Ask Language Log: an and ambiguity

« previous post | next post »

In this morning's mail:

My friend and I are avid Language Log readers. We were recently conversing over IM, and she was telling me about her boyfriend's great-aunt. Among the things she mentioned:

"She worked when women didn't work very much and never got married."

I interpreted her statement as my friend alluding to a time when women both didn't work and did not get married. After a few moments, I realized she was telling me that the great-aunt had a job and never got married; "when women" only modified "didn't work very much." We are unsure which reading is technically correct and therefore decided to ask.   Any insight you could provide would be greatly appreciated.

I'm not a syntactician, but I usually take the morning shift here at Language Log Plaza, so I'll do my best with this one — luckily, it seems pretty straightforward.

You're right that the two readings depend on two alternative structures, but it's probably not helpful to think about what "when women" modifies — the two words "when women" are not a syntactic unit, and their relation to the rest of the sentence doesn't involve what grammarians usually mean by "modification".

Rather, the two readings that you describe depend on a difference in which verb phrases and is taken to conjoin. On both readings, "never got married" is the second conjunct. The first conjunct might be "worked when women didn't work very much" (in which case the subject of both conjuncts is she), or it might be "didn't work very much" (in which case the subject is women). Your friend meant the first of these alternatives, while you (at first) understood the second.

If we put in just the relevant pieces of the trees, and aren't too fussy about node labels — I'm not a syntactician, remember — the difference is something like this:

(A)
(B)

Both structures are perfectly grammatical, and thus both interpretations are "technically correct", though perhaps the paired negatives "didn't" and "never" make your misinterpretation (B) easier than it should be. A simple way to prevent this problem would be to add a comma and another copy of she, thus creating a conjunction of sentences: "She worked when women didn't work very much, and she never got married".

But it's not hard to find plausible examples of both types of structure:

(A) We saved when times were good and used that to tide us over in lean seasons.
(A) I read it when I was in high school and never forgot it

(B) It almost broke my heart when he died and left me there alone.
(B) I started designing costumes when I was in high school and worked at a costume shop.

The conjunctions and and or are often technically ambiguous, but we usually don't notice it because the context or the content makes it clear what was meant.

When a recipe tells us to "Beat ripe bananas and sugar together until smooth", we know that it means "[ripe bananas] and sugar", because ripeness is not a concept that applies to sugar. In contrast, when we read that we should "Store ripe peaches and nectarines in the refrigerator", we interpret the instructions as dealing with "ripe [peaches and nectarines]", because why would you refrigerate unripe fruit?

In the case of the sentence that you asked about, real-world knowledge about the history of sex roles tells you that interpretation (B) doesn't make sense. But by the time you saw the problem, it was too late to avoid a conscious boggle.

[Update -- some more complete trees, submitted by Karen Davis:

There are a couple of apparent slips of the fingers -- the nodes labelled NP2 should be S, for example...]

Share:



26 Comments »

  1. Chris Allen said,

    April 28, 2009 @ 6:58 am

    I often daydream about asking the folks at language log interesting questions, so seeing you answer an E-mail from a reader is like seeing Santa Claus just before he climbs up the chimney.

  2. Stephen Jones said,

    April 28, 2009 @ 6:59 am

    As Geoff would rightly point out, a comma would resolve the ambiguity perfectly.

  3. Ton van der Wouden said,

    April 28, 2009 @ 9:06 am

    Stephen Jones said,
    "As Geoff would rightly point out, a comma would resolve the ambiguity perfectly."

    But as far as I know, most people do not use commas in their speech.

  4. D. Sky Onosson said,

    April 28, 2009 @ 9:22 am

    Diagram B seems to be at odds with the statement that the two words "when women" are not a syntactic unit. A non-linguist might be justifiably confused.

    [(myl) The floor is open for contributions of more complete trees...]

  5. Grep Agni said,

    April 28, 2009 @ 9:33 am

    @Ton van der Wouden:

    The conversation took place over IM. Commas are allowed there.

    For what it's worth, reading B is more natural for me.

  6. Rick said,

    April 28, 2009 @ 9:40 am

    Here some syntactic ways to resolve the ambiguity in speech as well as in writing:

    (A) She worked when women NEITHER worked very much NOR got married.

    (B) She worked when women did not work very much and SHE never got married.

  7. Bill Walderman said,

    April 28, 2009 @ 9:43 am

    "I started designing costumes when I was in high school and worked at a costume shop."

    Isn't this ambiguous, i.e, open to the (A) interpretation, too?

  8. Faldone said,

    April 28, 2009 @ 9:45 am

    I suppose it could be confusing for anyone who thinks there might have been a time in the recent past when women, as a rule, both did not work and did not get married. One wonders, however, how these women survived.

  9. Picky said,

    April 28, 2009 @ 9:54 am

    No, no .. they didn't work VERY MUCH. It's getting married that adds to a woman's workload.

    In a formal context that comma would have gone in, wouldn't it? It's part of this business of the situation allowing an informal style of writing, which one happily relaxes into … only to find (too late) that what one has written is riven with ambiguity.

  10. Stephen Jones said,

    April 28, 2009 @ 10:41 am

    But as far as I know, most people do not use commas in their speech.

    True, they use pauses and intonation instead.

  11. Jesse Tseng said,

    April 28, 2009 @ 12:01 pm

    In 50 years of IM'ing, Strunk & White have never let me down: "Be clear".

  12. Bloix said,

    April 28, 2009 @ 12:31 pm

    In speech one would be more likely to say,
    "She worked when women didn't work very much and she never got married."

  13. marie-lucie said,

    April 28, 2009 @ 12:32 pm

    Nobody ever thinks that their spontaneous speech or writing is unclear. "Make sure what you have written would be clear to someone who did not know the background" would be wordier but better advice.

  14. D.O. said,

    April 28, 2009 @ 1:22 pm

    Come on, if it takes only couple of seconds of confusion, it is not that much of an ambiguity. It would be a very dull world if each phrase had only one possible meaning (I will not give up Alice, no way). And it is not that clear how much IM conversation is like standard writing and how much it is more like speach. If we could only make LL bloggers comment on that.

  15. Nathan said,

    April 28, 2009 @ 1:39 pm

    @D.O.: IM or not, I think "She worked when women didn't work very much and never got married."–with its intended meaning, anyway–is the kind of sentence you would only ever see, never hear. It seems too formal for an ordinary conversational register, with the single subject and two main verb phrases. It's true that much of what happens on IM and text-messaging is less formal than other kinds of writing, but I still see a lot of sentences no one would ever speak. I think many of us are conditioned to use more formal language when we're writing, and it carries over even to IM.

  16. Rubrick said,

    April 28, 2009 @ 2:24 pm

    This isn't directly relevant to the discussion at hand, but is nonetheless directly inspired by the post. LLers often bemoan the sorry state of linguistics education, and that children are given virtually no exposure to "real" linguistics.

    Looking at these tree diagrams, a partial solution suddenly dawned on me: Mobiles! Don't those things just scream "Make me out of wire and paper and hang me from the ceiling!"?

    I guess there's a bit of a problem when the phrases swing around 180 degrees….

  17. Johanne D said,

    April 28, 2009 @ 3:06 pm

    marie-lucie,

    "Make sure what you have written would be clear to someone who did not know the background."

    That someone would include a translator! I've seen much worse. And if ever I ask the author to help me by "diagramming this sentence", I just know there's a blank stare at the other end of the phone line or e-mail. A lot of what I do is educated guesswork.To know ALL the background, I'd have to in the writer's head.

  18. Mary Kuhner said,

    April 28, 2009 @ 3:48 pm

    One of my most frustrating experiences as a scientific reviewer was taking a really opaque paragraph from a paper, painstakingly parsing it and rewriting it into clarity, and having the authors respond with "Obviously this doesn't need revision as the reviewer understood it perfectly." Since then I just write "I don't understand this paragraph" and don't try to help.

  19. marie-lucie said,

    April 28, 2009 @ 4:19 pm

    Rubrick: Mobiles! Don't those things just scream "Make me out of wire and paper and hang me from the ceiling!"?

    Some years ago when I was a graduate student in linguistics I did just that for the department's "open house", using wire coat hangers (those triangles are sometimes called that). Sure, they swung around, but that was not a real problem.

  20. marie-lucie said,

    April 28, 2009 @ 4:22 pm

    myl: I think that the "Adv" should say CP (complement clause), with "when" as COMP and "women" as NP2.

  21. Karen said,

    April 28, 2009 @ 6:24 pm

    I'm not that familiar with the program used to make those trees – it's probably possible to make them better, and label them properly, but at least the bits are separated out.

  22. comwave said,

    April 28, 2009 @ 8:44 pm

    @Mary Kuhner: Very nice to hear the same experiences. One ironic thing is that such frustrating writings are well accepted in many fields, which I guess means THEY don't read them but just scan them to see whether the letters are on paper.

  23. parse said,

    April 28, 2009 @ 9:21 pm

    "She never married and worked when women didn't work very much" is similarly ambiguous, but for some reason I have a hard time anyone would be confused by that version of the sentence. In the comment as reported I can see the confusion, though I'd be unlikely to draw the mistaken conclusion. When the phrase is reversed, I have to force myself to find the ambiguity in it, and I'm not sure why that is.

  24. Lance said,

    April 29, 2009 @ 9:51 am

    Thanks for posting! I knew I was misusing some linguistic jargon, but wasn't quite sure what the proper way to phrase my question was.

    I suppose it could be confusing for anyone who thinks there might have been a time in the recent past when women, as a rule, both did not work and did not get married. One wonders, however, how these women survived.

    Which is why I was confused. At first I thought she meant 'divorced' instead of 'married,' but then I pieced it together.

  25. dr pepper said,

    April 30, 2009 @ 4:52 pm

    Darn. The first glance at the title of this article suggested that someone had found a lost Jane Austin story.

  26. TootsNYC said,

    May 4, 2009 @ 12:22 pm

    As a copyeditor, I fix this sort of problem by flipping the phrases:

    She never got married and worked when women's didn't work very much.

    The problem is that the first "component" phrase has a preposition in it. It's safest to put those sorts of phrases as the last ones in the series.

RSS feed for comments on this post · TrackBack URI

Leave a Comment