Morphosyntactic Features: Categories, Phi-Features, and Syntax
Morphosyntactic Features and Word Classes
Morphosyntactic features affect the semantics of words. Minimal sets: nouns can be singular or plural. Two ways of thinking of morphosyntactic features: car [-plural] vs. cars [+plural]. A feature is present or absent in a word = privative features. Or binary valued: car [number: singular] vs. cars [number: plural]. A feature has one of two values. Not all features have the same properties (semantic effect or morphology). Category features: allow us to classify words into traditional major word classes: N, V, P, etc.
- a. Nouns [+N, -V]
- b. Verbs [-N, +V]
- c. Adjectives [+N, +V]
- d. Prepositions [-N, -V]
Phi-Features
Phi-features are a set of features that often occur together in languages: number, person, gender, and case. English lacks the gender syntactic feature (only semantic). Pronouns: a combination of person and number features. Case features: accusative, nominative, genitive, dative. Verbal features: subject-verb agreement. Tense feature [past]. Morphologically: addition of an –ed affix (play – played). Vowel change (run – ran). Suppletion (am – was). Suppletion + affixation (go – went). Interpretable feature: plural, category features are also interpretable. Uninterpretable features are essential for syntactic objects to be formed.
Uninterpretable Features and Syntax
Take any transitive verb, for instance, ‘watch’. Our Knowledge of Language includes the knowledge that ‘watch’ needs an object. ‘Watch’ is specified with [V]. It can also be specified with a feature that indicates that it needs to combine with an object. This is done with uninterpretable features: watch [V, uN]. But this uN, uninterpretable feature, is a problem. It needs to be deleted once the verb combines with an object [N], because it is not part of its meaning, it is simply a requirement. This relates directly with the architecture of the linguistic system we are assuming. Why? To form a sentence:
- We select a set of lexical items from the Lexicon (where the relevant words, morphemes, features, etc. are stored).
- These are combined by means of Syntax (the set of rules).
- The outcome of the syntax (syntactic structure) has to be interpreted by two components: the Interfaces, the Conceptual-Intentional system (meaning) and the Articulatory-Perceptual system (sound, its physical realization).
The process which gets rid of [uN] is called (feature) checking. The uN feature is checked once the v combines with an object. Words are arranged into phrases. Phrases –> constituents. Constituency tests: transposition, substitution, ellipsis, coordination.
Syntactic Structures
Joints: nodes (syntactic label). Node labels: C, E. Nodes are connected by branches. The top node E is the root node. Nodes with no branches beneath (A, B, D) are terminal nodes. Nodes with branches beneath (C, E) are nonterminal nodes. Immediate domination: E over D and C, and domination over A and B. Predicate (element that specifies who takes part in a proposition), arguments (participants in proposition). Verbs are classified according to the number of arguments they require: one-place to four-place. Theta roles: arguments play different roles: agent, patient, theme, experiencer, goal, benefactive, source, instrument, locative. Each theta-role should be assigned to a constituent, but a constituent cannot be assigned to more than one theta role.