Syntactic Categories – Diagnostics, Cross-Linguistic Variation & Methodology
Review of Lecture Context
- Lecturer apologises for lateness & room confusion; microphone needed for recording.
- Recap of Week 1:
- Syntax = study of word/phrase arrangement AND the relations/dependencies among them.
- Two intertwined dimensions of grammar:
- Structural/positional rules.
- Functional/dependency relations.
- Word‐order variation shows language‐specific rules (e.g. English “the child chased the dog” vs Ungarranggu* “dog child chased”).
- Some languages encode “who does what” via morphology rather than position.
Core Theme of Today: Syntactic Categories (Word Classes)
- Fundamental building blocks of sentences = words → grouped by syntactic categories (aka word classes / lexical categories / parts of speech).
- Rationale:
- Different words show parallel behaviour in distribution & morphology.
- Treating them as similar aids description & theory construction.
Problems with Purely Semantic Definitions
- Traditional school grammar: “noun = person/place/thing”, “verb = action”, “adjective = describing word”, etc.
- Quickly fails:
- “happiness” is abstract yet behaves like a noun (possessable, can be subject).
- “assassination”, “schoolhouse” ≠ simple persons/places.
- Evidence from other languages:
- Irish “doctor” acts as a verb morphologically (takes tense & agreement).
- Walpiri* (Western Desert) word for “small” can function as sentence subject, showing noun-like case marking.
Distributional/Structural Approach
- Key Principle: Category is determined by a word’s distribution (internal + external properties), not meaning alone.
- Internal (morphological) diagnostics:
- Derivational affixes (change class) → e.g. “happy” + “-ness” ⇒ noun.
- Inflectional paradigms (tense, number, case, etc.).
- External (syntactic) diagnostics:
- Dependents it licenses (modifiers, arguments, determiners, relative clauses).
- Functions the phrase can fill (subject, object, adjunct, head of PP, etc.).
- Method illustrated with the nonce word “flug”:
- Check for familiar derivational endings.
- Test inflections: \text{flug} \to \text{flugs}? \text{flugged}?
- Frame in contexts: “the flug”, “quick flug”, “John flugs”, “quickly flugs”, etc.
- Humorous English ambiguity: “Time flies like an arrow; fruit flies like a banana.”
- Same orthography “flies”; category distinguished by distribution (verb vs noun) & morphology (plural –s vs past –ew).
Language-Specific Nature of Categories
- English nouns vs Ungarranggu nouns:
- English: number marking (-s), limited case (pronouns), determiners precede noun.
- Ungarranggu: no number marking, rich \approx 14 case suffixes (ergative, locative, associative, instrumental, …), demonstratives follow noun, possession possible.
- Shared behaviours (can be possessed, modified) justify same label “noun” but criteria differ.
- Syntactician’s task: discover & justify categories for each language dataset; cannot import English diagnostics wholesale.
Typological & Ontological Tendencies
- Cross-linguistic clustering around semantic ontology but NOT deterministic.
- Broad mapping: events → verbs (predicates); entities → nouns (arguments); properties → adjectives/adverbs (modifiers).
- Counter-examples common: English “funeral” (event noun), Ungarranggu “small” (property noun).
- Degree of category inventory:
- Some languages proposed with 1 category (e.g. Cayuga, arguable): every lexical item takes agreement/tense.
- 2-way split (noun vs verb) more common.
- 3-way (adds adjectives); 4-way (adds adverbs) typical of many familiar languages.
Open vs Closed Classes
- Open: potentially limitless membership, productive morphology (nouns, verbs, adjectives).
- Closed: small fixed sets, idiosyncratic behaviour (determiners, pronouns, adpositions, auxiliaries, particles).
Sub-Classification within Major Classes
- Example: English count vs mass nouns (“butter” lacks plural -s).
- Subclasses arise when only a subset bears certain diagnostics.
Cross-Linguistic Variation in Nouns
Proper vs Common Nouns
- English: determiners/plurals restricted with proper names (*“the Questacon”, *“several Questacons” unless semantically coerced).
- Cebuano (Philippines): distinct actor/patient particles for names:
- Actor: si (proper), ang (common).
- Patient: ni (proper), sa (common).
Noun Class (Gender) Systems
- Chichewa (Bantu): 18 noun classes (≈ genders); agreement prefixes on verbs/adjectives/pronouns vary by class & number.
- e.g. Class 7 corn “chimanga” ⇒ verb object marker “chi-”.
- Class 3 lion “mkango” ⇒ object marker “u-”.
- Distinction between class/gender (agreement-triggering property of noun) vs case (morphological marking of syntactic role).
Possession Typology
- Inalienable vs alienable possession (Matukar Panau, PNG):
- Inalienable (body parts, kin): suffix on noun → “hand-2SG” = “your hand”.
- Alienable (transferable objects): separate possessive particle → “your house”.
- English shows hints of dual strategy (pre-nominal ’s vs “of” phrase), but not grammatically obligatory like some languages.
Practical Methodology for Analysis & Assignments
- Collect maximal data: paradigms, sentence frames, acceptability judgements.
- Identify internal markers (affixes, stem alternations).
- Map external distribution: allowed dependents, syntactic roles.
- Seek clustering → propose categories and possible subclasses.
- Support with multiple converging diagnostics; avoid sole reliance on meaning.
- Remember: criteria & labels are language-specific; use cross-linguistic tendencies only as heuristics.
Take-Home Summary
- Categories are defined by behavior (morphology & syntax), not by inherent meaning.
- Each language carves up the lexical space uniquely, though broad semantic correlations recur universally.
- Syntacticians must justify category assignments empirically, triangulating diverse evidence.
- Understanding variation (case vs agreement, alienable possession, noun classes, etc.) sharpens one’s analytic toolkit.
- Next lecture will extend from words to phrases (building constituency and phrase structure).