Navajo lexicalization & “phrasis” — themes, bases, phrase-level units
What’s claimed and how well supported
- Labels on this page
- SPECULATIVE ×11
- Strongest label used
- SPECULATIVE
- Main caveat (quoted)
- Chemistry / Zipf–molecule half is ◆ SPECULATIVE pedagogy only, or not established.
Counted automatically from the labels printed on this page. Counts are a reading aid, not a grade of the page.
Terms on this page (3), plain-language definitions
- Dopant
- Intentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase.
- 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.
Definitions quoted from the Chem for linguists primer glossary.
Date: 2026-10-03 PT
Status: Draft. Linguistics map first. Chemistry / Zipf–molecule half is SPECULATIVE pedagogy only, or not established.
Audience: syntax/semantics (less chemistry).
Does not: announce a joint field, rewrite Bridge A, invent Navajo forms/glosses/trees, invent Zipf exponents, or treat Grant’s dopantIntentional dilute impurity meant to tune electronic/optical/catalytic behavior without replacing the host as majority phase. guess as a finding.
Prior project notes (not primary literature): secondary index; secondary index; Derivational movement; secondary index. This note does not contradict sourced claims in those files.
Scout file: Causative split was absent when this pass began; a concurrent agent filed it the same evening (2026-10-03 PT). Causative remarks below were checked against that scout and do not contradict its sourced claims (McDonough boil pair / non-productivity; Gessner unaccusative-short vs unergative-long; thematic ł).
Abstract. What this note actually found: in Young & Morgan as reported by McDonough, the listed lexical unit of the Navajo verb is the verb base (classifier + stem), with a looser Athabaskan term verb theme covering non-compositional prefix + base combinations; productive morphology and paradigm conjugation sit around those listed units, not instead of them. The Athabaskan literature talks about lexicalized bases / themes, preverbs / incorporated postpositions, and idiomatic preverb–verb combinations — not English-style “phrasal verbs.” Hale: Navajo lacks verbal noun incorporation; Northern Athabaskan has it; Navajo does allow restricted noun-into-postposition incorporation. Williams et al. 2015 claim Zipf scaling is robust for mixed-length phrases, not only words. A power-law claim over chemistry molecules (as Grant means it) is not established in sources opened; biological metabolite / gene-expression power laws exist and are a different domain. Grant’s phrase↔molecule / word↔atom correspondence, his coinage “phrasisized”, and his dopant guess remain HIS SPECULATIVE hypothesis — recorded, not endorsed. Bridge A (concentration windows ↔ valencyArity: how many actants a predicate takes; formal vs semantic valency can diverge./saturation) is a different strand. The 2026-10-02 movement note is also different: arg1 merges in Spec,vP and does not move, on the drawn PA analysis.
- 0. Grant’s hypothesis (HIS, SPECULATIVE — not a finding)
- 1. Lexical units in the Navajo verb (Young & Morgan via McDonough; Hale)
- 2. Phrase-level units (literature’s terms beside Grant’s label)
- 3. Causative split — only as it bears on lexical vs derived
- 4. Zipf sources
- 5. SPECULATIVE correspondence (pedagogy only — after the linguistic map)
- 6. Sources opened (with URLs)
- 7. What this note refused
- 8. What to do when Grant’s Navajo texts arrive
SPECULATIVE Cross-domain section. Open after the materials.
0. Grant’s hypothesis (HIS, SPECULATIVE — not a finding)
⚠ HIS SPECULATIVE. Not a project finding. Not a joint field. Do not cite as prior art.
Grant’s working idea (paraphrased from his brief, labeled as his):
- Zipf’s law in language shows power laws at the level of phrases; in chemistry (on his view) power laws show up at the level of molecules.
- He believes there is a phrase ↔ molecule and word ↔ atom correspondence (pedagogical analogy after the linguistic map).
- Navajo is illuminating because verb arguments are theta roles, not full lexical items (same clarification as secondary index).
- There is still something like a phrasal verb: the verb and its phrase are lexicalized as a unit. His coinage for that is “phrasisized” (he knows he made it up). Use that word only as his label, in quotes, beside the literature’s own term.
- That unit is introduced into the language not as a normal phrase-level molecule but as phrase-level requirements for what a molecule should be. He guesses those requirements come from dopants, because the linguistic arguments are theta roles. He will bring actual Navajo texts later.
- Do not turn the dopant guess into a result.
Bridge A (concentration windows ↔ valency/saturation) is a different speculative strand — one sentence of pointer only: see Solubility windows; unrevised here.
Movement note (Derivational movement, 2026-10-02): different question — on the drawn PA analysis, arg1 merges in Spec,vP and does not move.
1. Lexical units in the Navajo verb (Young & Morgan via McDonough; Hale)
1.1 What is listed vs what is built
McDonough’s guide to Young & Morgan 1987 (opened PDF) is the main descriptive source for this section. Y&M themselves were not opened as the book; facts below are as reported by McDonough (and, where noted, by Hale citing Y&M).
| Term | What it is (sourced) | Listed or built? |
|---|---|---|
| Root | Abstract lexeme in CAPS across stem shapes (Appendix V) | Listed abstraction |
| Stem / stem set | Mode/aspect stem-shape alternations for a root | Listed paradigms of shapes |
| Verb base | Classifier + stem (final syllable of the core verb) | Listed; “these units are not productive; they need to be listed” (McDonough §3.3.1) |
| Verb theme | Stem + classifier + any prefixes that make up the meaning of a verb; addresses non-compositionality | Looser Athabaskan term; McDonough: Y&M do not use “theme” as a technical word-formation device |
| Core / minimal verb | Mode inflected for person/number (Base/Model Paradigm cell) + verb base | Built by combining a listed base with a paradigm cell |
| Full verb word | Core verb ± disjunct/conjunct prefixes that belong with that entry | Often non-compositional; dictionary entries are fully inflected words linked to conjugation classes |
McDonough’s organizing formula (her §3.3 / summary):
Model Paradigm (Mode.person.number) + Verb Base (VL.mode) → fully inflected verb word (Mode.Aspect)
Template is not the word-formation engine. McDonough: Y&M do not build words by concatenating the position-class template; paradigms and listed bases do. That matches the project’s earlier morphology notes: the template is a reference chart, not a free concatenator.
1.2 Example of a listed meaning unit (no invented forms)
McDonough’s example of a non-compositional unit that the looser “theme” talk is meant to capture (Y&M dictionary entry she cites):
- bi’dishnééh — glosses she reports: ‘to give it to him (a disease), to infect him with it’
- Verb base she reports: łnééh; prefix complex bi’di- treated as part of the meaning unit for that entry, not as free syntax glued on at use-time.
This is the descriptive pattern Grant cares about: something larger than a bare stem is listed, and person/mode still conjugate around it.
1.3 Hale on voice vs lexicalized classifier
Hale’s PA essay (opened PDF): the traditional “classifier” is a voice / transitivity marker in its productive uses; when it is “not merely a lexicalized, synchronically inert, component of a verb,” it participates in productive transitivity alternations (causative, passives, etc.). That split — productive voice vs listed inert classifier-on-stem — is exactly the lexical-vs-derived cut this note needs, without inventing new paradigms.
2. Phrase-level units (literature’s terms beside Grant’s label)
Grant’s coinage: “phrasisized.” Literature terms to put beside it:
| Literature term | What authors mean | Relation to Grant’s label |
|---|---|---|
| Verb base (Athabaskan descriptive) | Classifier + stem as listed unit | Smaller than a free English “phrase”; still a lexicalized chunk |
| Verb theme (Athabaskan descriptive) | Base + thematic / meaning-making prefixes; non-compositional | Closest ordinary term for “listed multi-morpheme meaning unit” |
| Preverb / incorporated postposition (disjunct; Rice; Alderete; Hale) | Adverbial–thematic / postpositional material left of the conjunct | Phrase-like material inside the verb word |
| Idiomatic / “truly lexical” preverb–verb combination (Hale citing Hargus & Tuttle p.c.) | Unique or idiomatic Pv+V pairings vs productive adverbial preverbs | Closest to English “phrasal verb / idiom” talk |
| Noun incorporation (Baker; Mithun; Rice; Hale) | Bare N in object-of-V or object-of-P position spelled inside the word | Verbal NI: not Navajo; Northern Athabaskan yes. Navajo: restricted N into postposition |
2.1 What the opened sources say about “phrasal” material
- Hale (opened): Navajo does not have noun incorporation into verbs. Northern Athabaskan cousins do. Navajo does allow incorporation of bare nominals into postpositions (restricted but morphologically transparent) — his water / taah pair.
- Hale on preverbs: Preverbs may attach high (subject in scope) or low (object/event-internal). Hargus & Tuttle (as Hale reports them) distinguish productive adverbial preverbs from “truly lexical” unique combinations and idioms. Surface leftward position of preverbs can hide that underlying low attachment — spell-out, not English-style particle shift as the analysis.
- McDonough / Y&M: Disjunct Position Ib = adverbial–thematic (“postpositional stems”); dictionary entries often list a prefix group with the base as one meaning unit (há- ‘for, after (to get)’ with łt’ááh, etc.).
- Alderete (ROA digest, secondary): Disjunct open-class-ish material includes preverbs (incorporated postpositions/adverbs) and, in languages that have them, incorporates; conjunct is the closed functional set.
Honest mapping for Grant’s word:
His “phrasisized” unit ≈ literature’s verb theme / lexicalized verb base / idiomatic preverb–verb combination, depending which grain he means when the texts arrive. It is not a stored English-style string of free words. In Navajo the “phrase” is usually inside one verb word.
2.2 Does the literature support his Navajo twist?
His twist: because theta-role prefixes are not lexical items, a “lexicalized phrase” in Navajo may be a template of role requirements rather than a stored string of words.
| Claim piece | Support in opened sources? |
|---|---|
| Arguments are often verb-internal person morphology, not free lexical NPs | Supported (Hale’s PA statement; Y&M/McDonough subject-on-mode; project morphology notes). Competing Speas/Perkins/Platero “agreement” reading remains live (Fernald 2000 as cited in the movement note) — still morphology-on-V either way. |
| Some multi-morpheme combinations are listed / non-compositional | Supported (McDonough on themes/bases; Hale on lexicalized classifiers and idiomatic preverbs). |
| Those listed units are “templates of role requirements” rather than stored word-strings | Partly supported as a fair reading of PA + thematic prefixes (slots that must be filled by role morphology). The literature does not use Grant’s wording and does not contrast “role template” vs “stored string” as a named debate. |
| Those requirements “come from dopants” | Not discussed. Chemistry is absent. Leave as HIS guess. |
Verdict on the twist: the linguistic half is compatible with PA / theme description and is not contradicted by the sources opened; the literature simply does not discuss the dopant / molecule framing. Do not force chemistry into the Athabaskan citations.
3. Causative split — only as it bears on lexical vs derived
See also the causative scout Causative split (v0.3, 2026-10-07: positional ł-causatives distinguish a causer who keeps a state from one who brings about a change, on one stem; v0.2, 2026-10-06: Goertz et al. 2006’s analytic causatives are reason clauses with no causer argument, so a causer enters the verb only through ł morphology in what is opened; v0.1, 2026-10-04: Hale & Platero 1996 primary names simple transitivity alternation vs complex causative; Japanese lexical vs -sase still not established). This section only borrows what bears on listed theme vs live derivation.
Productive / derivational side (reported):
- McDonough, from Y&M: yibéézh ‘it’s boiling’ (ø) vs yiłbéézh ‘he’s boiling it’ (ł) — the classic ø / ł transitivity alternation on root BÉÉZH (scout §3.1; same pair).
- Hale’s PA essay: classifier as voice v in productive uses; causatives and other transitivity alternations live there.
- Scout v0.1 (Hale & Platero 1996 primary; Gessner secondary): labile / unaccusative bases take the simple ł-only alternation; unergative complex causatives need more than ł (ł + y- + special object marking) — a family-internal length/complexity split, not the Japanese lexical/-sase cut.
Lexical / non-productive / thematic side (reported):
- McDonough’s three reasons ł is not a free productive transitivizer everywhere: (i) many ł transitives are not derived from an ø intransitive of the same root (her yíłchozh ‘I ate it (leafy matter)’ illustration); (ii) ł also appears in intransitives and neuters; (iii) classifier choice is not independent of stem sets — the verb base is the organizing unit.
- Hale’s matching hedge: classifier may be a “lexicalized, synchronically inert” piece of a verb.
- Scout (via Kibrik as reported): Young, Morgan & Midgette are said to list causatives as separate themes — which blurs derivational vs thematic listing in the lexicon.
Bearing on this note: the same morphology (ł) can be live valenceBonding capacity / oxidation-related electron count of an atom in context. derivation or listed theme material. That is the lexical-vs-derived split Grant needs when he asks whether a unit is “phrasisized” (listed) or built. It does not, by itself, decide the molecule analogy, and it is not established as the Japanese lexical vs productive -sase split (scout Answer).
4. Zipf sources
4.1 Language: phrases (opened)
Williams, Lessard, Desu, Clark, Bagrow, Danforth & Dodds (2015). “Zipf’s law holds for phrases, not words.” Scientific Reports 5:12209. DOI: 10.1038/srep12209. arXiv PDF opened: https://arxiv.org/pdf/1406.5181v2.pdf.
What they actually claim (no invented exponents beyond theirs):
- Word-frequency Zipf scaling is limited (~three to four orders of magnitude) before a break.
- Using random text partitioning (boundary-cut probability q; pure random q = ½), mixed-length phrases show Zipf-like rank–frequency scaling over as many as seven to nine orders of magnitude on large corpora (Wikipedia, NYT, Twitter, lyrics).
- For one-off partitions of A Tale of Two Cities, they report clause θ ≃ 0.78 and random-partition phrase θ ≃ 0.98 (their Fig. 1 / SI).
- Punchline they draw: the general semantic units for this kind of statistical analysis “can and must be phrases — not words.”
That is a real, opened source for Zipf at phrase level. It does not mention Navajo, Athabaskan, or chemistry.
4.2 Chemistry: power law over molecules?
Not established in sources opened for Grant’s chemistry-molecules claim.
What was opened in a neighboring domain:
- Sato, Horikawa, Kondo, Sato & Setou (2018). “A power law distribution of metabolite abundance levels in mice…” Scientific Reports. https://www.nature.com/articles/s41598-018-28667-5 — biological metabolites (MALDI IMS; lipids etc. in mouse tissue), not a general claim about chemical molecules as lexical analogues.
- Search hits also point to Furusawa & Kaneko (2003) Zipf in gene expression, and caveats that some omics abundance fits are lognormal / double PLN rather than pure Zipf (Lu & King 2009 — not fully opened here).
Recording rule: Grant may believe chemistry shows power laws at molecule grain. Until a chemistry (not metabolomics/transcriptomics) source is opened and checked, write “not established in sources opened” and keep his belief labeled HIS.
No Zipf exponent for molecules is invented or reported as a project number.
SPECULATIVE Cross-domain section. Open after the materials.
5. SPECULATIVE correspondence (pedagogy only — after the linguistic map)
⚠ SPECULATIVE pedagogy. Analogy only. Not a finding. Not a joint field. Bridge A untouched.
| Analogy slot | Linguistic side (sourced map above) | Chemistry side |
|---|---|---|
| Word ↔ atom | Word / stem / listed base as small lexical unit | Atom as small compositional unit — analogy only |
| Phrase ↔ molecule | Mixed-length phrase (Williams) or Navajo theme / idiomatic Pv+V as larger coherent unit | Molecule as bound multi-atom unit — analogy only; Zipf-over-molecules not established here |
| Navajo twist | Theta-role / person morphology saturates args; listed themes can be role-requirement packages more than stored free-word strings | Grant’s further guess: those requirements come from dopants — HIS, not sourced |
What the literature does:
- Supports recording Navajo argument structure as role morphology on V, and supports listed multi-morpheme meaning units (base / theme / idiomatic preverb).
- Does not discuss word↔atom or phrase↔molecule.
- Does not discuss dopants.
Do not force the chemistry. When Grant’s Navajo texts arrive, the job is to mark which units look like listed themes / idiomatic Pv+V / ordinary productive morphology — still linguistics-first.
6. Sources opened (with URLs)
| Source | URL / locator | Load status |
|---|---|---|
| McDonough, “How to Use Young and Morgan’s 1987 The Navajo Language” (Rochester CLS working paper) | https://www.sas.rochester.edu/cls/assets/pdf/working/fall-19/3mcdonoughsecondpaper.pdf | Opened cleanly |
| Hale, “On the Significance of Eloise Jelinek’s Pronominal Argument Hypothesis” (MIT essay PDF) | https://dingo.sbs.arizona.edu/~hharley/courses/StuttgartSummerSchool/HaleNavajoVerb.pdf | Opened cleanly (diacritics often lost in extract; forms quoted only where clear or as McDonough/Y&M practical orthography) |
| Williams et al. 2015, Sci. Rep. / arXiv PDF | https://arxiv.org/pdf/1406.5181v2.pdf; journal https://www.nature.com/articles/srep12209 | Opened cleanly (arXiv) |
| Sato et al. 2018 metabolite power law | https://www.nature.com/articles/s41598-018-28667-5 | Opened cleanly |
| Alderete ROA-557 (disjunct/preverb digest) | https://roa.rutgers.edu/files/557-1102/557-1102-ALDERETE-0-2.PDF | Search/digest use; not relied on for forms |
| BLS causative paper (Hale & Platero line) | https://journals.linguisticsociety.org/proceedings/index.php/BLS/article/download/1105/885 | Fetch timed out — not treated as read; causative facts taken from McDonough + Hale essay instead |
| Young & Morgan 1987 as the book | — | Not opened; cited via McDonough / Hale |
| Causative split | same-day concurrent scout | Absent at start of this pass; present by evening — causative § aligned to it |
7. What this note refused
- Inventing Navajo forms, glosses, trees, or conjugation tables beyond what McDonough/Hale printed cleanly.
- Inventing a Zipf exponent for molecules or for Navajo.
- Turning Grant’s dopant guess, phrase↔molecule belief, or “phrasisized” coinage into a finding or joint-field claim.
- Equating Bridge A with this strand, or reopening the 2026-10-02 movement analysis.
- Editing the website, pushing, or deploying.
- Quoting OCR-corrupt or diacritic-garbled strings as if they were clean data.
8. What to do when Grant’s Navajo texts arrive
- Tag each candidate unit with the literature’s term: verb base / verb theme / idiomatic preverb–verb / productive classifier alternation / ordinary inflection — not only with “phrasisized.”
- Ask whether the unit is a stored multi-morpheme meaning chunk or a live valence derivation (the ł split in §3).
- Ask whether arguments in the example are role morphology on V (PA-friendly) or lean on free lexical NPs — without forcing the Speas debate to a winner unless the text decides it.
- Keep chemistry / dopant / Zipf-molecule talk in a separate SPECULATIVE paragraph after the linguistic tags.
- If a form’s source page does not load cleanly, say so and do not reconstruct.
v0 — 2026-10-03 PT. Linguistics-first. Causative scout reconciled same evening. No joint field. Bridge A unrevised.
Corpus note: navajo-lexicalization-phrasis-v0.md. Listed bases and themes are same-domain linguistics. Grant’s phrase↔molecule line and the dopant guess are his and speculative. Not Bridge A. Not merged with movement or the typed-participant map.