Unification grammar uses attribute-value matrices (AVMs) to represent syntactic and semantic information about words and phrases, without requiring movement operations. AVMs model words as structured objects with attributes like PHON, SYNSEM, CAT, and VALENCE/SUBCAT. Unification grammar combines syntactic objects when their attributes are compatible, such as when a verb's SUBJ attribute matches the CONT and CAT features of an appropriate subject noun phrase. This achieves composition through structure-sharing rather than movement. Head-driven phrase structure grammar (HPSG) is a popular type of unification grammar.