Structural Constraints on Morphological Inflection
Cross-linguistically, we often find constraints on the realization of inflectional morphology. In Russian, for example, verbs show gender agreement with their subject (feminine, masculine, or neuter) only in the past tense. This fact about Russian has a historical explanation: the past tense form was originally an adjective, and Russian adjectives agree in gender with the nouns they modify. In Modern Hebrew, verbs agree in gender with their subjects only in the present tense (and hearing a Hebrew-speaking 3-year-old girl using correct feminine forms for all her present-tense verbs is a truly breathtaking experience). Again, this fact has a historical explanation: as with Russian past tense forms, Hebrew present forms were originally participles. Gender agreement is not optional in Hebrew and Russian. Instead, its morphological realization is context-dependent.
It is surprisingly easy to find languages where verb inflection is obligatory in some contexts but impossible in others. All depend on syntactic context, rather than on tense, which is expressed as an inflectional part of the verb itself in Russian and Hebrew.
In the Kujamaat Jóola portion, we will see that subject agreement is expressed obligatorily except in the past subordinate and positive imperative forms of the verb. In some related languages, like Balanta, however, verbs agree with their subject only in certain syntactic contexts. In Balanta, verbs may be marked for subject agreement, but generally only in the absence of a subject noun phrase (1a). When a subject noun phrase is present, a subject prefix on the verb does not express agreement. Instead, it indicates that the subject is focused (1b) (data from N’Diaye-Corréard 1970: 30):
Another example of a structural constraint on morphological inflection comes from the Central Khoisan language //Ani and its system of object agreement (Vossen 1985).1 Finite verbs (except in the imperative) bear affixes that agree with a pronominal object in person, gender, and number (2a) or with a nominal object for number and gender (2b):
The catch is this: if a nominal object is not marked for gender and number, object agreement does not appear on the verb:
Here, in contrast to (2b), ‘elephant’ does not bear a gender–number suffix, and object agreement morphology fails to appear on the verb.
In Arabic, the basic generalization is that subject–verb number agreement appears on the verb when the word order is SV (subject–verb) (4a) but not when it is VS (4b) (data from Ouhalla 1994: 43):
In (4b) third plural subject agreement is blocked, and instead we get default third person singular agreement.
In sum, in Balanta and Arabic subject agreement and //Ani object agreement, the realization of agreement is either obligatorily present or obligatorily absent, depending on the syntactic context. If you have had extensive experience with syntax you might want to investigate in greater detail the structural analyses behind these facts.
1 // represents a lateral click, ! a palatal retroflex click, and ʔ an alveolar click.
