Skip to main content

Connacher Murphy

Alliterative Affinities Revisited

In 2021, I wrote a brief piece, Alliterative Affinities: Do Parents Select First Names to Match Surnames? In short, the answer was yes—unless your name starts with a “J”. Unfortunately, the data for the article came from the 1930 Census. Now, I’ve updated the results with much more recent data.

I use up-to-date voter registration data from the North Carolina State Board of Elections. In this update, I’ll ditch the digression on baby names and whether names can have a causal effect on a child’s outcomes in later life.

Let’s say f is an individual’s first initial and l is their last initial. I denote the probability that an individual’s first name starts with “A” as P{f=A}, while the associated last-initial value is P{l=A}. We’ll consider the probability of an alliterative pair of names beginning with “A”, P{f=Al=A}. I then compare this value to an independent benchmark: P{f=A}×P{l=A}, which is the probability of an alliterative pair of names beginning with “A”, if first and last initials were independent. I then calculate the percent difference between these two values ΔAP{f=Al=A}P{f=A}×{l=A}1.

I repeat this process for all letters in the alphabet, ωΩ. I summarize Δω in Panel A of Figure 1 for the 18 most common alliterative combinations (as measured by the independent benchmark given above). For the remaining eight letters, the lower denominators result in some exceedingly large Δω values, all of which are positive. Indeed, Δω is positive for every letter except “J”; alliterative pairs of “J” are 3% less common than predicted by the independent benchmark. I found this result rather surprising, as I initially expected a positive and relatively large value for ΔJ. As far as I can gather, others are surprised too. Why are alliterative pairs of “J” so salient?

For the top 4 letters—“M”, “S”, “C”, and “B”—alliterative pairs are over 5% more common than would be expected under independence.

Figure 1. Alliterative Tendencies
alliterative-names

I now summarize these results across the alphabet. The probability that an individual has an alliterative name is ωΩP{f=ωl=ω}, and the probability under independence is ωΩP{f=ω}×P{l=ω}.

While the independent benchmark predicts an alliterative frequency of 5.35%, 5.74% of names are alliterative. Accordingly, alliteration is 7.3% more likely than would be expected if first and last initial were independent.

In the original article, I excluded married women due to the common practice of adopting a spouse’s last name. I do not observe marital status in this updated analysis. However, given the increase in age at first marriage and (I suspect) decreased tendency of taking a spouse’s last name, I expect gender to matter less compared to the 1930 data. When I restrict the sample to men, alliteration is 7.4% more likely than predicted under independence, a slight increase.