The trend in last letters of boys’ names: Why this is cool

Yesterday I showed Laura Wattenberg’s graphs showing the most popular last letters of boys’ names in 1906, 1956, and 2006. The quick story is that 100 years ago, there were about 10 last letters that dominated; 50 years ago, the number of popular last letters declined slightly, to about 6; but now, a single letter stands out: an amazing 36% of baby boys in America have names ending in N.

This is super-cool. As a commenter wrote, there should be some sort of award for finding the largest effect “in plain sight” that nobody has noticed before.

But, beyond pure data-coolness, what does this mean? My story, based on reading Wattenberg’s blog, goes as follows:

100 years ago, parents felt very constrained in their choice of names (especially for boys). A small set of very common names (John, William, etc.) dominated. And, beyond that, people would often choose names of male relatives. Little flexibility, a few names being extremely common, resulting in a random (in some sense) distribution of last letters.

Nowadays, parents have a lot of freedom in choosing their names. As a result, there are lots and lots of names that seem acceptable, but the most common names are not so common as they were fifty or a hundred years ago. With so much choice, what do people do? Wattenberg suggests they go with popular soundalikes (for example, Aidan/Jaden/Hayden), which leads to clustering in the last letter. Even so, the pattern with N is so striking, there’s gotta be more to say about it.

But I like the paradox: 100 years ago, the distribution of names was more concentrated but the distribution of sounds (as indicated by last letters) was broader. Nowadays, the distribution of names is more diffuse but the distribution of sounds is more concentrated.

Less constraint -> more diffuse distribution of names -> more concentrated distribution of last letters.

This must occur in other aspects of life. For example, consider food. We eat lots more different types of food than we used to, but a single ingredient–corn syrup–makes up more and more of our diet (or so I’m told). Again, lack of constraint (this time for economic reasons) leads to more diversity in some ways and more homogeneity (by choice) in others.

20 thoughts on “The trend in last letters of boys’ names: Why this is cool

  1. I think greater variation in girls names has something to do with it. It is no longer so clear what is a boy's name and what is a girl's name — so it is more important for the boy's name to include a sex marker. Aidan is a boy's name just as Aida is a girl's name.

  2. I agree completely that this is a very cool observation but can't see that the corn syrup analogy works at all. Indeed, it may well be the opposite effect – we are eating lots of corn syrup because it is being forced upon us by companies for (their own) economic reasons. Lots of people choose -n names because they like the sound of them. Very few people choose to eat corn syrup, it's simply unavoidable. I dunno, maybe it's just that naming, in some sense, has become more democratized over the last century and we are seeing this interesting pattern, while food has certainly become far more available over that time it is less democratic now than it was back then due to the rise of the industrial food giants. The pattern we see is much more depressing…

  3. Perhaps we are seeing the emergence of a linguistic marker — but I know little of linguistics.

    Consider the top of the male last letter frequency distribution in 2006:
    Last letter % of Males % of names
    n 35.3% 33.8%
    r 8.9% 6.9%
    l 7.7% 6.2%
    s 7.2% 6.8%
    e 6.7% 10.6%

    Now look at the same distribution for female last letter frequency in 2006:
    Last letter % of Females % of names
    a 40.0% 39.3%
    e 17.8% 18.6%
    n 13.5% 13.8%
    y 13.3% 11.3%
    h 6.5% 5.5%

    (The last column counts each name on the 1000 list equally.) Note that the "n" frequency for boys is about the same order of magnitude as the "a" frequency for girls.

    So perhaps we are culturally evolving to that letter becoming a marker of the child's sex.

  4. Mitzi said she was going to write in and set you straight on the linguistics, but I think she clued me in enough that I can do it for her. She wrote her MS thesis on Swedish nickname formation.

    When I enumerated the popular N names (e.g. "Dylan", "Mason", "Ethan", among my cousins' kids, "Dakota" being the odd cousin out), her immediate reaction was that it's because Germanic languages (like English) favor two syllable names with stress on the first syllable (like "Andrew" or "Robert"). Even if your full name doesn't meet that pattern, your nickname's likely to (e.g. "Dakota"/"Cody" [it doesn't matter that the root is from the native American language of the same name — we're speaking English]).

    Contrast this with Indian names like "Suresh", which typically stress the second syllable. My friend Mukesh spent hours teaching English speakers how to pronounce his name. Or even with the French (a romance language) pronounciation of Robert (stress on second syllable).

    The really cool part is that all of these names have reduced (unstressed) vowels in their second syllables (called "schwa", and written 'ə' in the international phonetic alphabet and in some dictionaries). So the "in" in "Dustin" is pronounced like the "an" in "Dylan" and the "on" in "Mason". (One reason English is so hard to spell is that unstressed vowels all sound the same.)

    With unstressed syllables, the consonants also tend to get reduced to simple nasals ("n" or "m"). So the "n" is predictable here, too, as it's preferred over "m" in English. So given names tend to end in unstressed syllables, we shouldn't be surprised at all the names ending in the sound "ən".

    I think this explanation neatly couples with the previous comment about the sociological forces at work. Once we start making up names or bringing more in, we exert our preferences for sounds over our preferences for conformity.

  5. I think the story is a little more nuanced – take e as a last letter of girl's names. In the 1890's & 1890's it accounted for 34% of all girl's names. Since then it has been steadily trending down to around 12% – i.e. basically the opposite trend to n in boys names.

    Other letters show interesting patterns when looking at the full 120 years of data. Plots and data to follow shortly.

  6. All this talk of names makes me think I should suggest "A Matter of Taste" by Stanley Lieberson, a book about names as an instance of 'fashion' for which a lot of cross country long-term data is available (also, it is theoretically less influenced by external organizations with commercial interests in the fashion we consume – the corn syrup, for example, or athletic shoe style).

    He has a lot of graphs and tables and notes many interesting patterns regarding phonemes and all that. It's been a little while since I read it and I don't remember if he picked up on the -n trend.

  7. Maybe the explanation is simpler. Parents are being more conformist in their choices of names, just conformist in a different way than parents were a century ago.

  8. (re Carpenter) So, unstressed vowel+consonant tends to go to "uh" plus "un" (final n spelling for males) and unstressed vowel tends to go to "uh" (final a spelling for females)?

  9. Bob: But doesn't your story explain too much? Under your story, wouldn't the N ending have dominated in 1906 as well?

    Hadley: Wow–that's so cool. Thanks!

    Mark: I read the Lieberson book and was disappointed. I mean, it was OK, but by the time I was through, I didn't feel I'd gotten much out of it.

  10. @ZBicyclist: Unstressed vowels all get pronounced as schwas in English. The "n" just tends to go along with reduced syllables in general, but you can find it replacing other sounds in fast speech. For instance, "ing" often becomes "in" in fast speech in unstressed syllables (e.g. "going"/"goin"). There are lots of such reductions; Americans use a flap for both "dd" and "tt" in "ladder" and "latter" unless very carefully articulatin'. Then if you're from Boston or Australia, go ahead and drop that final "r".

    If you actually try to segment real speech, it often just sounds like an undifferentiated noise in unstressed syllables. If you don't have context, it's pretty much impossible to transcribe conversational speech a syllable at a time.

    @Andrew: I think David's comment above helps here. If there are two effects, one preferring a certain kind of sound versus one preferring historically/culturally relevant names, then you have a balancing act.

  11. @ZBicyclist and Bob Carpenter
    It actually becomes a syllable 'n' (transcribed as an n with a little line underneath) the N can be a whole syllable in and of itself in english with no real vowel because it has enough phonetic substance (is a sonorant high on the sonority scale to be precise). The same is true of l and r. (consider paddle and power).

    I had a similar reaction about the gender marking in combination with phonological constraints. Interesting the statistical/phonological composition of a word (stress location, number of syllables, etc) reliably classifies it by its part of speech. See work by Monaghan, Chater, Christiansen, 2005, 2007. I guess I'm not surprised to see the same thing in names

    Bob's story nicely explains diachronic change. Sociolinguistics have shown that people tend to want boys to have masculine names because that helps you get ahead in the world. Parents give girls neutral/masculine names and the names gradually become considered feminine reducing the names available for naming boys that are strongly masculine… That in combination with constraints on phonological form in ENglish could have you end up here.

  12. @Hadley: Thanks. I particularly like the graph last-letter.png that lays out the entire string for both boys and girls.

    "n" is a fascinating graph. Pre-1900 it is low for girls, common for boys. Then it increases for girls and is flat for boys until, by about 1940, it's equally common in both sexes. Then it increases for males strongly and decreases slightly for females.

    This seems counter to the notion that once females start getting assigned a name, it becomes less common for males, although these are sounds rather than entire names.

    I got talked into teaching a forecasting course next spring; I think these would be good time series to have students try to forecast (with recent years a holdout sample).

  13. @Eric there's definitely a geographic trend (esp. in terms of the cultural make up of various regions). The SSA makes state level data available too – it should be too much work to extract that as well.

  14. Another possible factor is the native language of immigrants.
    East Asian languages (Chinese, Korean, Japanese) and many others only allow the nasal /n/ in syllable codas. Since name-final sounds are codas (or vowels), only nasal consonants will occur name-finally.

    There are two ways this may affect the distribution of name final consonants, to the extent that non-native speakers have more babies relative to the general population (a well-documented effect in immigrants). First, if parents name their child in their native language, it will drive up the relative frequency of word-final nasals. Second, even if parents name their child in English, they are likely to exhibit a first-language effect whereby they choose a name that is more in conformance with the phonotactics of their native language, e.g. choosing a name that ends in /n/ because that is legal in the native language.

  15. i just came across this and i the names i had chosen relate to this completely
    Aida for a girl
    Zain for a boy

  16. This effect where less constraint paradoxically seems to cause less diffuse variation somewhere else appears in creative endeavors a lot I think. It reminds me instantly of the careful design of creative constraints that go into jazz improvisation, for instance.

    The time people have to spend creatively thinking of baby names is finite, so perhaps there is a finite variability available there — the space covered by the 95% intervals is approximately constant. If you let up on constraints so that it expands in one dimension, other dimensions, like the choice of the final syllable, shrink.

Comments are closed.