In pairs of names, male names often precede female names. Wright & Hay (in press) investigated this bias and found that preferences for name-ordering are constrained by a combination of gender and phonology. They examined popular names in American English according to phonological constraints proposed by Cooper & Ross (1975) that govern ordering of words in binomial expressions; they found that male names are characterized by "First Position Phonology" (i.e., phonological properties that lend them to be preferred in first position), while female names are characterized by "Second Position Phonology." They also tested two of these constraints experimentally (syllable count and consonant clusters) and found further evidence that phonology plays a role in predicting ordering of names; in addition, they found that when phonology is controlled, an independent gender bias still exists, leading to an overwhelming tendency to place male names first.

In the present study, we investigated two additional constraints laid out by Cooper & Ross--sonority of initial and final segments. Cooper & Ross argue that words with initial sonorants precede words with initial obstruents in binomial pairs, but words with final obstruents precede words with final sonorants. We designed a name-ordering experiment that tested these constraints, as well as their interaction with gender. We constructed 100 pairs of names to independently manipulate each of these factors. 28 subjects were presented with the stimuli and asked to indicate
which ordering sounded most natural to them. An example is given below:

  1. Don, Matt

______ and _______ went sailing.

Our results show that a stepwise linear model retains all three factors as significant predictors of the direction and strength of subjects' preferred orderings (Gender coef: 2.9, Final Segment coef: 1.9, Initial Segment coef: .59, Overall Model: r2=.5, p<.001). The strongest predictor is gender and the weakest is the sonority of the initial segment. Further evidence that gender exerts the strongest influence comes from the condition where phonological biases are pitted against the gender bias: here subjects show a significant preference for placing male names first (i.e., subjects are more strongly influenced by the gender bias than by the phonological bias for name pairs which contain both).

Strikingly, the ranking of factors obtained from our experimental results correlates strongly with the degree to which these factors are over-represented in male names. Wright & Hay found a significant difference between final segments of male and female names (i.e., male names were significantly more likely to end with an obstruent), but a non-significant trend for initial segments. In addition, syllable count was the strongest phonological bias in male and female names - a phonological factor which Wright & Hay demonstrated so strongly influences subjects' preferred
rankings, that it can even overcome a conflicting gender bias. Thus, our results, together with those reported by Wright & Hay, suggest a deep "phonological conspiracy" in proper names in English. The more likely a phonological variable leads a word to be preferred in first position, the more likely it will be over-represented in male names.