Navajo derivational movement — verb-internal arguments, VPISH, questions
What’s claimed and how well supported
- Labels on this page
- REAL ×1 SPECULATIVE ×2
- Strongest label used
- REAL
- Main caveat (quoted)
- Chemistry half is ◆ SPECULATIVE pedagogy only, and it is not Bridge A.
Counted automatically from the labels printed on this page. Counts are a reading aid, not a grade of the page.
Terms on this page (8), plain-language definitions
- Host (matrix)
- Majority element/phase whose lattice dominates the material.
- Valency (ling.)
- Arity: how many actants a predicate takes; formal vs semantic valency can diverge.
- Valence (chem.)
- Bonding capacity / oxidation-related electron count of an atom in context.
- Dislocation
- Line defect enabling plastic slip at realistic stresses; edge vs screw are the two classic geometries.
- Vacancy
- Point defect: a missing atom site in the lattice.
- Grain boundary
- Interface between crystals in a polycrystal; barriers in Hall–Petch strengthening.
- Solid solution
- Solute atoms dissolved into the host lattice (substitutional or interstitial) without a separate precipitate phase.
- Solute
- Intentional minority species dissolved in the host (alloying addition).
Definitions quoted from the Chem for linguists primer glossary.
Date: 2026-10-02 PT
Status: Draft. Linguistics map first — same-domain derivations below are REAL. Chemistry half is SPECULATIVE pedagogy only, and it is not Bridge A.
Audience: A reader trained in syntax/semantics (VPISH, movement, theta roles) with less chemistry background.
Does not: announce a joint field, rewrite Bridge A, or treat a chemistry parallel as a linguistic result.
Prior project notes (not primary literature): secondary index; secondary index; secondary index; Linguistics: theta. This note does not contradict those files. Where they sketched the prefix template and the pronominal-argument debate, this note asks a narrower question: once theta-role bearers are verb-internal, where, if anywhere, they move, under the project assumption of the VP-internal subject hypothesis.
Do not merge with the typed-participant map. secondary index asks which participant types a mechanism family can wear. This note is movement. The two claims stay separate.
Abstract. In the Navajo sources opened for this note, person–number material inside the verb word is the locus of argument structure: either the arguments themselves (Hale’s statement of Jelinek’s pronominal-argument hypothesis) or agreement / mode-exponents that discharge or index those roles (the Speas / Perkins / Platero side, as Fernald 2000 reports it, and Young & Morgan’s paradigmatic treatment of subject-on-mode as McDonough describes it). arg1, on first use here, means the external argument — the agent/subject theta-role bearer — merged inside the verb phrase (VPISH), not a separate lexical NP. The derivation this note actually draws does not move arg1 to a lexical subject position outside the verb word. Hale’s essay puts arg1 in Spec of the voice head v and then treats surface order as spell-out; Young & Morgan, via McDonough, fuse subject person with mode inside the core verb. A movement of arg1 to Spec of Mode (Hale’s Infl/T analogue) is written only as a flagged overlay that the opened PA essay rejects. Polar questions, in the grammar that was opened, add a clause-initial particle da’ and/or an enclitic -ísh; the arguments do not extract. Content questions use independent h-words. In the clearest opened example the h-word sits in the object’s nominal position while 3–3 morphology stays on the verb. English-style wh-extraction of a verb-internal theta bearer is not what those examples show, and the dissertation that argues for some leftward h-word movement was not opened. The chemistry half, after the break, is a speculative pedagogical map of rearrangement inside one hostMajority element/phase whose lattice dominates the material., not a joint field, and not Bridge A.
- 0. What was opened, and what was not
- 1. Where the arguments are, before any movement
- 2. Worked derivations
- 3. Polar questions: a high marker, no argument movement
- 4. Content questions: an independent h-word; the verb-internal material stays
- 5. What the linguistic map is, in one place
- 6. Chemistry half — rearrangement inside one host
- 7. Sources actually consulted
- 8. What could not be verified
- 9. Open questions
0. What was opened, and what was not
Orthography below is Young & Morgan practical orthography when the opened source preserved it (acute vowels, nasal hooks, ł, glottal ’). Hale’s essay PDF and Fernald 2000’s scan lost diacritics in extraction for many later examples. Those strings are not reconstructed. If a form is quoted, it is from a stretch that came through cleanly, or the source itself printed a segmentation.
| Opened | Not opened (do not treat as read) |
|---|---|
| Hale, “On the Significance of Eloise Jelinek’s Pronominal Argument Hypothesis” (MIT essay PDF; no year on the title page that was fetched; internal citations include Hale 2000b, Hale 2001, Willie & Jelinek 2000) | Young & Morgan 1987 The Navajo Language as the book. Template facts below are as reported by McDonough and Hale from that book. |
| McDonough, “How to Use Young and Morgan’s 1987 The Navajo Language” (Rochester CLS working paper; file path fall 2019) | Willie 1991 dissertation; Willie & Jelinek 2000 article (fetch timed out) |
| Fernald, Chavez, et al. 2019, Navajo Questions: grammar and examples (Navajo Language Academy) | Schauber 1979 The Syntax and Semantics of Questions in Navajo (dissertation/book) |
| Fernald, Denny, et al. 2021, Navajo Verb Modes: grammar and examples (Navajo Language Academy) | Fernald & Platero NLA grammar Chapter 17 (HTTP 504) and Chapter 3 (HTTP 504) |
| Fernald 2000, “Generalizations in Navajo,” The Athabaskan Languages, 51–72 (Swarthmore scan; many example lines are OCR-corrupt and are not quoted as forms) | MITWPL catalog page for Speas 1990 (HTTP 500 on direct fetch). A search snippet of that abstract is not used as a tree. |
| DoITPoMS, “Diffusion mechanisms” (Cambridge teaching page) | Activation energies, diffusivities, concentrations |
| Wikipedia, “Segregation (materials science)” (digest, not a primary paper) |
Competing analyses, not a winner. Fernald 2000 §3.1.1, which was opened, states the split and does not take a side: work following Jelinek and Willie treats overt nominals as adjuncts and the verb-internal pronouns as the arguments; Perkins 1978, Platero 1978 and 1982, and Speas 1990 treat the verbal morphemes as agreement and the overt nominals as arguments. This note draws the pronominal-argument merge structure, because that is the structure Hale’s opened essay actually diagrams. It does not draw a Speas chain. The Young & Morgan / McDonough fact that sits beside both theories: in the descriptive grammar, subject person is an exponent on mode, not a free NP.
Bridge A (concentration windows ↔ valencyArity: how many actants a predicate takes; formal vs semantic valency can diverge./saturation) is a different speculative strand. It is untouched here. This note is about movement inside a system, not about count.
1. Where the arguments are, before any movement
1.1 The descriptive core (Young & Morgan via McDonough)
McDonough, reporting Young & Morgan 1987:g37–38, divides the verb into disjunct, conjunct, and stem. The positions she lists include, among others: position 0, object of a postposition; IV, direct-object pronouns; V, deictic subject pronouns; VII, mode/aspect conjugation; VIII, subject pronouns; IX, the so-called classifier (she prefers valenceBonding capacity / oxidation-related electron count of an atom in context.); X, the stem. She stresses that Young & Morgan do not build words by concatenating the template. The core verb, present in every fully inflected form, is mode inflected for person and number (positions VII+VIII as one paradigmatic piece) plus classifier+stem (IX–X).
Two of her quotations from Young & Morgan 1987:g206 are usable as-is:
- yishcha (ø) (vi): cry, weep. Segmentation she gives for the same shape elsewhere: (y)ish = Ø-imperfective, 1st singular, plus a null-classifier stem.
- yibéézh ‘it’s boiling’ vs yiłbéézh ‘he’s boiling it’ (Young & Morgan via McDonough, root BÉÉZH): null classifier vs ł classifier. This is a valence alternation, not the yi-/bi- inverse. It is recorded so it is not later confused with inverse voice. The separate causative scout treats this pair as the short ł transitivizer, records a longer morphological causative (biyeeshdloh), and refuses the Japanese lexical vs productive -sase split. Some ł is thematic, not derived. That scout is not this derivation. The lexicalization note (verb base vs verb theme) is not this derivation either. See Causative split and Lexicalization. Do not merge them with this page or with the typed-participant map.
Fernald, Denny, et al. 2021 give the same crying verb with an explicit segmentation and a third-person cell:
| Form | Their segmentation | Their gloss |
|---|---|---|
| yishcha | yi-sh-cha | I’m crying. |
| yinícha | yi-ni-cha | You’re crying. |
| yicha | yi-cha | S/he is crying. |
They also say that, usually, imperfective verbs have sh-, ni-, and Ø for first, second, and third person. The yi- in yicha is the third-singular cell of the Ø-imperfective paradigm in Young & Morgan’s base paradigms, as McDonough explains it — not a moved lexical subject, and not something this note will reanalyze beyond that.
1.2 Hale’s merge structure (the derivation this note draws)
Hale’s essay states the pronominal-argument hypothesis as he uses it: person–number morphology inside the verb word is the direct arguments. It is not agreement. There is no small pro beside it. An independent pronoun or noun, if present, is a contrastive or discourse adjunct, comparable to English left dislocationLine defect enabling plastic slip at realistic stresses; edge vs screw are the two classic geometries., not a second argument.
He identifies the traditional “classifier” with a voice head v, and mode (M) with the primary functional head above the verb, “corresponding to I(nfl) and T(ns).” Subject and mode are a portmanteau M/S. For a transitive, the object pronoun is the complement of V (its thematic position). The external argument is the specifier of v. For an unergative such as ‘cry’, the root does not force a specifier of V, so the pronominal argument is external to VP and is the specifier of v (his (23)).
That specifier of v is arg1 in this note: the external argument, merged VP-internally.
Hale’s essay then argues that, in a pronominal-argument language, those arguments are licensed in situ. He considers, and rejects, raising the object to Spec,M for Case. Surface order that does not match the merge tree is, on his proposal, phonological: the verb brings a bipartite skeleton (he cites Speas and McDonough), nuclear heads are spelled into that skeleton, and leftover material is prefixed. He writes that the voice head “is not ‘moved’ in the computational part of the derivation.” It remains in situ and is spelled out in the skeleton.
So the project’s question “where does arg1 → subject movement happen?” has a negative answer in the analysis this note draws. It does not happen. arg1 is merged in Spec,vP and is realized as the subject exponent inside M/S, still inside the verb word. The landing site of a subject movement is not a site Hale’s opened essay uses.
1.3 The flagged VPISH+EPP overlay (not the drawn analysis)
Grant’s assumption for this project is that the external argument is merged inside the verb phrase and moves to subject position. The only landing site that can be named from a label Hale actually uses is Spec,M. Reason: he himself says M corresponds to Infl/T, which is the familiar subject position (Spec,TP) under that correspondence, and his rejected earlier story had arguments raise to agree and to satisfy the EPP.
That step is written, in brackets, under examples 2 and 3, and marked NOT DRAWN. Adopting it would contradict the essay that supplies the merge sites. It is recorded so the map shows where the step would be, and that the literature opened here does not warrant it.
Either story keeps the theta bearer inside the clause and inside the verb word. Neither story turns arg1 into a separate lexical NP that then leaves the sentence.
2. Worked derivations
Notation: brackets are derivational, not a claim that the source printed a tree in this shape. “Spell-out” means the morphological/phonological realization Hale and McDonough describe, not a new operation.
2.1 Intransitive: yishcha ‘I’m crying’
Surface and gloss (Fernald, Denny, et al. 2021): yishcha, segmented by them as yi-sh-cha, ‘I’m crying.’ Intransitive. Young & Morgan practical orthography. McDonough, citing Young & Morgan 1987:g206, marks the same shape as intransitive with a null classifier: yishcha (ø) (vi) ‘cry, weep’.
Numeration, PA reading drawn here.
| Piece | Status |
|---|---|
| Stem cha | Lexical verb (V). Hale treats the root of ‘cry’ as not forcing a specifier of V (unergative). |
| Classifier | Null. Voice head v. Not an argument. |
| sh- | arg1: 1st-person external argument, a verb-internal pronominal, not a lexical NP. |
| yi- | Not a moved subject. In the Ø-imperfective 1st-singular cell, Young & Morgan’s paradigm (via McDonough) is yish-: mode and subject as one exponent. Fernald et al. write yi-sh- as a teaching segmentation of that cell. |
No internal argument is merged.
Derivation drawn (no arg1 movement).
- Merge the verbal stem: V = cha. No complement.
- Merge voice: [v′ v∅ V].
- VPISH merge of arg1. Merge the 1st-person pronominal as specifier of v:
[vP arg1:1sg [v′ v∅ [V cha]]]
This is the external argument inside the verb phrase. It has not moved. - Merge mode (Ø-imperfective) above vP:
[MP M∅ [vP arg1:1sg [v′ v∅ [V cha]]]] - Spell-out, not movement. Subject person is the exponent on mode (McDonough / Young & Morgan): M/S = yish-, classifier null, stem cha. Surface yishcha. arg1 never becomes a separate word.
Prose. The crier is arg1, merged in Spec,vP. The heard word is one verb. There is no lexical subject to raise, and the opened analyses do not raise the prefix out of the verb.
Flagged overlay, not drawn. If an EPP feature on M were imposed:
[MP arg1:1sg [M′ M∅ [vP t [v′ v∅ [V cha]]]]]
Landing site: Spec,M, because M is Hale’s Infl/T. Hale’s essay does not do this. The trace would still be inside the verb’s extended projection.
Third person of the same verb, same source: yicha (yi-cha) ‘S/he is crying.’ The subject exponent is null in their table (Ø for third person); yi- is the paradigm vowel, not an overt arg1 that has moved to a specifier you can hear.
2.2 Transitive, both arguments verb-internal: ni-sh-hozh ‘I tickle you’
Surface and gloss (Hale’s essay, (1)): ni-sh-hozh. His gloss: 2s-1s-tickle, ‘I tickle you.’ His fuller linear parse, (6): Obj ni, Mode Ø, Subj sh, classifier (the ł he says is phonologically deleted; footnote 1), stem ghozh. Orthography in the fetched text is ni-sh-hozh / ghozh without tone marks on this example. This note does not add tones the fetch did not show.
Numeration.
| Piece | Status |
|---|---|
| ghozh | Lexical V (stem). |
| ł (deleted on the surface, per Hale’s footnote) | Voice v, not an argument. |
| ni- | Internal argument: 2nd-person object pronoun. Verb-internal. Not a lexical NP. |
| sh- | arg1: 1st-person external argument. Verb-internal. |
| Mode Ø | Functional head M, portmanteau with the subject. Not an argument. |
Derivation drawn.
- Merge the internal argument as complement of V (Hale’s thematic position for this object):
[VP ni:2sg [V ghozh]] - Merge voice:
[v′ ł [VP ni:2sg [V ghozh]]]. - VPISH merge of arg1 as specifier of v:
[vP arg1:1sg [v′ ł [VP ni:2sg [V ghozh]]]] - Merge M:
[MP M∅ [vP arg1:1sg [v′ ł [VP ni:2sg [V ghozh]]]]] - No syntactic movement of arg1, and no syntactic movement of the object. Hale rejects object-raising to Spec,M. Spell-out puts the subject exponent with mode and realizes the object further left in the word: surface ni-sh-hozh. Both theta bearers are still inside the verb word.
Prose. ‘You’ is the complement of the stem. ‘I’ is arg1 in Spec,vP. The order you hear (object prefix before subject prefix) is not, on this essay, the result of moving arg1 to Spec,TP. It is spell-out into the verb’s skeleton. That is the map’s point for the simple transitive: the theta-role bearers do not leave the verb word to become lexical NPs.
Flagged overlay, not drawn. EPP movement of arg1 only (the external argument, not the object):
[MP arg1:1sg [M′ M∅ [vP t [v′ ł [VP ni:2sg [V ghozh]]]]]]
Landing site Spec,M. The object stays the complement of V. Hale considers a different movement — the object to Spec,M for Case — and rejects it as contrary to the pronominal-argument idea that arguments are licensed in place. This note does not draw that object movement either.
2.3 Independent noun plus verb-internal arguments: ’Awéé’ bi-’nii-sh-hóósh
Surface and gloss (Hale’s essay, (4), early pages, diacritics largely intact): ’Awéé’ bi-’nii-sh-hóósh. Gloss he prints: baby, 3-INCH-1s-tickle, ‘I start to tickle the baby.’ He calls -’nii- inchoative. He says the true arguments are bi- (third-person object) and sh- (first-person subject). ’Awéé’ is an adjunct, like English the baby in “The baby, I will start to tickle him/her.”
Numeration.
| Piece | Status under the PA derivation drawn here |
|---|---|
| ’Awéé’ ‘baby’ | Independent nominal. Not a theta-role bearer. Adjunct, coindexed with bi-. Not in the numeration as an argument. |
| bi- | Object argument (verb-internal). |
| ’nii- | Inchoative morphology (Hale’s INCH). Not an argument. This note does not resegment it. |
| sh- | arg1, 1st person, verb-internal. |
| Stem -hóósh | V. Hale does not further segment the stem in this example. |
Derivation drawn (PA). The argumental derivation is the same shape as §2.2, with object bi- and arg1 sh-, plus whatever inchoative heads Hale bundles as INCH (not reanalyzed here):
[MP M [vP arg1:1sg [v′ v [ … bi:3 … V:hóósh ]]]]
The independent noun is not merged as the complement or as arg1. It is an adjunct to the verb-sentence:
[adjunct ’Awéé’ [VS bi-’nii-sh-hóósh]]
No movement builds that adjunct from a theta position. On this analysis there is nothing for it to move from. arg1 is still sh-, inside the verb.
What is not drawn. The agreement analysis Fernald 2000 reports (Perkins, Platero, Speas): ’Awéé’ would be the object argument, and bi- would be agreement, possibly forming a chain with that nominal. A VPISH derivation of that analysis would merge a lexical object and might move a lexical subject. This note does not draw it. The essay that was opened draws the adjunct analysis, and Fernald’s scan does not supply a tree for the other side that survived OCR. Competing analyses are both live. The shared descriptive core, already in the prior morphology note, is that the verb word is well-formed without the noun.
2.4 Postpositional argument, still inside the verbal complex
A ditransitive string with two independent NPs was not recovered from a clean source. What was opened is smaller and real.
Fernald, Chavez, et al. 2019, (1): Dibé nee hólǫ́. Gloss: sheep, 2-with, exist.NI. ‘You have sheep.’ The second person is on the postposition nee (‘2-with’), not on a free pronoun. Dibé is an independent noun (the sheep). The existential verb is hólǫ́. This note does not resegment hólǫ́ or nee past their gloss.
Under the PA reading used in §2.2–2.3, the postpositional object prefix is the argument of P, and an independent noun linked to it is an adjunct. Hale’s essay says that pattern explicitly for postpositions, but the dog/water examples in that PDF were diacritic-corrupt and are not quoted. The sheep sentence is quoted from the questions grammar instead. This note does not claim Hale glossed Dibé nee hólǫ́.
Drawn, only as far as the gloss supports:
[PP 2 [P with]] is the verb-internal (here, postposition-internal) participant.
Dibé is an independent nominal beside that PP.
There is no external-argument prefix on hólǫ́ in the gloss they printed (exist.NI). This note does not invent a null arg1 movement for the existential. If a third-person subject exponent is null, that is the usual imperfective/neuter third-person zero Fernald, Denny, et al. 2021 state for verbs generally — an inference, flagged as such, not a gloss of this existential.
McDonough, quoting a Young & Morgan entry: bi’dishnééh ‘to give it to him (a disease), to infect him with it.’ Her parse of the prefix group bi’di-: bi- indirect object, ’(a)- third-person indefinite object, di- ‘extension’, plus ish as Ø-imperfective 1st singular, verb base ł-nééh. Three indexed participants (indirect object, indefinite object, subject) sit in the prefix complex. arg1 is the 1st-person exponent ish, again inside the core verb, not a lexical causer NP. No movement is shown by the dictionary entry. This is the sourced “more than one internal participant” case. It is not a yi-/bi- pair.
2.5 The yi-/bi- inverse — not given a tree
The complicated case the project hoped to derive in full is the direct/inverse alternation. Hale’s essay refers to it (obviative third person when both arguments are third person and the object is not the topic; linkage to overt NPs “in accordance with Willie and Jelinek 2000”). The example sentences in the fetched PDF were not legible. Fernald 2000 discusses yi- vs bi- verbs and cites Willie, but the Navajo lines in that scan are OCR-corrupt. Willie & Jelinek 2000 itself did not load.
Refused tree. No surface sentence, no gloss line, and no bracketed inverse derivation are given here. Prior project notes mention the alternation; they are not a substitute for a form this note re-opened. What can be said without a new example: on the PA side, yi- and bi- would be verb-internal arguments (or voice marking over them), still inside the verb word; on the agreement side they would index nominals that might be arguments. Choosing between those for a specific sentence requires the sentence.
Pointer (2026-10-04 evening): legible yi-/bi- examples are now opened in Inverse ranking (Willie & Jelinek 2000; Hale “Conflicting Truths”). Still no tree here.
3. Polar questions: a high marker, no argument movement
Source: Fernald, Chavez, et al. 2019, yes/no section. Orthography as on that page.
Declarative they start from, (1): Dibé nee hólǫ́. sheep, 2-with, exist.NI. ‘You have sheep.’
With the particle, (2): Da’ dibé nee hólǫ́? Q, sheep, 2-with, exist.NI. ‘Do you have sheep?’
They star da’ in every non-initial position (after dibé, after nee, and at the end). So da’ is a clause-initial question particle. This note calls it a high interrogative marker. It does not assign it the label C from a paper that was not opened. The distribution is what a high head (C, or a particle in Spec or adjunction to the clause) would have; the category label beyond “clause-initial particle” is not in the opened grammar.
With the enclitic, (4): Dibéésh nee hólǫ́? sheep-Q, 2-with, exist.NI. ‘Do you have sheep?’ They note the vowel of -ísh assimilates to the vowel of dibé. -ísh occurs once. It cannot attach to da’ (*Da’ísh …). It can focus a later constituent when it is not on the first phrase; their longer rope/sheep examples show focus by placement, not by extracting an argument out of the clause.
Both, (12): Da’ shééhonísínísh? Q, 1-2-know.NI-Q. ‘Do you know me?’ The gloss says the verb is neuter imperfective, object 1st person, subject 2nd person. This note does not resegment shééhonísin beyond that gloss.
Derivation for (2), and the negative result.
Start from the declarative structure of §2.4. Question formation, on this evidence:
[da’ [dibé [nee [hólǫ́]]]]
- da’ is merged at the left edge of the clause.
- Dibé stays where it was.
- nee (the 2nd-person postpositional argument) stays where it was.
- The verb stays where it was.
- No argument moves. arg1, if identified with the second person on nee or with a null subject of hólǫ́, is not extracted. The question is a particle (and, in other examples, an enclitic) over a clause whose theta-bearing morphology has not gone anywhere.
For (12), same negative result. The 1st-person object and 2nd-person subject are inside shééhonísin. -ísh is an enclitic on that verb word. da’ is initial. Nothing in the example shows those person markers leaving the verb.
Tag ya’ (their (14)–(15)) is final and is not analyzed here beyond noting that it, too, does not extract an argument in the examples.
4. Content questions: an independent h-word; the verb-internal material stays
Source: Fernald, Chavez, et al. 2019, content-question section. They call these h-words (Navajo counterparts of who/what/where, mostly in h-; díkwíí ‘how many’ is also a content-question word). Question particles -sh, lá, and -shą’ are glossed Q. They say -sh and lá focus a phrase and -shą’ marks a topic, and that these particles often sit right after the h-word but can sit elsewhere. They explicitly distinguish content-question -sh from yes/no -ísh.
Example (1), the one used for the derivation:
Mary ha’át’íí nayiisnii’?
Mary, what, 3-3-buy.P.
‘What did Mary buy?’
ha’át’íí ‘what’ is an independent nominal, not a prefix. It sits between Mary and the verb — the position of an object nominal in an SOV clause — not at the left edge, past Mary. The verb is glossed 3-3-buy, perfective: third person for two arguments, on the verb. This note does not resegment nayiisnii’.
Derivation, only as far as the example warrants.
Distinguish two kinds of participant:
- Verb-internal 3–3 morphology on nayiisnii’. Under the PA drawing, those are the theta bearers (subject and object pronouns inside the verb). Under the agreement drawing, they are agreement. Either way they are inside the verb word in the surface string.
- ha’át’íí, an independent DP. It is not the verb-internal theta bearer. It occupies the object’s nominal slot in this sentence. Mary is an independent nominal in the subject’s nominal slot.
No step in this example moves a prefix out of the verb. No step moves ha’át’íí to the front of the clause. The h-phrase is in the object position, which is the in-situ pattern for this sentence. This note does not add a covert movement to Spec,CP. Nothing opened here argues that covert movement, and Hale’s essay is about argument morphology, not about wh.
arg1, if it is the third-person subject exponent inside nayiisnii’, stays in the verb. Mary is a separate nominal. On the PA drawing she is an adjunct linked to that exponent, not arg1 itself, and she has not moved. On the agreement drawing she might be the subject argument; this note still would not move her, because the sentence does not show movement.
Other opened examples, not given false movement stories.
- Háágóósh díníyá? where-to-Q, 2-go.I. ‘Where are you going?’ The h-word is clause-initial, with -sh. The second person is inside díníyá (their gloss: 2-go.I) and stays there. Whether háágóó is fronted or is simply the position of a directional was not checked against a declarative counterpart. Gap: no movement arrow for this sentence.
- Díkwíí lá ninááhai? how.many, Q, 2-winter. ‘How old are you?’ díkwíí is initial; -lá follows it; the second person is on the verb. Same gap.
- Nléí ’asdzą́ą́ sáníshą’ háí ’át’į́? their (5): a demonstrative phrase with topic -shą’, then háí ‘who’, then a neuter verb. The h-word is not a prefix. Topic morphology is an enclitic on the nominal phrase. Not analyzed as extraction.
What the literature is often said to claim, and what was not verified. Secondary descriptions (not opened as the book) attribute to Schauber 1979 both in-situ embedded questions and some unbounded leftward movement of h-words, including in direct-discourse complements, and they say Navajo is not a language with obligatory English-style movement to C. Chapter 17 of the NLA grammar, whose search snippet says an h-word can be in its base position as object of ‘buy’ and can sometimes be displaced leftward, returned HTTP 504. This note does not adopt those claims as findings. The finding from pages that did open is the negative one the map needs: the verb-internal person marking does not leave the verb, and the clearest wh-nominal stays in object position rather than extracting.
If a later reading of Schauber shows optional leftward displacement of an independent h-phrase, that would be movement of a DP inside the clause, still not movement of the prefixal theta bearer out of the verb word, and still not movement out of the sentence. That derivation is not drawn.
5. What the linguistic map is, in one place
| Construction | What moves, on the evidence opened | Where arg1 is |
|---|---|---|
| Intransitive yishcha | Nothing syntactic. arg1 is merged in Spec,vP and spelled out in the mode–subject cell. | Inside the verb word. |
| Transitive ni-sh-hozh | Nothing syntactic, in the derivation drawn. Object raising to Spec,M is considered by Hale and rejected. | Spec,vP at merge; surface exponent inside M/S. Still inside the verb word. |
| Noun + verb (’Awéé’ …) | The noun is an adjunct on the PA drawing, not a moved argument. | sh- inside the verb. The noun is not arg1. |
| Polar da’ / -ísh | No argument movement. A high particle and/or an enclitic. | Unmoved. |
| Content question Mary ha’át’íí nayiisnii’ | The h-word is an independent DP in object position. Prefixal 3–3 stays on the verb. No extraction is shown. | Inside the verb, if identified with the subject exponent. Mary is a separate nominal. |
Spec,TP by name. This note does not use Spec,TP as the landing site of a movement it draws, because the opened Navajo analysis uses Mode, not T, as the head above vP, and does not move arg1 there. The flagged overlay’s landing site is Spec,M, Hale’s correspondent of Spec,TP. Reason for that name, if the overlay were ever adopted: EPP/subject position of the extended projection. Reason it is not adopted: the essay that defines M says the movement is unmotivated.
Agree. The PA derivation does not have a separate Agree step: the prefix is the argument. The agreement derivation would have Agree between a lexical nominal and a prefix. Fernald 2000 records that split and does not decide it. This note does not invent a feature inventory for either side.
6. Chemistry half — rearrangement inside one host
⚠ SPECULATIVE pedagogy. Not a joint field. Not prior art. Not a claim that Navajo syntax and solid-state diffusion are the same theory. Definitions below are one sentence each, from the pages opened. No activation energies, no concentrations, no isotherm equations. Bridge A (how many / whether saturated) is a different strand and is not used.
Mechanisms, defined only far enough to see the parallel.
- Interstitial diffusion (DoITPoMS): a small atom sits in a gap between lattice sites and moves to a neighboring gap, still inside the same crystal. It does not need a vacancyPoint defect: a missing atom site in the lattice., and it does not leave the solid.
- Vacancy-mediated substitutional diffusion (DoITPoMS): an atom that occupies a normal lattice site moves into an adjacent vacancy; the vacancy appears on the site the atom left. Direct place-swapping is not the practical mechanism. The atom remains in the solid.
- Segregation to a surface or grain boundaryInterface between crystals in a polycrystal; barriers in Hall–Petch strengthening. (Wikipedia, “Segregation (materials science)”): enrichment of a soluteIntentional minority species dissolved in the host (alloying addition). at a surface, grain boundary, or other defect, by partitioning out of solid solutionSolute atoms dissolved into the host lattice (substitutional or interstitial) without a separate precipitate phase.. The solute is still in the material. This is not phase separation into a macroscopic second region, and it is not evaporation. The same Wikipedia page notes that at high temperature, evaporation from a surface can occur and then the McLean description no longer holds. Evaporation is the case that leaves. It is the wrong parallel for the Navajo facts below.
Corpus reminders, not new measurements: prior notes already treat hydrogen in titanium as an interstitial and carbon in iron as an interstitial alloying addition (Same-element H; Fe–C). Those files are the project’s exhibit of a species on a site inside a host. This note does not add numbers to them.
Cautious map.
-
A verb-internal theta prefix that never becomes a separate lexical NP is like a species that occupies a site inside the host (an interstitial gap, or a substitutional site) rather than forming a separate phase. The Navajo verb word is the “host” in the analogy only. The prefix is not an atom.
-
arg1 → subject, in the derivation that was actually drawn, is not a hop: arg1 is merged in Spec,vP and realized in the mode–subject cell without a movement step. The speculative likeness is weak and should be said weakly: the participant changes how it is addressed (from thematic specifier to the audible subject exponent) while remaining inside one verb word. If one insists on the flagged EPP overlay (arg1 moving to Spec,M), the physical likeness would be a species moving from its original site to a privileged site in the same crystal — an interstitial hop, or a substitutional atom moving into a vacancy. Example that was looked up, not invented: interstitial diffusion as DoITPoMS describes it, of the kind the corpus already illustrates with H in Ti. No energy is stated. The overlay is not the linguistic analysis, so this sentence is a likeness for a step the syntax note refused to adopt.
-
Polar questions, which do not extract an argument, are like asking a yes/no about the system without removing a constituent. The mechanism-level likeness is site occupancy in place: a site is occupied or not, or a local arrangement has one state or the other, and the report does not take the species out of the solid. No laboratory technique is named. Resistivity, Hall effect, and similar measurements were not checked for this note and are not recruited.
-
The tempting wrong parallel for a wh-question is leaching, evaporation, or any path that removes an atom from the material. That parallel would fit English-style extraction of a constituent out of the clause. It does not fit the Navajo example that was opened. In Mary ha’át’íí nayiisnii’, the h-word stays in the object’s nominal position and the verb-internal 3–3 marking stays in the verb. The matching speculative picture, if any, is a rearrangement or a label inside the same solid — closer to an in-situ hop or to segregation that remains in the solid — and not a species leaving. If Schauber’s optional leftward h-movement is later confirmed, it would still be displacement inside the clause, which would still not be evaporation. Until that text is opened, the chemistry side should not be drawn more tightly than this mismatch.
-
Surface segregation that remains in the solid (Wikipedia: partitioning to a surface or grain boundary from solid solution) is available as a second “privileged site, still inside” picture, beside the interstitial hop. It is not a better match for polar questions, and it is not Bridge A. Equilibrium segregation is still one material system. A second phase, or evaporation, would break the analogy the same way a theta bearer leaving the clause would break the linguistic map.
Not claimed. Diffusion does not encode theta roles. A vacancy is not an argument. Mode is not temperature. The point of the half is only the geometry Grant asked for: a participant can change site, or be queried, without leaving the system it is in.
7. Sources actually consulted
URLs fetched or searched on 2026-10-02:
- Hale essay PDF: https://dingo.sbs.arizona.edu/~hharley/courses/StuttgartSummerSchool/HaleNavajoVerb.pdf
- McDonough working paper: https://www.sas.rochester.edu/cls/assets/pdf/working/fall-19/3mcdonoughsecondpaper.pdf
- Fernald et al. 2019, yes/no: http://talkingdictionary.swarthmore.edu/navajo_questions/?q-binary=
- Fernald et al. 2019, content questions: https://talkingdictionary.swarthmore.edu/navajo_questions/?q-content=
- Fernald et al. 2021, imperfective: https://talkingdictionary.swarthmore.edu/navajo_verb_mode/?m-imperfective=
- Fernald 2000 scan: https://works.swarthmore.edu/cgi/viewcontent.cgi?article=1238&context=fac-linguistics
- DoITPoMS diffusion mechanisms: https://www.doitpoms.ac.uk/tlplib/diffusion/diffusion_mechanism.php
- Wikipedia, Segregation (materials science): https://en.wikipedia.org/wiki/Segregation_(materials_science)
Cited inside those sources and not opened as texts: Young & Morgan 1987; Young 2000; Jelinek 1984; Willie 1991; Willie & Jelinek 2000; Speas 1984, 1990; McDonough 1990, 1996, 2000; Rice 2000; Perkins 1978; Platero 1978, 1982; Schauber 1975/1979; McLean 1957 (via the Wikipedia segregation page).
8. What could not be verified
- A clean yi-/bi- sentence with gloss. Several PDFs that should contain one were corrupt or timed out. No pair is invented.
- Schauber 1979 on whether, and when, h-words move.
- NLA Chapter 17’s base-position vs left-displacement examples (HTTP 504).
- Speas 1990’s own trees (catalog fetch failed). Only Fernald 2000’s one-sentence report of that side is used.
- A formal published claim that da’ is C⁰. Distribution is clause-initial; the category label is not upgraded.
- Whether háágóó in Háágóósh díníyá? is fronted or base-generated.
- Full diacritics on Hale’s later examples (dog into water; roll the ball; full template word). Unquoted.
- Any chemical number (barrier, concentration, enrichment ratio).
- That interstitial diffusion of H in a named metal has a measured barrier in a paper opened today. The corpus note is the project’s prior statement; DoITPoMS is the mechanism definition.
9. Open questions
- Open Willie & Jelinek 2000 or Young & Morgan 1987 on paper and copy one yi-/bi- pair with their gloss before any inverse derivation is drawn.
- Open Schauber 1979 far enough to say, with her examples, which question types move an h-phrase and which do not. Then decide whether any of those movements displace a verb-internal argument or only an independent DP.
- If a configurational analysis is drawn later, say whose (Speas 1990 vs another), and do not reuse Hale’s in-situ PA brackets for it.
- Keep this movement strand separate from Bridge A. Saturation counts and in-system movement answer different questions.
Draft 2026-10-02 PT. No joint field. Bridge A unrevised. Movement, not the typed-participant map.
Corpus note: navajo-derivational-movement-v0.md. Linguistics derivations are same-domain. The chemistry half is speculative pedagogy and is not Bridge A.