This involves assigning feature values to phrases or words, e.g. marking a Noun Phrase as singular, or a Verb Phrase as past tense. A feature-based syntax has been modelled by the Text Encoding Initiative (Sperberg-McQueen & Burnard (eds) 1994), in which syntactic information may be represented by means of attribute-value pairs. If necessary, this kind of information can be added in a compact way by adding various subscripts to syntactic segments. Table 7 is an example from the SUSANNE Corpus (showing only part of a sentence), with, in the right hand column, a singular (proper) noun phrase (Nns), and past tense verb phrase (Vd).
Similarly, the ENGCG annotation scheme gives subcategorisation information, specifying values such as PERS (personal pronoun), NOM (nominative) and SG1 (first person singular), and for verbs <SV> (intransitive), <SVO> (monotransitive) and <SVOO> (ditransitive) . Subclassification is also coded in TOSCA.