Categorical and gradient homophony avoidance: Evidence from Japanese

Abstract

Many languages have been claimed to have phonological patterns that are sensitive to the need to avoid homophony -for example, a rule that is blocked if it would create a surface form that is identical to another word in the language. Such patterns always involve comparisons between words in the same morphological paradigm (e.g., singular and plural forms with the same stem). The lone exception to this generalization is Ichimura (2006), who argues that a nasal contraction pattern in Japanese is blocked by potential homophony between verbs with different stems. We present experimental evidence that homophony avoidance is not part of the correct synchronic description of the environment in which this pattern applies; rather, nasal contraction does not productively delete stem-final vowels. However, homophony avoidance does appear to affect the probability with which contraction applies. We conclude that homophony avoidance affects phonological behavior, but that absolute homophony-related blocking is restricted to morphological paradigms.

Homophony avoidance in the grammar: Russian nominal allomorphy

“…subsets of the lexicon relevant for comparison. Based on the results from the corpus study as well as previous findings (Baerman 2010, Bethin 2012, Kaplan & Muratani 2015, Pertsova 2015), the domain of homophony avoidance, and therefore the comparison set, appears to be the morphological paradigm. However, note that evidence in favour of broader, between-word homophony avoidance effects exists as well (Silverman 2009, Ogura & Wang 2018).…”

“…Frameworks dealing with contrast more generally, such as Preserve Contrast Theory (Łubowicz 2012, 2016), utilise broader comparison sets, even those containing nonce stems themselves. However, because the OT modelling in §3.3 and §4.2.2 imposes a strict limit on the comparison set, and because other literature on homophony avoidance has also found that the phenomenon is limited to the inflectional paradigm (Baerman 2010, Bethin 2012, Kaplan & Muratani 2015, Pertsova 2015), I conclude that homophony avoidance has a narrower scope in the grammar than in contrast preservation more generally. Frameworks that deal with contrast in other domains, such as Dispersion Theory (Ní Chiosáin & Padgett 2010, Flemming 2017) and Preserve Contrast Theory, although they differ from the current study in empirical coverage, also suggest the existence of a synchronic contrast preservation mechanism, and are, therefore, complementary to this study, rather than contradicting it.…”

“…This paper introduces a pattern of homophony avoidance in the Russian nominal paradigm which must be attributed to a synchronic restriction against homophony. The existence of a synchronic anti-homophony constraint has been previously regarded as unsubstantiated (King 1967, Mondon 2009, Sampson 2013, Kaplan & Muratani 2015). I believe the main reason for the scepticism is the well-established framework for diachronic homophony avoidance and the lack of a clear method of distinguishing between diachronically emergent contrast preservation and a synchronic anti-homophony constraint.…”

“…Results from the experiment suggest that speakers are less likely to apply an optional nasal assimilation process when this results in homophony with a competing word. However, it has been subsequently claimed that the results are best analysed as positional faithfulness rather than homophony avoidance (Kaplan & Muratani 2015). It is still a matter of debate whether examples of synchronic restrictions on homophony exist.…”

Category clustering: A probabilistic bias in the morphology of verbal agreement marking: Supplementary Material

“…According to expectations established by the general corpus frequency, the number of tokens of combined ma-, mai-1NS.P + mai-NEG is unexpectedly small, and highly unlikely to be a result of chance (χ 2 p < 0.001). Homophony avoidance is the obvious explanation for the unexpectedly rare usage of the ma-, mai-1NS.P variant in combination with mai-NEG, following other instances of probabilistic morphological homophony avoidance attested in recent literature (Baerman 2011;Kaplan & Muratani 2015). This is however the only instance we know of that involves avoidance of homophony in potentially adjacent morphemes, as opposed to paradigmatic alternants.…”