Blaming the messenger?

« previous post | next post »

R.L. writes to ask whether the phrase "a life-threatening diagnosis" is unfairly blaming the messenger.

The way to understand this common phrase, I think, is to consider the typical interpretation of phrases of the form "NOUN diagnosis". Thus a "cancer diagnosis" is a diagnosis of (the condition) cancer", not a diagnosis that is cancer. And similarly for "AIDS diagnosis", "ADHD diagnosis", "diabetes diagnosis", "asthma diagnosis", etc.

This pattern generalizes to cases where a "MODIFIER diagnosis" is interpreted as a diagnosis of a MODIFIER condition, not a diagnosis that is MODIFIER. Some examples:

The proportion of people with a psychotic diagnosis engaged in work/study in a mainstream integrated setting following intervention (53.8%) did not differ significantly from the proportion of people with a non-psychotic diagnosis (56.6%).

According to a study conducted in 2001, it takes patients an average of 10 years and four different psychiatrists to arrive at a bipolar diagnosis.

I recently learned that he is still alive, now 15 years since his terminal diagnosis.

The phrase "life-threatening diagnosis" seems to work in the same way — it's not a diagnosis that is literally itself life-threatening, but rather a diagnosis of a life-threatening condition.

This sort of thing can happen with a noun like diagnosis that has a complex corresponding semantic frame, I think — someone diagnoses some condition in someone else using some method, etc.   Modifiers can then relate semantically to one of the parts of this structure. Thus in "pathological diagnosis"or "histologic diagnosis", it's the diagnostic method that's pathological or histologic.



28 Comments

  1. David Eddyshaw said,

    November 8, 2014 @ 4:02 pm

    "Male lavatory" occasionally strikes me as a somewhat startling transferred epithet.

  2. Guy said,

    November 8, 2014 @ 5:13 pm

    It's important to remember that dependents to nonverbs are often grammatically unmarked or undermarked as to semantic roles when there are several such roles available. This is not surprising, as "words taking complex marked complementation patterns" is probably a very good syntactic definition for the class of verbs.

    This is one reason why deverbal nouns are prone to this kind of ambiguity. And nouns denoting events, processes, and relationships tend to have this issue for the same reason that verbs are often semantically associated with actions, events, and relationships. In this case, the relationship between "diagnose" and "diagnosis" is transparent.

  3. Irina said,

    November 8, 2014 @ 5:21 pm

    Schiphol Airport has "disabled toilets".

  4. Ø said,

    November 8, 2014 @ 5:58 pm

    Well, when people seem to say "The diagnosis is cancer" I suppose they might really be saying "The diagnosis is 'cancer'". It would be hard to hear the difference.

  5. Guy said,

    November 8, 2014 @ 6:08 pm

    Now I'm wondering if saying "the diagnosis is 'cancer'" is ambiguous between the person being diagnosed as having the disease cancer and being diagnosed as being afflicted with the word "cancer", albeit not suffering from the illness cancer at all.

  6. Guy said,

    November 8, 2014 @ 6:16 pm

    But more seriously, I think it's true that "diagnosis" is ambiguous between being a message, a formulation of an opinion, and an identified disease itself. That is, I think it's true that sometimes "diagnosis" doesn't denote "the messenger" at all, but rather the thing that what diagnosed – the disease identified by the "diagnosis" in the "messenger" sense of the word. Sort of like how "I saw the destruction" could mean you witnessed the event or you witnessed the state of the destroyed thing after the event.

    [(myl) That's a good point: "diagnosis" perhaps should be given a sense "the disease state identified by a diagnostic process". However, none of the dictionaries that I've consulted list such a sense.]

  7. D.O. said,

    November 8, 2014 @ 8:01 pm

    I don't believe diagnosis could mean "diagnosed condition" itself, more probably it is a form of hedging. Someone has diagnosis of some affliction could mean that probably that's what it is, but there is a room for doubt.

    What about other "messenger" words? Say, threatening report is used sometimes with the meaning of "report on something threatening".

  8. Gregory Kusnick said,

    November 8, 2014 @ 8:52 pm

    In Chinatown, when Roman Polanski tells Jack Nicholson he's a very nosy fella, perhaps that could fairly be characterized as a life-threatening diagnosis (in the "messenger" sense).

  9. Jerry Friedman said,

    November 8, 2014 @ 9:37 pm

    When does diagnosis unequivocally mean a disease state? I'm not coming up with anything.

    Only slightly sidetracking: Has LLog ever logged the remarkable rise of diagnosed with?

  10. Eric P Smith said,

    November 8, 2014 @ 9:47 pm

    To use diagnosis to mean the disease state is to confuse the measurement with the thing that is measured. I think that is quite a common confusion. A few years ago I read in BBC News Online:

    Thousands of firefighters, aided by the army, are battling several major fires, and the number of dead is expected to rise as fires are put out.

    Then don't put the fires out!

  11. Guy said,

    November 8, 2014 @ 9:54 pm

    @ Jerry Friedman

    If you also graph "diagnosed" by itself for reference, the rise looks less meteoric. A large part of the rise is apparently more discussion of medicine in the corpus or something.

    I don't know that "diagnosis" ever unambiguously refers to the disease (and I would say in the usages I'm thinking of it refers more to the abstract concept of the disease than a specific instantiation of illness in a particular person). But it seems suggested in uses like (as mentioned above) "the diagnosis is cancer", or "we have the same diagnosis". Though trying to parse out exactly what the referent is is subtle, like trying to find semantic referentiality in clauses, which can be referring to facts, to situations, to knowledge, to ideas, to potentials, etc. I don't know how useful it is to try to parse out these fine distinctions in every case.

  12. Guy said,

    November 8, 2014 @ 10:20 pm

    @Eric P Smith

    I think you're reading a causative meaning into "as" that isn't present in that sentence.

    In any event, if "diagnosis" can refer to both the evaluation and the disease, that wouldn't be confusion, although perhaps it would be confusing. "X is green" can have a meaning that conveys the color of X or it can say that X is the abstract concept of the color green. Also compare:

    She thinks she is taller than John is.
    She thinks John is taller than he is.

    In the natural interpretations of these two sentences, the first refers to the two heights she thinks they have whereas the second compares a height she imagines to a height that is real.

    Unforgivably imprecise? Perhaps, but such is human language.

  13. Jerry Friedman said,

    November 8, 2014 @ 10:30 pm

    Guy: Okay, here's an ngram of the comparison to "diagnosed". I moved the initial point to eliminate some noise at the beginning that unfortunately made the rise look less meteoric was irrelevant.

    Thanks for the comments on shades of meaning of diagnosis.

    I think Eric P Smith's point isn't his humorous post hoc ergo propter hoc but the substitution of "the number of dead" for "the number known to be dead".

  14. Guy said,

    November 8, 2014 @ 11:50 pm

    That is definitely a trend, and it teaches me I shouldn't just eyeball lines instead of looking at the actual ratio.

  15. Derwin McGeary said,

    November 9, 2014 @ 3:07 am

    Just a small observation. I have heard Russians use "У него диагноз" ("u nego diagnoz" – he has a diagnosis) to refer to people who behave oddly and seem as though they might be mentally ill . It does not imply that a professional has offered her opinion, and I think it is unambiguously using "diagnosis" to refer to the disease itself.

  16. tpr said,

    November 9, 2014 @ 4:05 am

    How can we tell whether our ability to correctly interpret what someone means by "a life-threatening diagnosis" is because "life-threatening" fulfills one of the grammatically acceptable roles of pre-nominal modifiers, or because of whatever processes allow us to recover the intent of ungrammatical utterances like those of non-native speakers and infants?

  17. Eric P Smith said,

    November 9, 2014 @ 4:42 am

    @Jerry Friedman

    You correctly understood my point: thanks for explaining it on my behalf.

  18. Emiel said,

    November 9, 2014 @ 4:52 am

    A parallel example might be "A dangerous assumption". Here it's also the thing being assumed that is dangerous. Not the assuming party.

  19. Bob Ladd said,

    November 9, 2014 @ 6:01 am

    I think diagnosed with arises from the same kind of slippery grammatical subcategorisations you get with lots of verbs. Back in the early days of generative syntax there was a lot of discussion of pairs like Bees swarmed in the garden vs. The garden swarmed with bees or He smeared paint on the wall vs. He smeared with wall with paint. Some of these involve stable alternations, but others (like the referential/semantic equivalence of older substitute X for Y with current substitute Y with X) change diachronically. It seems to me diagnose with is the result of allowing the object of the verb diagnose to include not only the disease state (They diagnosed cancer) but also the patient (They diagnosed him with cancer).

    If I believed that language change was a symptom of the decline of civilisation, that change in the grammar of substitute would be high on my list of evidence.

  20. Eric P Smith said,

    November 9, 2014 @ 9:47 am

    @Emiel: Can you explain what you mean? It seems to me that in "a dangerous assumption" it is quite literally the assumption – the assuming – that is dangerous. If I have taken a hallucinogenic drug and I jump off a cliff assuming that I can fly, it is precisely and literally my assumption that is dangerous.

  21. Guy said,

    November 9, 2014 @ 12:33 pm

    @tpr

    The main test of grammaticality s usually frequency of usage and acceptability to a majority of speakers in the relevant language community. But keep in mind that even if it could be shown that this isn't an acceptable role of an attributive modifier, it could still be grammatical by virtue of being a complement licensed by "diagnose" – like "criminal" as in "criminal lawyer"

    @Eric P Smith

    I'm afraid I still don't grasp the thing that bothers you about that. If the fires are still raging, as they must be during the time indicated by "as the fires are put out", shouldn't we expect the number of dead people as well as the number of people known to be dead to rise?

    Or are you working with an ambiguity illustrated between "the number of dead is rising" and "a number of dead are rising"?

  22. David Donnell said,

    November 9, 2014 @ 7:41 pm

    Am I wrong in seeing this as a sort of 'inverse example': when a TV news voice-over announces "The weather has been brought to you by such-and-such corporation." (Dude, NO; the weather was brought to us by nature!)

  23. Lars said,

    November 10, 2014 @ 4:29 am

    @Derwin McGeary – in Sweden the situation is the reverse, in a sense. At least in a school setting.

    In any given classroom there will be a number of kids that the generally poorly trained and motivated teachers find it hard to handle (because they behave as kids normally do). But if one of them actually, officially 'has a diagnosis' — OCD, ADD, ADHD, Aspergers, dyslexia, you name it, and all the kids know it, but school employees aren't allowed to discuss it with the other parents — well, in that case the school gets money for 10 assistant teacher hours per week (of which 8 are spent on the other kids).

    So from the teacher's viewpoint 'having a diagnosis' sort of means 'no more annoying than the rest, but s/he has an excuse and s/he is someone else's problem now'.

  24. tpr said,

    November 10, 2014 @ 5:13 am

    @Guy

    The main test of grammaticality…

    "A life-threatening diagnosis" certainly passes the kinds of grammatical tests that "colorless green ideas" passes, but more relevant is the acceptability of the intended interpretation. My question was about how we can tell whether we deduce the intention of the speaker via an established rule for interpreting pre-nominal modifiers or by falling back on strategies that aren't really part of language at all, the ones that help us to deduce intentions on the basis of gestures, looks, grunts and from utterances in languages we don't speak.

    Perhaps a good example is when a non-native speaker of English says something like "She is hardly working" when it's clear from context that they mean "She is working hard". A person familiar with this kind of error will be able to recover the correct intention, but it's still an error. Both sentences are grammatical, so the error is in what the speaker thinks the conventional interpretation of "hardly working" is.

  25. Michael Watts said,

    November 10, 2014 @ 9:58 pm

    I feel certain that there's a standard term in poetry for this kind of construction, where a description is syntactically applied to a particular noun but understood as actually describing a different noun. Here's an example from Kipling:

    As I pass through my incarnations, in every age and race
    I make my proper prostrations to the gods of the market-place
    Peering through reverent fingers, I watch them flourish and fall

    Obviously, the fingers aren't reverent; the narrator, or the pose he takes, is. I'm almost sure I learned the term for this from my Latin teacher years ago, but I can't call it to mind.

  26. Jerry Friedman said,

    November 11, 2014 @ 12:55 am

    Michael Watts: "Transferred epithet", as David Eddyshaw said? "Hypallage"?

  27. Michael Watts said,

    November 11, 2014 @ 3:08 am

    That would be it. Thanks! Anyway, (a) it seems a little weird to call this specific instance of it "unfairly blaming the messenger", since the same thing happens all the time in all sorts of positive and negative contexts, and (b) it seems a little weird to write a whole post on the question without even mentioning that it's a well-known effect. Focus on the particular word "diagnosis" is misplaced (IMO); "fingers" doesn't exactly have a complex semantic frame.

  28. Michael Watts said,

    November 11, 2014 @ 3:10 am

    For a different example, what is the semantic frame allowing transferred epithet in the offer "have a quiet cup of tea"?

RSS feed for comments on this post