This paper presents results of aresponse-latency (RL) experiment with English-speaking learners of German that investigated towhat extent—if any—two different groups of second language (L2) learners permitraising of the thematic verb. The framework under which the study was conducted involvesvarying theoretical predictions derived from the native-language (NL) transfer view of Schwartzand Sprouse (1994, 1996), the gradual-development view of Vainikka and Young-Scholten(1994, 1996), the underspecification view of Eubank (1993/1994), and a local-impairmentview that presents a more parsimonious solution to the L2 developmental problem than standard“no access” views. The L2 results reported here show that learners respond indifferent ways to stimulus sentences with raised and unraised verbs depending on thesubjects' level of development. Surprisingly, it is the less advanced learners who exhibitan RL preference for apparently raised-verb experimental stimuli; the more advanced learners donot differentiate between raised-verb and unraised-verb stimuli. Analysis of these findingsreveals that the less advanced group may only project VPs, consistent with thegradual-development view of Vainikka and Young-Scholten. Crucially, however, the LocalImpairment Hypothesis is the only view that is consistent with the results from the moreadvanced learners, especially when they are seen in developmental context.