In this article I argue that the longstanding practice of depicting Italian Americans as opera lovers stems from a tradition associating Italian immigrants with mechanical music devices. As a growing number of Italian unskilled labourers entered the United States in the second half of the nineteenth century, they were stereotyped as street musicians, and especially as organ grinders, in mainstream popular culture. Beginning in the 1900s, recording manufacturers strove to make home phonographs appealing to the middle class by breaking the chain of mechanical, social and racial associations that connected the phonograph with earlier musical devices such as the barrel organ, and with those who played them. Because of the prominent marketing role that record labels assigned to Italian opera, this commercial strategy had important consequences for the genre as well as for Italian immigrants, who leveraged opera's renewed visibility and audibility into an effective vessel for social and political empowerment.

In the past twenty years the existence of a sense of ethnic belonging among immigrant groups of European ancestry in the United States has become the focus of frequent debates and polemics. This article argues that ethnicity cannot be understood if it is abstracted from concrete social practices, and that analyses of this construct need to be based on ethnographic observation and on the study of actual talk in interaction. This interactionally oriented perspective is taken to present an analysis of how Italian ethnicity is constructed as a central element in the collective identity of an all-male card playing club. Linguistic strategies, particularly code-switching, are central in this construction, but their role becomes apparent only when language use is analyzed within significant practices in the life of the club. Code-switching into Italian is used as an important index of ethnic affiliation in socialization practices related to the game and in official discourse addressed by the president to club members through the association of the language with central domains of activity.I would like to thank two anonymous reviewers and the editor of this journal, Barbara Johnstone, for their insightful suggestions, which have substantially contributed to the shaping of this article.