In his comments on Inside Higher Ed's recent story on oral history and IRBs, Eaton Lattman, Dean of Research at Johns Hopkins Krieger School of Arts and Sciences, writes that "many oral history projects probably do not meet the OHRP definition of research, and are therefore free from the need to go to an IRB for approval. They are not hypothesis-based projects and they do not produce generalizable results."
This interpretation of generalizability is consistent, as best I can tell, with the intent of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, which first established generalizability as part of the definition of human subjects research. It also resembles Columbia University's recent statement on oral history.
Dean Lattman's comment contrasts with definitions put forward by Michael Carome of OHRP and consultant Jeffrey Cohen, who maintain an oral history project is generalizable research if it lacks a hypothesis but draws conclusions, informs policy, or creates an archive for use by future researchers.
Thus, Hopkins joins Columbia and Northwestern in explicitly freeing oral history--as practiced by oral historians--from IRB review. While such advances are welcome, more welcome still would be a statement along the same lines from OHRP itself.