There has long been a debate in phonological theory about the roles of substance and markedness, with linguists taking diametrically opposed positions on the importance of these to phonology. Perhaps the most frequent phonological diagnostics used to determine whether a feature is marked or unmarked are emergence-of-the-unmarked diagnostics, including neutralization and epenthesis. In this talk I present a cross-linguistic study of neutralization of place of articulation, concluding that this survey does not provide evidence for a universal place of articulation markedness hierarchy.

It is often claimed that coronal and glottal places of articulation are what one finds in positions of neutralization as a result of neutralization (e.g., de Lacy 2006 for recent work on this). In fact in languages in which no contrast exists between places of articulation in a word-final position, that place of articulation can be restricted to (using stops as an example) coronal (e.g., Finnish) or laryngeal (e.g., Yagaria), but it can also be labial (e.g., Nimburan) or velar (e.g., some Fuzhou). In languages which contrast two places of articulation in this position, the following contrasts are possible (taking into account only languages with labial, coronal, velar distinction): coronal-labial (e.g., Kiowa), coronal-velar (e.g., some Chinese dialects), labial-velar (e.g., some Vietnamese dialects). All possible combinations of two places of articulation thus are found cross-linguistically.

Neutralization in the cases discussed above is passive in the sense that they are based on distribution, rather than synchronic alternations. Neutralization to the different major places of articulation is found actively as well, with synchronic alternations. While active neutralization to coronal and glottal places of articulation is well reported, in addition active neutralization can result in a labial place of articulation (e.g., Manam) as well as in a velar place of articulation (e.g., some Spanish dialects).

This study of neutralization of place of articulation suggests that neutralization is not very helpful as a diagnostic in assessing cross-linguistic universal phonological markedness as cross-linguistically the full range of place features can appear in neutralization positions. Two questions arise. First, why does neutralization not yield information about universal phonological markedness when it s generally perceived to do so in the literature? A basic principle is at work here: in the absence of contrast, the phonetic realization is phonologically indeterminate. Second, why is it so commonly believed that certain places of articulation can be identified as universally unmarked? This follows from the important role of articulatory and perceptual factors in realizing the output of neutralizaiton; however, other reasons, including social factors, can override these. Thus emergence-of-the-unmarked phenomena reveal that a number of different factors can play a role in determining the output for a particular language, calling into question substantive theories of featural markedness.