I did a mental double take the first time I heard my wife, Gigi, say the word policeman. She gave the second and third syllables roughly equal stress and said -man with an ash sound (what was traditionally referred to as a short vowel), represented in International Phonetic Alphabet (IPA) as /æ/. It came out the way I would say Batman or milkman. To me, the -man in policeman has a reduced stress and a schwa vowel (/ə/ in IPA), as in woman.
I actually must have done a physical double take because she asked me what the matter was. I told her. “You’re crazy,” she explained, adding that everyone said policeman the way she did.
The issue has stuck with me — not so much because of policeman, which is sexist and falling out of use, but gunman, which sadly seems to come up more and more every year. Gigi says it like policeman — with equal stress and a short “a.” I say gunmən, and so do most of the announcers on the radio, but not all of them. I searched the NPR website and of the last 10 uses of the word on the air, the count was eight for my way.
A couple of weeks ago, a Sean Penn movie called The Gunman opened, and attracted so little interest that it will probably be gone from the theaters by the time you read this. Anyway, in the trailer, when the deep-voiced announcer says the title (at about the one-minute mark), he sounds as if he is trying to split the difference, and it comes out more like the plural gunmen.
Here’s what John Wells has to say in the Longman Pronunciation Dictionary:
-man mən, mæn —This suffix may be weak or strong. (i) In most well-established formations, written as one word, it is weak, mən: policeman pə ˈliːs mən. (ii) Where written hyphenated or as two words, and in new formations, it is usually strong, mæn: spaceman ˈspeɪs mæn. Note batman ‘army servant’ ˈbæt mən, but Batman (cartoon character) ˈbæt mæn
Seeking further insight, I constructed and distributed via social media a Qualtrics online survey asking people how they pronounced these words and then others of similar construction. Gigi is from Massachusetts and I’m from New York, which led me to ask respondents to indicate the region they are from. At this point, 275 people have taken the survey, so it’s far from scientific, but it is instructive. (Please feel free to take it and add to the database.) Below is a chart showing the percentage who reported schwa, reduced-emphasis pronunciation of -man. The blue bar represents respondents as a whole, and the red those from New England.
The striking thing isn’t the regional variation, which (in this small sample) is negligible, but rather the breakdown of the words into two fairly distinct categories. Robust disagreement breaks out only about postman, pronunciation of which is roughly evenly divided among both New Englanders and respondents as a whole.
But why this particular breakdown? I posed the issue to Mark Liberman, the University of Pennsylvania linguist who started Language Log with our own Geoffrey Pullum, and Mark wrote a post about it, including a report on his his own pronunciation, which seems to jibe with the survey:
Language Log commenters, discerning as always, offered various hypotheses for the variation. One wrote:
I notice that where the -man is reduced, the first word is someone who does something, whereas the words where the derivative suffix is given a secondary stress, the man doesn’t do that thing, e.g. a mailman doesn’t mail but a policeman polices and a gunman guns. The exceptions seem to be clansman and journeyman. …
However, as Mark noted in a comment to the comment, “There are quite a few other exceptions: a gentleman doesn’t gentle, nor does an oarsman oars, or a swordsman swords. And in the other direction, a bossman does boss.”
Another suggested:
Maybe the difference has to do with whether the -man word can also be thought of as a group of men (as a collective abstraction — a trade, profession, or organization) or is primarily understood in terms of a solo man? The first group of words describes men we also often think of in terms of a group of like people, the second group describes men we more often think of as individuals.
That makes sense, kind of, but an explanation based on relative newness of the term (also suggested by John Wells) seems to be the best answer. Another commenter wrote:
I think that for most of these compounds, [mən] vs [mæn] has more to do with the age of the formation than any semantic reasons. Apparently, postman dates to the 1520s, while I think mailman must be more recent. The longer a compound has been part of the language, the more likely it is to be subject to vowel reduction.
Familiarity also plays a large part, which is why British English reduces saucepan while American English doesn’t, as Americans are more likely to call it a pot.
Nobody had anything to say about why some people, like Gigi, go against the grain. I await Lingua Franca readers’ insight. In the meantime, here’s a clip posted on Language Log where the gang from Friends address the issue: