I recently came across a reference to the Henny Youngman “take my wife” joke, which turns on the ambiguity of that phrase, with two very different uses of take, one of them very restricted in its syntax and discourse function, the other free in both respects.
Henry “Henny” Youngman (original Russian surname Yungman) (March 16, 1906 – February 24, 1998) was a British-American comedian and violinist famous for “one-liners”, short, simple jokes usually delivered rapid-fire. His best known one-liner was “Take my wife — please.”
In a time when many comedians told elaborate anecdotes, Youngman’s comedy routine consisted of telling simple one-liner jokes, occasionally with interludes of violin playing. These gags depicted simple, cartoon-like situations, eliminating lengthy build-ups and going straight to the punch line. He was known as the King of the One Liners, a title bestowed upon him by columnist Walter Winchell. A typical stage performance by Youngman lasted only fifteen to twenty minutes, but contained dozens of jokes, delivered in rapid-fire fashion.
Here’s a video from an appearance by Youngman on the Ed Sullivan Show (5/8/66), with the iconic joke folded in close to the end:
Take my wife at first seems to have an idiomatic use of take, characterized by NOAD2 as follows:
[ usu. in imperative ] use as an instance or example in support of an argument: let’s take Napoleon, for instance.
Or, with a clear lead-in:
Linguists are often musicians. Take Ray Jackendoff. Ray is an accomplished clarinetist and also a scholar of musical cognition.
[Background from Ray's home page, where describes his work in syntax, semantics, and cognitive linguistics, adding:
Jackendoff is also a classical clarinetist, performing frequently in recital and chamber music in the Boston area. Through his musical interests, he has collaborated with composer Fred Lerdahl on a theory of musical cognition modeled on generative linguistics. The 25th anniversary of their book A Generative Theory of Tonal Music was celebrated in 2008 with conferences in Paris and Dijon and also at Tufts.]
Besides the special discourse function of this take in take NP, it is severely restricted syntactically. As NOAD2 suggests, it’s almost entirely confined to imperative (or imperative-like) sentences; I couldn’t describe the little discourse above by saying:
I took Ray Jackendoff. / I suggested we should take Ray Jackendoff. / etc.
Instead, I’d have to use take as an example or something of the sort.
(Of course, it’s very common for idioms to have very specific pragmatic functions — think of the idiom Take that!– and not uncommon for lexical items and idioms to be severely restricted syntactically; for instance, the English modals, and the obligative be of You are to leave this instant, are limited to occurrence in finite forms.)
Back to Youngman. Having delivered take my wife in such a way as to suggest it has example-providing take, Youngman pauses and then, with a gesture, adds the request particle please, forcing the audience to switch interpretations in mid-stream, to the more neutral and literal transfer-of-possession take (which is discoursally and syntactically unrestricted). That gives the one-liner its surprise value.