There is a very good new paper by Lawrence Edwards and Robert Z. Lawrence. In this case the conclusion is clearer than the abstract:
…the fear of rising US wage inequality from developing-country imports in recent years are unwarranted. While conventional trade theory makes such expectations plausible our investigation reveals they are far off the mark. At the most disaggregated level for which comprehensive skills are available we have found that the US industries competing with developing country imports are not particularly intensive in unskilled labor. Moreover, the relative effective prices of the US industries that are unskilled labor-intensive have actually increased rather than decreased since the early 1990s. Changes in effective US prices form whatever cause have not mandated changes in relative ages. Neither have changes that can be ascribed to import prices mandated increases in wage inequality.
The most likely explanation for the data is:
The goods exported by developing countries are highly imperfect substitutes for those produced by developed countries. This means that for the most part, unskilled US workers are not competing head to head with their counterparts in developing countries.
You can find ungated copies here, though in some browsers they seem to create problems.