A syntactic category is a syntactic unit that theories of syntax assume. Word classes, largely corresponding to traditional parts of speech (e.g. noun, verb, preposition, etc.), are syntactic categories. In phrase structure grammars, the phrasal categories (e.g. noun phrase, verb phrase, prepositional phrase, etc.) are also syntactic categories. Dependency grammars, however, do not acknowledge phrasal categories (at least not in the traditional sense).
Word classes considered as syntactic categories may be called lexical categories, as distinct from phrasal categories. The terminology is somewhat inconsistent between the theoretical models of different linguists. However, many grammars also draw a distinction between lexical categories (which tend to consist of content words, or phrases headed by them) and functional categories (which tend to consist of function words or abstract functional elements, or phrases headed by them). The term lexical category therefore has two distinct meanings. Moreover, syntactic categories should not be confused with grammatical categories (also known as grammatical features), which are properties such as tense, gender, etc.
Defining criteria
At least three criteria are used in defining syntactic categories:
The type of meaning it expresses
The type of affixes it takes
The structure in which it occurs
For instance, many nouns in English denote concrete entities, they are pluralized with the suffix -s, and they occur as subjects and objects in clauses. Many verbs denote actions or states, they are conjugated with agreement suffixes (e.g. -s of the third person singular in English), and in English they tend to show up in medial positions of the clauses in which they appear.
The third criterion is also known as distribution. The distribution of a given syntactic unit determines the syntactic category to which it belongs. The distributional behavior of syntactic units is identified by substitution. Like syntactic units can be substituted for each other.
Additionally, there are also informal criteria one can use in order to determine syntactic categories. For example, one informal means of determining if an item is lexical, as opposed to functional, is to see if it is left behind in "telegraphic speech" (that is, the way a telegram would be written; e.g., Pants fire. Bring water, need help.)
Lexical categories vs. phrasal categories
The traditional parts of speech are lexical categories, in one meaning of that term. Traditional grammars tend to acknowledge approximately eight to twelve lexical categories, e.g.
Lexical categories
adjective (A), adposition (preposition, postposition, circumposition) (P), adverb (Adv), coordinate conjunction (C), determiner (D), interjection (I), noun (N), particle (Par), pronoun (Pr), subordinate conjunction (Sub), verb (V), etc.
The lexical categories that a given grammar assumes will likely vary from this list. Certainly numerous subcategories can be acknowledged. For instance, one can view pronouns as a subtype of noun, and verbs can be divided into finite verbs and non-finite verbs (e.g. gerund, infinitive, participle, etc.). The central lexical categories give rise to corresponding phrasal categories:
Phrasal categories
Adjective phrase (AP), adverb phrase (AdvP), adposition phrase (PP), noun phrase (NP), verb phrase (VP), etc.
In terms of phrase structure rules, phrasal categories can occur to the left of the arrow while lexical categories cannot, e.g. NP → D N. Traditionally, a phrasal category should consist of two or more words, although conventions vary in this area. X-bar theory, for instance, often sees individual words corresponding to phrasal categories. Phrasal categories are illustrated with the following trees:
The lexical and phrasal categories are identified according to the node labels, phrasal categories receiving the "P" designation.
Lexical categories only
Dependency grammars do not acknowledge phrasal categories in the way that phrase structure grammars do. What this means is that the interaction between lexical and phrasal categories disappears, the result being that only the lexical categories are acknowledged. The tree representations are simpler because the number of nodes and categories is reduced, e.g.
The distinction between lexical and phrasal categories is absent here. The number of nodes is reduced by removing all nodes marked with "P". Note, however, that phrases can still be acknowledged insofar as any subtree that contains two or more words will qualify as a phrase.