My posting on Susan Sarandon’s “The Curious Case of Benjamin Britten” didn’t say anything about the details of her error, which is of a very common type. Let me explain.
This is a word retrieval error. (Word here is a brief label for lexical units in general, taking expressions that are words but have words as parts — like the proper names Arnold Zwicky and Benjamin Britten and the compound nouns chemistry teacher and cherry pie — as well as other composite expressions that are larger than words, like American century and still larger idioms like point of entry.)
The thing about lexical units is that we’re “stored them away” (in the usual metaphor for these things; the point is that there’s some sense in which we “know them already”). What happens when we try to retrieve a lexical unit from storage?
The standard story is that two things can go wrong: you can pull up an item that’s semantically related to the one you’re looking for (a semantic error), or you can pull up an item that’s phonologically related to the one you’re looking for (technically, a Fay/Cutler malapropism, but I’ll just use the label phonological error). More complex things can go wrong — you could retrieve two competing items and produce a blend of them, for example — but I’ll stick to simple errors in retrieval here.
[In a different sort of mistake, you successfully pull up the item you're looking for, but it's not the one most other people use -- a private meaning, if the divergence is semantic, a classical malapropism, if the divergence is phonological.]
There are plenty of pure semantic errors, for example producing an opposite instead of the intended target: “It was too hot” instead of “It was too cold”. (Even here, there are probably some phonological effects favoring semantic errors where the intended target and the actual production share phonological properties, in particular their metrical pattern.)
For phonological errors, the contribution of meaning relationships is clear and strong. To start with, phonological errors almost always preserve syntactic category: nouns substitute for nouns, verbs for verbs, and so on. At a finer grain, mass nouns substitute for other mass nouns, abstract nouns for other abstract nouns, transitive verbs for other transitive verbs, and so on. At an even finer grain, the substitutes tend to be from the same semantic domain as the targets: nominal is a more likely substitute for numeral than numinal is, especially if the speaker is a linguist.
So proper names are particularly likely to substitute for other proper names, as with Britten for Button in the Sarandon example this all started with. The phonological relationship is important; Benjamin Britten is a more likely substitute for Benjamin Button than Benjamin Bratt is (despite the shared /b/ and /t/; the metrical pattern is different), and it’s also a more likely substitute for Benjamin Button than Benjamin Franklin is (despite the shared trochaic pattern; the consonants are mostly different).
Still another effect is familiarity. Substitutes tend to be more familiar items (to the person who produces them) than the targets they replace. (Not an astounding insight, but then many things you think are true, and obviously should be true, aren’t.) It’s reasonable to assume that Benjamin Britten (the real person) was more familiar to Sarandon (from years of experience) than Benjamin Button (the fictional character, recently impelled to prominence by the release of the movie) was.
Sarandon wasn’t fated to say “Benjamin Britten”, but it was a trap waiting for her.
[Note: this is not a posting on all the types of "mistakes", inadvertent and advertent, that are out there, just a little riff on one small part of this world. It's a rich world, which you can sample here (a somewhat outdated, but user-friendly, and fairly compact, survey). Links to some materials on slips are assembled here.]