{"items": [{"author": "Leland", "source_link": "https://plus.google.com/114575814092755493073", "anchor": "gp-1434522147840", "service": "gp", "text": "I think we've got good evidence that this isn't true (though, see caveat below). See Warner et al 2004 [1] on \"subphonemic differences\" in Dutch (with summary of some other findings in other languages). IIRC there's some discussion of orthographic influences in the introduction, and then their main experiment on this is section 6.\u00a0\n<br>\n<br>\nSubphonemic differences, in this context, are systematic differences in pronunciation that are not contrastive. This is a pretty weird phenomenon, but we think we have evidence for it in some languages. One problem that has been raised with this experimental evidence, though, is that it could be purely the effect of orthography. In fact, in some of the cases where we've found subphonemic differences, if you present the stimuli orally (i.e. telling the subject \"say a word that means...\" or sim. rather than presenting the written word itself) the effect goes away.\n<br>\n<br>\nWhich brings us to the caveat: All the studies that I know about this are specifically cases where the speaker is reading a written text out-loud. We've got no evidence (either way, afaik) about whether such an effect holds in normal speech.\n<br>\n<br>\n[1] PDF : \nhttp://www.cs.indiana.edu/~port/teach/641/Warner.incomplt.neutrztn.JPhon2004.pdf\n\u00a0", "timestamp": 1434522147}]}