the noun; completion is NOT entailed.

E2. (28a) naaltsoos ʼałtso ʼadeeshłííł ‘I will write the whole paper (in an hour)’ (p. 65)

Σ     {ρA, ρTh}; future mode (neutral viewpoint) + 1s          SOURCED (gloss)
A     verb word closed: no viewpoint bound, telicity unset      SOURCED (ES-1, ES-4)
B     adjoin naaltsoos + 'ałtso ⇒ telic ('in an hour')          SOURCED (p. 65)
Result: telicity is fixed only at clause level. Counterexample to "all
event structure is assembled before nouns" (ES-7).

E3. ‘boil’ vs ‘cry’ (H&P 1996)

boil: Σ = {ρTh} (internal); + ł ⇒ {ρA, ρTh}                  SOURCED (alternation) / HYPOTHESIS (Σ)
cry:  Σ = {ρA} (external only); + ł ✗; y- + ł + causee ⇒ causative   SOURCED
Result: the change-of-state subevent (an internal theme) licenses the
simple causer slot inside the word, before any noun.

8. Already known vs new

ClaimStatus
Viewpoint (mode) is in the verb word; mode is fused with the subjectKnown (SPF 2007; Hale 2003)
Perfective = bounded, not completedKnown (SPF 2007)
Telicity marked by some morphemes but not grammaticalizedKnown (Smith 1996; Midgette 1996; SPF 2007)
Change of state ↔ simple ł transitivization; causative y-Known (Hale & Platero 1996)
The word assembles role slots and viewpoint together, before nouns (ES-6)HYPOTHESIS (a new combination): follows from combining the PAH (Willie) with Smith/Hale. Nobody we read states it as one claim
Telicity can be settled outside the word (ES-7)Fact known (SPF (28)); its bearing on Grant's thesis is new (HYPOTHESIS), and it limits the thesis
Aspect restricts role frames (ES-8)OPEN, no evidence yet

9. Open questions

#QuestionWhat would settle it
ES-aDoes Midgette 1996 tie telic markers to a theme or incremental object?Read Midgette 1996, pp. 305–30
ES-bSmith 1996's lexical / super-lexical / formal split: does any category select a role frame?Read Smith 1996, IJAL 62:227–63; Smith 1997 Navajo chapter
ES-cY&M 1987:144–189 mode–verb-class co-occurrence constraints: are the classes defined by roles?Read Y&M; cross-check with the dictionary
ES-dIs ni- (cessative/terminative) restricted to verbs with a goal or theme slot?Y&M; Kari 1979; Axelrod 1993 (Koyukon parallel)
ES-eDoes telicity under the perfective change with the object's quantity (bare vs ʼałtso), as incremental-theme theory predicts?Speaker judgments
ES-fRice 2000:283–303 treats stem suffixes as derivational. Do they encode situation type close to the root, under the role slots?Read Rice 2000