calliope-ts 0.0.4 → 0.1.0

package/dist/calliope/syntax.js

@@ -51,6 +51,205 @@ export function isInherentlyGiven(w) {
 export function isLightNominalHead(w) {
     return isPronoun(w) || isInherentlyGiven(w);
 }
+/** Focus-sensitive particles (Krifka §4.4.6, Association with Focus): a
+ *  premodifying "just / only / even / merely" marks its sister as the FOCUS —
+ *  the ASSOCIATE bears the accent, the particle is a functor and reduces
+ *  ("the carriage held but just OURSELVES": "just" = "only", so the accent
+ *  belongs to the pronoun it exclusivises).  Returns the associate, or null
+ *  when the word is not a premodifying focus particle.  A postmodifying or
+ *  clause-final "just/only" ("he only just made it") is temporal/degree use
+ *  and returns null. */
+const FOCUS_PARTICLES = new Set(['just', 'only', 'even', 'merely']);
+export function focusAssociateOf(w, words) {
+    if (!FOCUS_PARTICLES.has(bare(w)))
+        return null;
+    if (!/^RB/.test(w.lexicalClass))
+        return null;
+    const gov = w.dependency?.governor;
+    // canonical case: the particle is an ADVMOD premodifier of its associate
+    if (gov && gov !== w && !isPunct(gov) && gov.absoluteIndex > w.absoluteIndex)
+        return gov;
+    // parse-robust fallback: the particle immediately precedes a pronoun the
+    // parser attached elsewhere ("but just Ourselves" with "just" hung on the verb)
+    const next = words.find(x => x.absoluteIndex === w.absoluteIndex + 1 && x.syllables.length > 0);
+    if (next && (PRON.test(next.lexicalClass) || feat(next, 'PronType') === 'Prs'))
+        return next;
+    return null;
+}
+/** Marginal-modal use of dare/need (Quirk et al.): the BARE-infinitive
+ *  complement construction — "dare he aspire", "What the hand dare seize" —
+ *  is the AUXILIARY usage (the one that licenses subject inversion), so the
+ *  verb reduces like a modal and the beat belongs to its infinitive.  The
+ *  LEXICAL usage takes a to-infinitive ("dared to aspire") or a nominal
+ *  object ("dared the leap") and keeps full content stress — the gate is the
+ *  complement's shape, not the lemma alone. */
+const MARGINAL_MODALS = new Set(['dare', 'dares', 'dared', 'durst', 'need']);
+export function isMarginalModalUse(w, words) {
+    if (!MARGINAL_MODALS.has(bare(w)) || !/^VB/.test(w.lexicalClass))
+        return false;
+    // A clausal-complement child that is a bare verb (no "to" mark below it).
+    const comp = words.find(d => d !== w && d.dependency?.governor === w &&
+        /^(XCOMP|CCOMP)$/.test(d.canonicalRel ?? '') && /^VB/.test(d.lexicalClass));
+    if (comp) {
+        const hasTo = words.some(m => m !== comp && m.dependency?.governor === comp &&
+            (m.lexicalClass === 'TO' || bare(m) === 'to'));
+        return !hasTo;
+    }
+    // Surface fallback for mis-parses: an immediately-following bare infinitive
+    // ("dare seize"), or inverted pronoun subject + bare verb ("dare he aspire").
+    const at = (k) => words.find(x => x.absoluteIndex === w.absoluteIndex + k);
+    const n1 = at(1);
+    if (n1 && n1.lexicalClass === 'VB')
+        return true;
+    const n2 = at(2);
+    if (n1 && n1.lexicalClass === 'PRP' && n2 && /^VBP?$/.test(n2.lexicalClass))
+        return true;
+    return false;
+}
+/** A CAPITALIZED personal/possessive pronoun in mid-sentence ("but just
+ *  Ourselves", "For His Civility" — Dickinson's reverential capitals).
+ *  English capitalizes pronouns only sentence-initially and in "I", so a
+ *  capital here is the poet's deliberate typographic FOCUS mark: the pronoun
+ *  is referential/contrastive — Wagner's "functor role" ceded to a full
+ *  referring use, Krifka (78): an accented pronoun signals narrow focus. */
+export function isCapitalizedFocalPronoun(w, words) {
+    if (!PRON.test(w.lexicalClass) && feat(w, 'PronType') !== 'Prs')
+        return false;
+    if (!/^[A-Z]/.test(w.word))
+        return false;
+    const lower = bare(w);
+    if (lower === 'i' || lower.startsWith('i')) {
+        // "I" and its contractions are conventionally capital — no signal there.
+        if (lower === 'i' || /^i(ll|m|d|ve)$/.test(lower))
+            return false;
+    }
+    // Conventional capitalization is NOT a focus mark: a pronoun opening a
+    // QUOTED sentence ("Who said, "It is just as I feared!"") or following
+    // sentence-internal terminal punctuation or a colon is capitalized by
+    // orthographic rule, not by the poet's emphasis.  Scan back over the
+    // punctuation tokens abutting the pronoun; a quote mark or sentence
+    // punctuation among them disqualifies.  (Dashes and commas do NOT
+    // disqualify — Dickinson's mid-clause capitals stand.)
+    for (const x of [...words]
+        .filter(p => p.absoluteIndex < w.absoluteIndex)
+        .sort((a, b) => b.absoluteIndex - a.absoluteIndex)) {
+        if (/[A-Za-z]/.test(x.word))
+            break; // reached a real word
+        if (/["“”'‘’.!?:]/.test(x.word))
+            return false; // conventional capital
+    }
+    // Mid-sentence only: some earlier alphabetic word must exist.
+    return words.some(x => x.absoluteIndex < w.absoluteIndex && /[A-Za-z]/.test(x.word));
+}
+/** A POSTPOSED preposition: orphaned from its complement, which stands
+ *  immediately to its LEFT as a wh-word — the inverted pied-piping of "what
+ *  FOR is this form" (= "for what"), and the bare "Who WITH?", "Where FROM?".
+ *  Such a preposition cannot procliticize (no rightward complement to lean on)
+ *  and takes its STRONG form (Selkirk 1996: function words are weak only where
+ *  they can cliticize onto a following host) — the same phonology as clause-
+ *  final stranding, caught here in the wh-adjacent inversion the transitivity
+ *  test misses.  Subordinators never match ("what if…" is a mark, not case),
+ *  and neither does a preposition with a genuine rightward complement
+ *  ("who for the WORLD would…", "when in ROME…" stay reduced). */
+export function isPostposedPreposition(w, words) {
+    const pos = w.lexicalClass;
+    const rel = w.canonicalRel ?? '';
+    if (pos !== 'IN' && pos !== 'TO' && rel !== 'CASE')
+        return false;
+    if (rel === 'COMPMARK' || rel === 'ADVMARK')
+        return false;
+    const prev = words
+        .filter(x => x.syllables.length > 0 && x.absoluteIndex < w.absoluteIndex)
+        .sort((a, b) => b.absoluteIndex - a.absoluteIndex)[0];
+    if (!prev || prev.absoluteIndex !== w.absoluteIndex - 1)
+        return false;
+    if (!/^(WDT|WP|WP\$|WRB)$/.test(prev.lexicalClass))
+        return false;
+    // No genuine rightward complement: neither a following nominal governor (UD
+    // attaches the prep as CASE dependent of the noun it precedes) nor a
+    // following nominal/pronominal child.
+    const gov = w.dependency?.governor;
+    if (gov && !isPunct(gov) && (NOUN.test(gov.lexicalClass) || PRON.test(gov.lexicalClass)) &&
+        gov.absoluteIndex > w.absoluteIndex)
+        return false;
+    if (words.some(d => d.dependency?.governor === w &&
+        (NOUN.test(d.lexicalClass) || PRON.test(d.lexicalClass)) &&
+        d.absoluteIndex > w.absoluteIndex))
+        return false;
+    return true;
+}
+/** Subject–auxiliary INVERSION without interrogative force: the aux/modal
+ *  precedes its clause's subject in a sentence that is NOT a question —
+ *  optative "So, may then each moment drip off…" / "May she rest", negative
+ *  inversion "Never did I see", conditional inversion "Had I known".  The
+ *  inverted auxiliary is the overt exponent of the operator that triggered the
+ *  inversion (wish / negation / condition — Krifka: illocutionary and polarity
+ *  operators are accentable focus exponents), so it is NOT the reduced given-
+ *  auxiliary of a plain declarative: it anchors a light beat ("HAD I but
+ *  known", "never DID I see").  In a QUESTION the inversion is discharged by
+ *  the interrogative operator instead (wh-word or final "?") and the aux stays
+ *  reduced ("won't you GUIDE me…?", "do I have to BLESS?"). */
+export function isInvertedOperatorAux(w, words) {
+    const rel = w.canonicalRel ?? '';
+    if (rel !== 'AUX' && rel !== 'AUXPASS' && w.lexicalClass !== 'MD')
+        return false;
+    if (!/^(MD|VB)/.test(w.lexicalClass))
+        return false;
+    const head = w.dependency?.governor;
+    if (!head || head === w || isPunct(head))
+        return false;
+    const subj = words.find(d => d.dependency?.governor === head &&
+        /^(NSUBJ|NSUBJPASS)$/.test(d.canonicalRel ?? ''));
+    if (!subj || subj.absoluteIndex < w.absoluteIndex)
+        return false; // normal order / imperative
+    if (words.some(x => x.word.includes('?')))
+        return false; // overt question
+    const firstAlpha = words
+        .filter(x => /[A-Za-z]/.test(x.word))
+        .sort((a, b) => a.absoluteIndex - b.absoluteIndex)[0];
+    // A wh-fronted clause is interrogative even when verse carries the "?" on a
+    // later line.
+    if (firstAlpha && /^(WDT|WP|WP\$|WRB)$/.test(firstAlpha.lexicalClass))
+        return false;
+    return true;
+}
+/** Semi-modal "have to" (necessity): have/has/had governing a TO-infinitive
+ *  open complement with no intervening object — "do I have to bless", "we had
+ *  to laugh".  This is the reduced ("hafta") functor use: under broad focus
+ *  the whole chain subordinates and the beat belongs to the infinitive (or
+ *  falls to the subject by alternation).  The LEXICAL uses keep full content
+ *  stress and never match: possession with a purpose infinitive ("I have
+ *  BOOKS to read" — the object intervenes, the infinitive is ACL on the
+ *  noun), causative ("have him CALL me" — bare infinitive), existential /
+ *  plain possession ("what do I HAVE here" — no infinitive at all).  A
+ *  narrow/verum focus on the necessity itself ("Who do I HAVE to bless?") is
+ *  a marked reading that needs a cue the text must supply — an all-caps HAVE
+ *  is honored as that typographic cue. */
+const HAVE_FORMS = new Set(['have', 'has', 'had', 'hath', 'having']);
+export function isSemiModalHaveTo(w, words) {
+    if (!HAVE_FORMS.has(bare(w)) || !/^VB/.test(w.lexicalClass))
+        return false;
+    if (/^[A-Z]{2,}$/.test(w.word))
+        return false; // typographic narrow focus
+    const kids = words.filter(d => d !== w && d.dependency?.governor === w);
+    if (kids.some(d => /^(DOBJ|OBJ|IOBJ)$/.test(d.canonicalRel ?? '')))
+        return false;
+    const comp = kids.find(d => /^(XCOMP|CCOMP)$/.test(d.canonicalRel ?? '') &&
+        /^VB/.test(d.lexicalClass));
+    if (comp) {
+        const hasTo = words.some(m => m !== comp && m.dependency?.governor === comp &&
+            (m.lexicalClass === 'TO' || bare(m) === 'to'));
+        if (hasTo)
+            return true;
+    }
+    // Parse-robust surface fallback: have + "to" + verb, immediately adjacent.
+    const n1 = words.find(x => x.absoluteIndex === w.absoluteIndex + 1);
+    const n2 = words.find(x => x.absoluteIndex === w.absoluteIndex + 2);
+    if (n1 && (n1.lexicalClass === 'TO' || bare(n1) === 'to') &&
+        n2 && /^VB/.test(n2.lexicalClass))
+        return true;
+    return false;
+}
 /** A child relation that makes a verb BRANCH (carry an internal argument) — the
  *  specifier-restriction trigger (Wagner §6.3.2): a transitive/clausal VP whose
  *  branchingness blocks subordination of its subject. */