The intuition that NSs rely on the verb's morphology to recover the reference of the subject is very old. On the one hand, the verb in languages like Italian, Spanish or Swahili productively and distinctively encodes person and number, whereas in languages like French or English it does not. On the other hand, null subjects are possible in many languages only when they are recoverable from verbal information. As already mentioned, for example, NSs are possible in Irish in the synthetic form, which has person/number information (see McCloskey and Hale, 1984), but not in analytic form, with no person/number information.

Hebrew presents a different aspect of the same phenomenon. As I discussed in Section 2.3.3, this language only allows for NSs in contexts where person/number inflection is available (either on the verb or on some other functional category; see Shlonsky, 2009), which is typically in 1st/2nd person in the past and future tense, as well as with certain types of negation in the present.

A third type of example is illustrated by languages like Pashto (see Huang, 1984, 535). This is a split ergative language that follows a nominative–accusative pattern for present verbs. It has a rich agreement system, and the verb agrees with the subject in transitive and intransitive clauses, as illustrated in (1), where the verb razi ‘comes’ agrees with the person and number with the subject Jān and zә ‘I’ agrees with xwr-әm ‘eat’, respectively.