Results on the spine: first-principles reviews 1–13, restated as derivations on one clause structure
Verbs on this page → dictionary: yiztał ‘kick’ · néidííłtsʼin ‘hit with the fist’ · jizloh ‘rope’ · yiyiisxį́ ‘kill (one)’ · dabiiłtsą́ ‘see’ · yi-sh-cha ‘cry’ · neezdá ‘sit down; make sit; keep (somewhere)’ · doogááł ‘arrive, will go / come (future)’ · ní ‘say’ · nizin ‘think; wish (with an optative clause)’ · dah diiyá ‘leave, set out (one person)’ · neezdá ‘sat down; with yikáá’ ‘on it’: sat down on it’
2026-11-14 overhaul (approved by Grant for the live site, 2026-11-14). This page is new. It restates what first-principles reviews 1–13 (2026-11-10 to 2026-11-14) found, using the single clause spine set out on the derivation page. Nothing here is new data. Every Navajo example is copied from the example index under its NX number, with its judgment and translation as printed there. Hypothesis numbers (H4″, H23 and so on) are the reviews’ own, and each section links to the review that states and tests them. Where a review’s wording and this summary differ, the review is the record and this page is the restatement.
Every result below is stated the same way. First the sourced data. Then the derivation on the spine: what is merged where and which theta role it gets there, the movements that are assumed in every tensed clause (the verb V→v→T, the subject from Spec,vP to Spec,TP), and then what else moves, which is where the real claims are. Each derivation ends in a logical form whose scope order matches c-command. Labels: SOURCED for what a source prints, HYPOTHESIS for our analysis, OPEN for what is undecided. The same results, with the noun phrase, the verb word and the older topics added, are placed in one table on One structure for Navajo (2026-11-15).
On this page: How to read this page · 1 Nouns and participants · 2 The inverse · 3 Question words · 4 Sentential negation · 5 Constituent negation: hanii · 6 Questions with negation · 7 Quantifier scope and da- · 8 Relative clauses · 9 Questions with “you think” and “she said” · 10 Reported speech · 11 =go clauses · 12 Passives · 13 Telicity, light d and light n · 14 The question grid · 15 What resists, and what is open
How to read this page
HYPOTHESIS The spine is the same in every derivation (full statement: derivation §S2):
[CP … [NegP [TP DP_subj [T′ [vP ⟨DP_subj⟩ [v′ [VP … ⟨V⟩] ⟨v⟩]] V+v+T]] Neg⁰] C⁰] 1. Merge V with its internal argument(s): V assigns their roles (Theme, Patient, Goal …). 2. Merge v; merge the external argument in Spec,vP: v assigns its role (Agent, Experiencer …). 3. Merge T. Working default: V→v→T; the subject moves Spec,vP→Spec,TP; lower copies are silent. 4. Sentential negation, when present: Neg⁰ above TP. Questions, when present: C⁰ above that. 5. What else moves: stated case by case, below. Each extra step is a HYPOTHESIS. 6. LF: scope order = c-command at the position where the expression is interpreted.
Head direction is not fixed here. With T⁰ final (hypothesis HS-1 on the headedness page), the raised verb is pronounced last in TP and nothing else has to move for verb-final order. With T⁰ initial (HS-2), the remnant VP must also move above T in every clause. The derivations below are written for HS-1, and each one says what HS-2 would add. Which is right is OPEN. “Subject” means the DP in Spec,TP; it is not a synonym for Agent.
1. Nouns and participants: the one-noun rule (H4′, H4″)
SOURCED The same two nouns swap who saw whom when the verb changes from yi- to bi-: ashkii atʼééd yi yiiłtsą́. (NX-001; Speas 1986:249 (4.6a)), translated “The boy saw the girl”; ashkii atʼééd b-iiłtsą́. (NX-002; Speas 1986:249 (4.6b)), translated “The girl saw the boy”. With two nouns and ‘kick’: łį́į́ʼ dzaanééz yiztał (NX-029; Willie 1991:71 (37a)), translated “The horse kicked the mule.”; dzaanééz łį́į́ʼ biztał (NX-030; Willie 1991:71 (37b)), translated “The mule was kicked by the horse.” A lone noun can go with the agent when its meaning rules out the other role: ʼashkii yiyííbį́į́ʼ (NX-027; Willie 1991:61 (22)), translated “The boy picked them. (c.f. berries)”. (review 1 and review 2.)
HYPOTHESIS H4″ (H4 in review 1, H4′ in review 2, restricted to third persons in review 10): a single nominal is linked to the lowest third-person participant, unless its meaning cannot bear that participant’s role. Only the override follows from the theta-role premise; the default ranking is an added hypothesis.
Derivation of NX-029 (direct). Merge V ‘kick’ with the mule, which gets Patient. Merge v; merge the horse in Spec,vP, where it gets Agent. Merge T; V→v→T; the horse moves to Spec,TP. What else moves: nothing under HS-1 (agent, patient, verb is the printed order). LF: ∃e[kick(e) ∧ Agent(e,horse) ∧ Patient(e,mule)].
Derivation of NX-030 (inverse). The same Merge steps and the same roles. Working default as before. What else moves: the patient moves to a position above the raised agent (H3, below), and the verb is spelled out with bi-. LF: ∃e[kick(e) ∧ Agent(e,horse) ∧ Patient(e,mule)], the same as NX-029. The inverse changes which phrase is highest, not who does what to whom. Its English passive translation is not evidence of Navajo passive morphology.
2. The inverse: the patient moves over the agent (H3, H6′, H9′)
SOURCED Willie says the inverse patient “must be definite” (1991:77). A polar question can be hosted by the inverse’s first noun, łííʼ-ísh iiʼniʼ biisxį́ (NX-053; Willie 1991:92 (76b)), translated “Did the lightning kill the horse?”, but not when that position is taken by the fronted agent, *iiʼniʼish łííʼ biisxį́ (NX-052) (review 2). The question clitic =sh can move off its question word onto the first word in the direct form, Díí=sh háí yiztał (NX-085; Willie 1991:211 (16)), translated “This person kicked who?”, but not in the inverse, *Díí=sh háí biztał (NX-086; Willie 1991:211 (17)), translated “[This person was kicked by who?]” (review 3).
HYPOTHESIS H3: in the inverse the patient (O) is moved to a position above the agent (S); the upper copy is pronounced and the lower copy keeps the theta role, so the Move assigns no new role. H6′: only a definite or presuppositional (“which one”) O can make this Move. H9′: yi- needs a distinct third-person co-argument above it; elsewhere b-. H21b (review 7): the moved O passes through an outer Spec,vP, where da- also counts it.
On the spine. The working default raises the agent to Spec,TP in both voices. What else moves in the inverse: O, from inside VP to a position above the raised agent. Its exact landing site is OPEN: an outer Spec,vP passed on the way (H21b), then a position above Spec,TP. Its trigger is also OPEN; the animacy ranking (Witherspoon 1977, Creamer 1974) describes when the inverse is used, but it is not yet stated as a trigger.
Order of operations (review 3’s first such result). In NX-085 the direct agent ‘this person’ is in Spec,TP and moves on to Spec,CP, where =sh is written after it. In *NX-086 the demonstrative is the inverse patient that H3 has already raised, and it cannot host =sh. So the H3 Move happens before the question clitic is placed. Whether the block is intervention (H13-int) or a focus restriction on the host (H13-host) is OPEN. LF of NX-085: ASK_x:person(x)[∃e[kick(e) ∧ Agent(e,DEM) ∧ Theme(e,x)]]. Tree: headedness §9.4.
3. Question words sit where nouns sit (H11, H12, H14, H36)
SOURCED A question word stands where a noun with the same role would stand, in both voices: John háísh yiztsʼǫs (NX-067; Willie 1991:204 (7b)), translated “John kissed who?”; John háísh biztsʼǫs (NX-069; Willie 1991:204 (8b)), translated “John was kissed by who?” Without =sh and with a modal clitic, the same word is an indefinite: John háí-shį́į́ yiztał (NX-084; Willie 1991:210 (15)), translated “John (probably) kicked someone.” (Willie 1991:210). Phrases such as “which boy” and “how many boys” behave like ordinary nouns (review 11).
HYPOTHESIS H11: a question word is an indefinite plus Q; =sh is or spells out Q. H12: Q can be displaced to the clause edge (H12-move, syntax; H12-pf, phonology only: OPEN). H14: a question-word object can be fronted over the subject when meaning keeps the roles apart (Willie pp. 206–207). H36: a noun plus a following “which” or “how many” is one nominal with one theta role.
On the spine. The question word is merged in its theta position, like any noun, and nothing else has to move: C⁰[Q] takes scope over the clause and is linked to the question word by Agree. Fronting (H14) is one extra Move to Spec,CP. LF of NX-067: ASK_WH(λx ∃e[kiss(e) ∧ Agent(e,John) ∧ Patient(e,x)]).
4. Sentential negation: da is the negation (N1, N2′, H23, H17, H18a)
SOURCED One doo … da is one “not”: Nidlóóhísh doodaiiʼ dooísh nidlóoh da? (NX-205; Navajo Questions database entry 804), translated “Are you cold or not?” (review 4). The position of da, not of doo, decides which clause is negated: Jáan Mary doo chidí neidiyoołnih da yóʼní. (NX-180; Perkins & Fernald 2014:88 (38a)), translated “John expects Mary not to buy a car.”; Jáan Mary doo chidí neidiyoołnih yóʼní da. (NX-181; Perkins & Fernald 2014:88 (38b)), translated “John doesn’t expect Mary to buy a car.” (review 6). A minimizer said before doo is still read under the negation: Béeso łáʼí ndi doo naashʼáa da. (NX-287; Perkins & Fernald 2014:104 (20 (§5.3))), translated “I haven’t so much as one dollar.”; Díí chidí bikeeʼ łáʼí ndi doo ditsʼid da lá. (NX-288; Perkins & Fernald 2014:105 (21 (§5.3))), translated “Not a one of these tires is any good.”; without a negation it is out, *Béeso łáʼí ndi naashʼá. (NX-289) (review 8). An h-word with -da under doo means “any-”: Doo háí-da níyáa-da. (NX-218; Hale & Platero ms. (1a)), translated “No one has arrived.” A bare -í- on the verb stands in for a missing argument, and which argument is missing depends on order: Doo béégashii yizloh-í-da. (NX-228; Hale & Platero ms. (9)), translated “Nobody roped the cow.”; Béégashii doo yizloh-í-da. (NX-229; Hale & Platero ms. (10)), translated “The cow didn’t rope anything.” (review 5).
HYPOTHESIS N1: doo … da contributes exactly one ¬. H23 (review 8): the negation is the final head Neg⁰, spelled da; doo (and tʼáadoo) is not its specifier but a lower marker adjoined inside the complement of Neg⁰, in the same clause. N2′: ¬ covers the clause that da closes. The strict N2 of review 4 (“¬ covers exactly what lies between doo and da”) is contradicted by NX-287/288 and is kept only as history. H17: the -da on an h-word is agreement with Neg⁰, not a second ¬. H18a: the -í- argument really moves, into Neg⁰’s domain, and is then written on the verb word (a PF step).
Derivation of NX-287. Merge V ‘have’ with the minimizer phrase, which gets Theme. Merge v; merge the silent first-person holder in Spec,vP (first person is marked only in the verb word). Merge T; V→v→T; the holder moves to Spec,TP. Merge the lower marker doo inside TP (H23; its exact site is OPEN). Merge Neg⁰ da above TP. What else moves: the minimizer phrase moves to a position left of doo but still inside TP, so da still c-commands it. This movement is needed on any analysis; under H23 it does not take the phrase out of the negation. LF: ¬∃x[dollar(x) ∧ MIN(x) ∧ ∃s[have(s) ∧ Holder(s,speaker) ∧ Theme(s,x)]]. Under HS-2 the remnant VP would also move above T, staying under Neg⁰.
OPEN One example resists the default. Shibéeso łáʼí ndi ʼádin. (NX-292; Perkins & Fernald 2014:106 (31 (§5.3))), translated “I don’t have even one bit of money.” The negative verb ʼádin carries the negation inside the verb word. If its single argument raises to Spec,TP like any subject, T (which holds the negative verb) does not c-command it, yet the minimizer is read under the negation. So either that argument is interpreted at its lower copy (reconstruction), or it does not raise. Which one is open (H26, review 8).
5. Constituent negation: hanii negates one phrase (H20)
SOURCED ʼAshxosh hanii; shinááʼ yeeʼ hanályįįh. (NX-242; Perkins & Fernald 2014:93 (4 A (§3.2))), translated “I am not sleeping; I was just resting my eyes.” *Hérii dibé nayiisniiʼ hanii. (NX-246; Perkins & Fernald 2014:95 (17 (§3.4))), translated “It is not buying sheep that Harry did.” (intended reading). hanii cannot license the strong NPIs that sentential negation licenses: *Yishtłʼóí hanii, ʼashxaał gaʼ. (NX-285; Perkins & Fernald 2014:104 (16 (§5.2))), translated “(I’m not weaving anything; I’m drumming.)” (intended reading); *Béeso ndóʼ hanii naashʼá. (NX-298; Perkins & Fernald 2014:107 (8 (§5.4))), translated “(I don’t have any money.)” (intended reading) (review 6 and review 8).
HYPOTHESIS H20: hanii is merged right after a maximal projection (XP), and that XP is its focus. ¬ applies to the focus, and the rest of the clause is presupposed. It is not Neg⁰ and does not take the clause. LF of NX-242: assertion ¬[ιP.P(speaker) = sleep], presupposition ∃P.P(speaker), and the second clause supplies the correction.
On the spine, *NX-246. With V in T and the subject in Spec,TP, the object stays in VP and the verb is in T, so “object + verb” is not a phrase. H20 then predicts that hanii cannot follow “object + verb”, which is the printed star. An analysis in which the verb does not move (labelled contrast) leaves “object + verb” as one phrase and needs an extra rule to exclude it. This is one starred example, it depends on H20, and its two follow-up predictions (no whole-clause focus reading; hanii possible right after the object) need a speaker or another source: OPEN.
Grant’s split holds in the grammar. Sentential negation (Neg⁰ da, with doo) and constituent negation (hanii on an XP) differ in what they license, not only in meaning (H17 sharpened in review 8).
6. Questions with negation: Q over ¬, and WHY over ¬ (H15)
SOURCED A negative yes/no question asks about the negative proposition: Łeetsʼaaʼísh tʼáadoo táásinigiz ? (NX-308; Perkins & Fernald 2014:121 (5 (§10))), translated “Didn’t you wash the dishes?”, with the source’s answers (review 8); Doo ndóʼísh yiyą́ą da? (NX-184; Perkins & Fernald 2014:107 (5-NDO)), translated “Don’t you eat at all?” In every negative why-question the question word and its marker stand left of the negative frame: Haʼátʼéego shąʼ tʼáadoo díníyáa da? (NX-138; Fernald et al. 2022 (4)), translated “Why didn’t you go?”; Haʼátʼéegoshąʼ tʼáadoo ʼazeeʼálʼį́į́góó nisíníyáa da? (NX-210; Navajo Questions database entry 695), translated “Why didn’t you go to the hospital?” (review 4). A negative clause with doo hanii … da and no question word is read as “why not”: Doo hanii kótʼéego ʼánílééh da? (NX-247; Perkins & Fernald 2014:97 (28 (§3.6))), translated “Why donʼt you make it like this?” (review 6).
HYPOTHESIS On the spine. C⁰[Q] is merged above NegP and c-commands Neg⁰, so the question scopes over the negation. Where -ísh is written (before the negative word in NX-308, on or after doo elsewhere) is a fact about its host, not about where Q is interpreted (H12). H15: a reason or manner question phrase is merged or moved above NegP and scopes over ¬. LF of NX-184: ASK_POL(¬ EVEN ∃e[eat(e) ∧ Agent(e,you)]) (tree on review 8). LF of NX-210: ?r. REASON(r, ¬∃e[go(e) ∧ Mover(e,you) ∧ Goal(e,hospital)]). What gives NX-247 its question force is OPEN.
7. Quantifier scope and the plural prefix da- (H22a/b, H21a/b)
SOURCED tʼáá ʼałtso ʼatʼééké ʼashkii łaʼ deiztsʼǫs (NX-267; Willie 1991:80 (60)), translated “Every girl kissed a boy. (one boy)” / “Every girl kissed a boy. (many boys)” (review 7). da- needs a plural agent wherever the agent is pronounced: *ʼashkii ʼatʼééké dayiiłtsą́ (NX-273; Willie 1991:82 (63d)), translated “(The boy saw the girls.)” (intended reading); *ʼashiiké ʼatʼééd dabiiłtsą́ (NX-277; Willie 1991:83 (65b)), translated “(The boys were seen by the girl.)” (intended reading). In the inverse it also restricts the patient: *ʼatʼééd ʼashiiké dabiiłtsą́ (NX-278; Willie 1991:83 (65c)), translated “(The girl was seen by the boys.)” (intended reading).
HYPOTHESIS On the spine. “Every girl” is merged in Spec,vP and raised to Spec,TP; “a boy” is merged in VP. So the universal c-commands the indefinite, which gives the “many boys” reading directly: ∀x[girl(x) → ∃y[boy(y) ∧ ∃e[kiss(e) ∧ Agent(e,x) ∧ Patient(e,y)]]]. The “one boy” reading, ∃y[boy(y) ∧ ∀x[girl(x) → ∃e[kiss(e) ∧ Agent(e,x) ∧ Patient(e,y)]]], needs one more step, which is what else has to happen: either “a boy” moves covertly above the subject (H22a) or łaʼ introduces a choice function bound at the top of the clause (H22b). Which is right is OPEN. Pronounced order alone does not fix scope.
HYPOTHESIS H21a: da- is a head (Dist⁰, a placeholder name) merged right above vP that agrees with the closest argument it c-commands; under VPISH that is the agent in Spec,vP. On the spine the verb’s path V→v→T then passes through Dist⁰ (the Head Movement Constraint allows no skip). H21b: the inverse patient passes an outer Spec,vP inside Dist⁰’s reach, so it is counted too. Why demonstrative, possessed and pronoun patients are exempt is not understood: OPEN.
8. Relative clauses: nothing moves out (H27b, H28′, H31, H34)
SOURCED With no noun inside, the enclitic picks the patient under yi- and the agent under bi-: yiztał=yę́ę (NX-312; Willie 1991:152 (1c)), translated “the one he kicked”; biztał=yę́ę (NX-313; Willie 1991:152 (1d)), translated “the one who kicked him”. An overt pronoun stays inside its postposition phrase: bichʼįʼ yáshtiʼ=yę́ę (NX-314; Willie 1991:152 (1e)), translated “the one I spoke to”. With one noun inside, the noun takes one participant and the enclitic the other: ʼashkii yiztsʼǫs=yę́ę yóóʼeelwcd (NX-331; Willie 1991:162 (22)), translated “The one who kissed the boy ran away.” / “The boy, who was kissed by her, ran away.” With a fourth-person participant, the enclitic goes with the third person: haztał =yę́ę (NX-319; Willie 1991:154 (4)), translated “the one who kicked that person”. A clause with =ígíí can name a fact: Mary łį́į́ʼ nayiisniiʼ=ígíí yíniiʼ (NX-372; Willie 1991:180 (67)), translated “That Mary bought a horse, I heard about it.” / “I heard that Mary bought a horse.” (review 9 and review 10).
HYPOTHESIS H27b: the enclitic is an overt head D⁰ that takes the whole clause (CP) as its complement and is linked to one participant inside by Agree; nothing moves out of the clause. Its rival H27a (a silent operator moves to Spec,CP, leaving a gap) is not supported: the overt pronoun in NX-314 shows there is no gap. H28′: the enclitic is linked to the lowest third-person participant that has no noun inside; if every third-person participant has a noun, to the upper one. H31: a noun, a relative DP and the enclitic are third person, and can be linked only to a third-person participant. H34: an unlinked D⁰ names the fact the clause expresses.
On the spine. Inside the relative, the clause is built exactly as a main clause: Merge with theta roles, V→v→T, the subject to Spec,TP, and in the inverse the H3 Move. Then D⁰ merges with CP and Agrees with one participant. What else moves: nothing. LF of NX-312: ιx ∃e[kick(e) ∧ Agent(e,he) ∧ Patient(e,x)]. LF of NX-313: ιx ∃e[kick(e) ∧ Agent(e,x) ∧ Patient(e,him)]. LF of NX-331, reading (a): ∃e2[run_away(e2) ∧ Agent(e2,r)], r = ιx ∃e1[kiss(e1) ∧ Agent(e1,x) ∧ Patient(e1,boy)]. Willie’s own rule that the head is always an absolutive is contradicted by her NX-319, NX-331 (a), NX-314 and NX-316.
OPEN Not understood: *ʼatʼééd biztał=yę́ę ʼashkii nábidííłtsʼin (NX-345) against ʼatʼééd biztał=yée ʼashkii léiʼ nábidííłtsʼin (NX-346), which differ only in léiʼ.
9. Questions with “you think” and “she said”: two clauses side by side (H40a)
SOURCED John háísh yiztsʼǫs nínízin (NX-378; Willie 1991:214 (25)), translated “John kissed who, you think?”, with the paraphrase “(Who do you think John kissed?)”. hádááʼshąʼ dah diiyá ní (NX-384; Willie 1991:217 (31)), translated “When did he leave, she said.” hádááʼshąʼ ʼání, dashdiiyá=hágíí ʼáhóótʼiid (NX-385; Willie 1991:218 (32)), translated “When did she say it, his leaving took place.” Willie reports that (31) has only the reading in which “when” asks about the leaving (review 11).
HYPOTHESIS H40a: the clause before nínízin or ní is a complete clause with its own C⁰ and Q; the “think/say” clause is a separate CP adjoined to it, whose Theme is a silent pronoun resolved to the first clause. A “when” in the first clause can then modify only that clause, with no extra condition. H40b: the first clause is the Theme complement of “think/say”. It gets the order and the paraphrase for free, but it needs H40c (nothing of the matrix clause may be pronounced before the embedded clause) to block the missing reading. On present data H40a is cheaper; H40b is not excluded.
On the spine. Both clauses are built with the working default. What else moves: nothing; the question word stays in its own clause. LF of NX-384 under H40a: ASK_WH(λt ∃e2[leave(e2) ∧ Agent(e2,he) ∧ at(e2,t)]), plus the adjoined ∃e1[say(e1) ∧ Agent(e1,she) ∧ Theme(e1,q)].
10. Reported speech: who “I”, “you” and “he” can be (H41–H44)
SOURCED Every example is a verb glossed “arrive” followed by a verb of saying (Willie 1991, ch. 5 §5.3, pp. 143–147). deesháał ní (NX-386; Willie 1991:143 (21)), translated (a) “He said "I will be there".” or (b) “He said that I would be there.”; doogááł ní (NX-387; Willie 1991:143 (22)), translated “Heᵢ said heⱼ will be there.”; deesháał jiní (NX-388; Willie 1991:144 (23)), translated (a) “That personᵢ said that heᵢ would go.”, with (b) *“They say I will go.” and (c) *“That person says that I will go.” excluded as readings; jidoogááł jiní (NX-391; Willie 1991:145 (26)), translated (a) “Heᵢ said that heᵢ would come.” or (b) “They say that person will come.”; díínááł nijiní (NX-394; Willie 1991:147 (30)), translated “That person said [of you] that you should go.” No Navajo string on these pages is starred; the stars exclude readings. Willie notes that these are “the only subordinate clause types that do not have an enclitic that serves as determiner or nominalizer” (p. 149) (review 12).
HYPOTHESIS H41: a first or second person in the reported clause is read either in the actual speech situation or in the reported one (H41a with H40a: the reported clause is a root clause whose context is fixed by the say clause; H41b with H40b: an optional context-shifting operator in the embedded C⁰). H42: where Navajo has a dedicated form for “the sayer” (a shifted “I”, or a fourth person with a fourth-person sayer), a plain third person is read as someone else; this derives Willie’s (27b), two plain third persons are disjoint. H43: the fourth-person subject of jiní is either a specific respected person or nobody in particular (the quotative “it is said”). H44 (a stipulation): with a specific fourth-person sayer, a first or second person in the reported clause must be shifted, unless that participant is also an argument of “say”, as the second-person object of nijiní is in (30). Why a respected sayer forces the shift is NOT UNDERSTOOD.
On the spine. All eighteen clauses use the working default: V→v→T, and the silent subject moves Spec,vP→Spec,TP. Because every subject is silent and T⁰ is final, nothing else moves. The only order question is how the say clause comes second: a final V (H40b) or adjunction to the right (H40a; if adjunction can only be leftward, the reported clause needs one extra Move). LF of NX-387: p = FUT ∃e2[arrive(e2) ∧ Agent(e2,y₃)], ∃e1[say(e1) ∧ Agent(e1,x₃) ∧ Theme(e1,p)], x₃ ≠ y₃ (under H40b p is inside the Theme by c-command; under H40a by the silent Theme pronoun).
OPEN H40a against H40b is not decided by the person facts. Each analysis derives (21)–(29) with the same added H41 (one version each), H42, H43 and H44, because those facts are about who a pronoun can refer to, not about constituent structure. One smaller point favours each side: the missing enclitic (p. 149) suits H40a (H40b needs H40d, a bare CP can be an argument only of “say” and “think”); (30) is easier to state under H40b, as binding by the object of “say”. H40a stays favoured only on review 11’s “when” contrast (§9). The obligation reading of a second person (28) is derived by neither (OPEN).
11. =go clauses: the second clause reuses the first clause’s people (H45–H49)
SOURCED Willie 1991, ch. 5 §§5.0–5.2, pp. 132–142. yiztał=go néidííłtsʼin (NX-399; Willie 1991:135 (7)), translated “As heᵢ kicked himⱼ, heᵢ hit himⱼ.” A fourth-person patient in the second clause with all third persons in the first is starred: *yiztał=go náhodííłtsʼin (NX-403; Willie 1991:137 (10)). jizloh=go ʼáádóó yiztał (NX-407; Willie 1991:138 (13)), translated “That personᵢ roped himⱼ, and then heₖ kicked himⱼ.”, against *hazloh=go ʼáádóó yiztał (NX-408; Willie 1991:138 (14)). A bare noun as the second patient is starred, *yiztał=go ʼáádóó ʼatʼééd néidííłtsin (NX-409; Willie 1991:140 (16a)), but an indefinite one is fine: yiztał=go ʼáádóó ʼatʼééd léiʼ néidííłtsin (NX-410; Willie 1991:140 (16b)), translated “Heᵢ kicked himⱼ and then heᵢ hit a girl.” A main-clause noun subject follows the =go clause: yah ʼííyáa=go ʼatʼééd dah neezdá (NX-397; Willie 1991:134 (4)), translated “When he came in, the girl sat down.” Event against person: Mary yicha=go yiiłtsą́ (NX-411; Willie 1991:141 (17)), translated “I saw Mary crying.”; Mary jicha=go hoołtsą́ (NX-412; Willie 1991:141 (18)), translated “I saw Mary when she was crying.” (review 13).
HYPOTHESIS H45 (replaces H32c, Willie’s “parallel processing” as a stipulation): in two transitive clauses joined by =go, (a) the second clause’s patient, if definite, must be one of the first clause’s participants, while an indefinite patient (léiʼ) may be new; (b) the second clause’s agent is the first clause’s agent whenever their persons match; (c) a plain transitive verb cannot have the same individual as agent and patient. H45 predicts every good and every starred sentence on these pages, including *(10), (13) against *(14), and *(16a) against (16b). It is still mostly a stipulation: why only two transitive clauses force pairing, and why the patient half is strict, is NOT UNDERSTOOD. H46: a noun where a silent pronoun would do names someone else. H48: a pronoun that refers to an event is third person; the fourth person refers only to persons, which derives the event/person contrast of (17)–(20) from person alone. H49: a noun in Spec,CP with no theta role can be a sentence topic (Mary in (18)). Review 12’s H42 explains only half the pattern (why coreference is possible, not why it is forced), and its general “the dedicated form wins” version is contradicted by NX-395.
On the spine. All forty clauses use the working default, and nothing here resists it. H47: the =go clause is a CP, with =go in C⁰, adjoined at the left of the main clause at the TP level or higher. It is forced by NX-397: the girl is the main-clause subject, raised to the main Spec,TP, and the =go clause is pronounced before her. Objects stay in VP (*NX-406, NX-410); the topic of (18) is merged in Spec,CP, not moved. LF of NX-399: WHEN(∃e1[kick(e1) ∧ Agent(e1,x₃) ∧ Patient(e1,y₃)], ∃e2[fist.hit(e2) ∧ Agent(e2,x₃) ∧ Patient(e2,y₃)]); the =go clause c-commands the main TP, matching its place as WHEN’s first argument.
OPEN (9a)/(9b), a source conflict. For yiztał=go nábidííłtsʼin (NX-401; Willie 1991:136 (9a)) and biztał=go néidííłtsʼin (NX-402; Willie 1991:136 (9b)), Willie’s printed subscripts keep who does what across the yi-/bi- switch (H45b-θ: pairing by first-merge theta role), but her prose says the alternation can “mark a change in thematic roles” (p. 136), and (9b) ends in “(back)” (H45b-pos: pairing after the inverse Move). The source does not settle it, and we do not pick one by guessing which part is a slip. Also OPEN: H48a against H48b (is the =go clause the Theme of “see” in (17), or the usual time clause?), favoured H48b on economy only; and why H49 does not rescue review 10’s *NX-359 (NOT UNDERSTOOD).
12. Passives: a silent indefinite agent in Spec,vP (H10b, H16a)
SOURCED Questions whose verbs the source glosses as passive ask about a place or an amount, never the doer, and have no by-phrase: Háadi lá dibé nidahaniih? (NX-213; Navajo Questions database entry 2199), translated “Where are sheep being sold?”; Naʼazhǫǫshgi díkwíí shį́į́ shaa woobįįh? (NX-215; Navajo Questions database entry 2076), translated “How much will I lose at the hoop and pole game?” (review 4). A relative built on a passive picks out the patient: biʼdistał=yę́ę (NX-315; Willie 1991:152 (1f)), translated “the one who was kicked” (review 9).
HYPOTHESIS H10b / H16a: the passive still merges an agent in Spec,vP, as VPISH requires for the role, but it is silent and indefinite, cannot be linked to a noun, and is existentially closed. The rival H10a / H16b (no agent in syntax; the verb’s meaning adds ∃x Agent) is not excluded. On the spine the silent agent raises to Spec,TP like any subject (as drawn in review 4’s passive-question figure); whether the theme can raise instead is OPEN. LF of NX-213: ?λl.∃x∃e[sell(e) ∧ Agent(e,x) ∧ Theme(e,sheep) ∧ Loc(e,l)]. Prediction (H10b): no question, relative or noun ever picks out the silent agent. Every sourced case so far fits.
13. Telicity, light d (withdrawn) and light n (OPEN)
OPEN Telicity. In thirteen reviews no pair differed only in telicity, so no general surface effect of telicity is supported or contradicted. Two narrow claims are contradicted. First, =yę́ę attaches to a state, łichííʼ=yę́ę (NX-310), to activities, and to the telic “fix”, chidí ʼínáályaa=yę́ę yáʼátʼééh (NX-367), so it is not restricted to either kind (review 10). Second, the form of =go does not change with telicity: the same =go attaches to “come in” and “sit down” (NX-395–398), to “kick” and “rope” (NX-399–410) and to “cry” and “carry” (NX-411–414) (review 13; which of these are bounded is our classification, not the source’s). In review 13 the event/person contrast of (17)–(20) holds with the verbs, and so their telicity, kept the same, so that contrast is not driven by telicity. Whether =go is read “while” with activities and “when / as” with bounded events is OPEN (the verbs differ). Review 12 had one verb only. Mode and viewpoint labels (perfective, imperfective) are not telicity tests.
SOURCED Light d: withdrawn (Grant, 2026-11-14). Grant: 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.” It is no longer a hypothesis under test. The record of the tests stays: in reviews 1–13 a silent determiner on bare nouns never did any work; a silent d that is always definite is contradicted by the bare noun translated “a horse” in NX-372 (review 10); in review 12 a silent d with the meaning of =hágíí over a reported clause would wrongly make the reported content a fact. The determiner-like elements that do work are all overt: the enclitics =yę́ę and =ígíí, -í-, léiʼ and łaʼ. See derivation, light d withdrawn.
OPEN Light n. Grant (2026-11-14): if DP and TP are built alike, the noun phrase may have a light n, parallel to light v in the clause. This is a structural question, not a premise of the spine, and it is separate from light d’s job of covering particular facts. Nothing in reviews 1–13 bears on it (review 13 says so explicitly).
14. The question grid (cumulative after review 13)
SOURCED cells name indexed examples; OPEN cells have none verified. An empty cell is not a judgment that such sentences are ungrammatical. “Passive” follows the source’s gloss; an English passive translation of a bi- verb fills no passive cell.
| Question cell | Active / direct | Inverse / bi- | Passive / impersonal (source gloss) |
|---|---|---|---|
| Polar, affirmative | NX-119–134, 142, 203, 204, 291, 307 | NX-053; *NX-052 | OPEN (NX-217 candidate only) |
| Polar, negative | NX-153–161, 184, 205–209, 308 (answer NX-309) | OPEN | OPEN (declaratives NX-216, NX-236 only) |
| Wh, affirmative | Review 3 set; NX-113, 212, 220, 221, 374–376, 378–385 | Review 3 set | NX-213, 215 |
| Wh, negative | NX-138, 210, 211; NX-247 (by reading) | OPEN | OPEN |
Still empty: polar affirmative × passive; polar negative × inverse and × passive; wh negative × inverse and × passive. These are the cells that would test whether the inverse Move (H3) or a passive interacts with Q and Neg. Reviews 12 and 13 had no questions, negation or passives, so no cell changed; the inverse declaratives (8), (9a), (9b) of review 13 are translated with English passives but glossed INV by the source, so they fill no passive cell.
15. What resists the default, and what is open
- OPEN NX-292 (ʼádin): its single argument must reconstruct or not raise (§4).
- OPEN *NX-246: fits the default given H20; its two follow-up predictions need a speaker or a source (§5).
- OPEN Head direction of T⁰: HS-1 (final, nothing else moves) against HS-2 (initial, remnant VP moves). See headedness and clause spine.
- OPEN Prefix order inside the verb word: whether the sourced prefix orders fit Grant’s hypothesis that the word is linearized opposite to the clause because of where its heads are Merged (restated 2026-11-15; unified structure §3, derivation §S2).
- OPEN The inverse Move: its trigger and landing site (§2).
- OPEN Where doo adjoins under H23, and so whether a subject left of doo needs a second move (NX-223/229, NX-288).
- OPEN H22a against H22b; H40a against H40b (review 12’s person facts are neutral, §10); H10a against H10b; H13-int against H13-host; H48a against H48b (§11).
- OPEN (9a)/(9b) (NX-401/402): Willie’s printed subscripts and her prose disagree, so H45b-θ against H45b-pos is a source conflict (§11).
- OPEN NOT UNDERSTOOD: H44 (why a respected sayer forces “I” and “you” to shift, §10); the core of H45 (why two transitive clauses force pairing, and why the patient half is strict, §11); why H49’s topic parse does not rescue *NX-359.
- OPEN Light n (DP/TP symmetry with light v; Grant, 2026-11-14): a structural question, untested (§13). Light d is withdrawn.
- OPEN *NX-278 and *NX-345 / NX-346: not understood.
