Inflectional paradigms come in two possible flavours: either the forms line up with the featural system whose values they express, or they do not. For example, the formatives in Table 5.1 from Chamorro (an Oceanic language of Guam) can be readily characterised in terms of aspect-mood distinctions: ha- is the realis 3sg prefix, u- the irrealis prefix, and reduplication marks the incompletive. Contrast this with the three tense-aspect paradigms from Gulmancema, a Gur language of Burkina Faso. There is a suffix -di which is found with each verb, but it cannot readily be characterised in featural terms, because it can in fact be used for any feature: with a verb like ‘pass’ it is used for the imperfective, with a verb like ‘love’ it is used for the perfective, and with a verb like ‘hear’ it is used for the aorist and perfective. In this case a description of the possible forms in the paradigm (unsuffixed, or with suffix -di) is separate from the feature values behind it.

What I would like to look at here is a set of paradigms that falls in between these two extremes, where the distribution of forms appears to approximate the feature system they express. That is, there is neither complete correspondence (as in Chamorro) or complete divergence (as in Gulmancema). This is either a challenge to morphological theory, which cannot readily accommodate this, as far as I know, or it demonstrates that my interpretation of the facts is wrong, in which case this may be a useful exercise in clarifying what is or is not a possible morphological analysis. The example involves number marking on the verb in Hualapai, a Yuman language spoken in Arizona. Verbs distinguish four forms which, I will argue, can be understood as representing four degrees of number, ranging from low (singular) to high (greater plural). The morphological paradigm in turn follows an incremental principle where the addition of a morphological marker corresponds to an increase in the number value. However, the terms by which this morphological system is constructed do not correspond to the values of the morphosemantic system. In effect, the morphological system approximates the morphosemantic system without quite matching it. In §5.2 I present the structure of the verbal paradigm and argue for my particular interpretation of it in terms of morphosemantic number. In §5.3 I present the morphological paradigm and argue for an interpretation in terms of incremental quantity. In §5.4 I attempt to relate the two, arguing for what this might mean for morphological theory. All the data here is taken from the reference grammar by Watahomigie et al. (1982, revised version from 2001), or their dictionary (2003).