This study presents results from a corpus-based analysis of the expression of attitude, emotion, certainty and doubt (stance) in a large corpus of British and American conversation. Stance marker frequencies were assessed through an automated procedure for identifying stanced lexical items occur-ring in particular grammatical frames. The frequencies were analyzed with a multi-variate statistical procedure known as factor analysis which identifies co-occurrence patterns (factors). These factors can be understood to be the most salient moods of stance. Three factors were identified as character-istic: 1) informal AFFECT (American dialect-based), 2) boulomaic planning (American work-based) versus small talk (British dialect-based), and 3) hedged opinion (British dialect-based). Social norms were identified by examining the factors in light of discourse context and interpersonal relationships among speakers. Cross-cultural misunderstandings seemed particularly likely in work contexts, where Americans preferred boulomaic verbs (want, need), and British preferred evidentials (know, maybe). Differences in informal adult conversations are also potentially important, where Americans used many more affect markers (such as love, crazy). More work on pragmatic or functional domains using multi-variate analysis is proposed in the conclusion.