Verb Raising and Affix Lowering. It is argued that verb raising is more economical and easier than affix-lowering (Xxxxxxx 1989). According to Ionin and Xxxxxx (2002), the reason why affixal agreement is more difficult than suppletive agreement is that L2 learners first associate S-V agreement with verb raising, as this operation is more economical for L2 learners. In other words, L2 learners do not initially analyze 3SG -s as a morphological reflex of agreement. Evidence for the early acquisition of suppletive agreement comes from a few observations made in Ionin and Xxxxxx (2002). First, L2 learners, L1 Russian children, demonstrate significantly higher proficiency in using suppletive agreement with the verbs of be than affixal agreement; they show great sensitivity to incorrect uses of agreement with the verbs of be, which suggests that the learners have mostly mastered the feature specification of be forms. In other words, categories and features governing agreement under I(nfl) are intact in L2 grammar. Another potential evidence for L2 learners taking verb raising as an initial step for acquiring agreement is provided by overgeneralization of be in non-progressive contexts (e.g., the lion is go down.) by Ionin and Xxxxxx (2002) observe that the vast majority examples with the overgeneralized be do not intend to be progressives with the missing -ing on the main verb, but rather express generic/habitual or past-tense meanings, mostly with stative verbs which do not take the progressive form. Another observation is that when the overgeneralized be is used, the main verbs do not carry any inflectional morpheme in most cases. Ionin and Xxxxxx thus conjecture that this overgeneralization of be verbs is an indication of making tense and/or agreement on the main verb when the learners have not acquired affixal agreement. Based on the findings in their study, Ionin and Xxxxxx (2002) conclude that the syntax of agreement is intact in Chinese learners’ L2 grammar and that frequent omission errors are attributed to problems with mapping from existing agreement feature to their surface morphological representation, in line with Xxxxxxxx (2000) and Prévost and White (2000). They further demonstrate that omission errors are dominant in spontaneous oral production. When finite forms (inflected forms) are used by L1 Russian children, they are almost always used with the appropriate verb forms with the correct tense-person-number features, which supports findings from Prévost and White (2000).
