This paper addresses the question of how to predict which alternatives are active in scalar implicature calculation, and the nature of this activation. It has been observed that finger implicates 'not thumb', and a Manner-based explanation for this has been proposed, predicting that if English had the simplex Latin word pollex meaning 'thumb or big toe', then finger would cease to have the implicature 'not thumb' that it has. It has also been suggested that this hypothetical pollex would have to be sufficiently colloquial in order to figure in scalar implicature calculation. This paper makes this thought experiment into a real one by using a language that behaves in exactly this way: Spanish has pulgar 'thumb' (< pollex), a non-colloquial form. We first use a fill-in-the-blank production task with both English and Spanish speakers to guage the likelihood with which a speaker will produce a given form as a way of describing a given digit. Production frequency does not perfectly track complexity, so we can then ask whether comprehension follows production frequency or complexity. We do so using a forced choice comprehension task, which reveals cross-linguistic differences in comprehension tracking production probabilities. A comparison between two RSA models -- one in which the speaker perfectly replicates our production data and a standard one in which the speaker chooses based on a standard cost/accuracy trade-off -- illustrates the fact that comprehension is much more closely tied to production probability than to the mere existence of sufficiently simple alternatives.