For lack of space it was not possible to include the following information in the paper.

In addition to selecting the best fitting punctuation symbol, subjects were asked to indicate for every response whether they were sure about their choice or not. These certainty scores reveal that subjects are reasonably confident about their responses when the stimulus ends in an L% tone (89% ‘sure’ responses), a little less confident when the stimulus ends in a H% tone (62% ‘sure’ responses) and rather unsure when they respond to a level ending contour (43% ‘sure’ responses). This means that subjects are the least confident when they have to respond to a contour ending in an unspecified boundary tone. 
	The fact that the %-contour is responded to by 34% full stop responses could be due to the relative unnaturalness of the synthesized contour, as mentioned in the main text, but it is also possible that subjects interpret this contour as some sort of bored exclamation. In the absence of a fourth response category (exclamation mark), they select the closest fitting one: the full stop.
	Looking at the percentages of full stop, comma and question responses, no significant effect of subject is found for the full stop responses (one-way ANOVA F(25,1014) =1,264, ins.). There is, however, a small effect of subject for the percentage of comma and question responses (F(25, 1014) = 2,557, p<.001; F(25, 1014) = 2,954, p<.001 respectively). Post hoc analyses reveal largely overlapping groups. This means that subjects generally respond in a similar manner, but that there are small differences between subjects, mainly in the number of comma and question responses.