<p>Black Pentecostal churches in America are known for their intensive use of music during worship services, including congregational singing and dancing. This paper deals specifically with their ritual trance dance, called shouting, and describes the micro interactions among congregations where it occurs. By doing so, I aim to discuss an important but usually overlooked aspect of the mechanism through which (possession) trance dance emerges. Anthropological inquiries have tried to understand the trance dance as a cultural representation; either by presenting how it is interpreted within each culture, or showing how it is culturally constructed, or both. Such perspectives are based on the semiotics that imparts conceptual meaning to the trance dance. While those perspectives of understanding shouting are undoubtedly important, they do not focus on its indeterminacy. Indeterminacy partly means its improvisational character: there are scenes and settings where shouting is expected, but no one can really predict when it will occur. Indeterminacy also means the ambiguous character in which a shouting is interpreted by others. Moreover, the heart of Pentecostal belief is also grounded in that indeterminacy: one must be "filled" with the Holy Spirit as a gift, and no one is supposed to be able to control it. To make that "indeterminate" process visible, this paper starts by showing that shouting is dependent upon acts of other people surrounding the shouter, and how a shout emerges through the interactions between them. A typical occurrence of shouting goes as follows: During a pastor's preaching or right after the choir finishes singing, someone who is visible to most in the congregation starts shouting, or gesturing as if he/she will shout. The musicians, by watching it, start playing "shout music," and several members of the congregation begin shouting, as if to follow the music. There, its interactive characteristic is obvious. There are even more micro interactions among the people in the congregation, and we can find them when those surrounding a shouter become "involved" in the shouter's act. In previous studies, such involvements are mentioned only as the acts of preventing shouters from hitting people or things, or keep them from hurting themselves. However, when looked at more closely, one can observe that certain types of involvement support or even encourage shouting. Moreover, there are cases when someone who does not normally shout starts to do so after being touched by others in certain ways. The interaction concerning shouting is one type of "focused gathering" (Goffman, 1985/1961), which I call "Engrossment-Involvement Relationship (E-I.R.)" (here, shouting is paraphrased as "engrossment"). I shall use the E-I.R. as an analytical unit, rather than shouting itself, for it enables an examination of the dynamism of how individuals are motivated to shout; in other words, its indeterminate character. Shifting the focus to E-I.R. allows us to notice cases where shouting does not arise smoothly but requires others' involvement during the gradual emergence of shouting. In such cases, the individual who is about to shout will gesture in ways slightly different than that of shouting itself-i.e., by exaggerating the sound of stomping, and so on. I call that gesture "demanding," because in an interactional sense it can be seen as demanding others' involvement and engrossment (shouting). The gesture occurs slightly before the shouting, and is geared more toward the public. Similarly, there are times when others' involvement comes out prior to shouting. I call that the gesture of "expecting," because it leads to the expectation that someone will start shouting. Those terms are necessary to examine cases where shouting fails to occur satisfactorily. The shouting can</p><p>(View PDF for the rest of the abstract.)</p>