Early Alzheimer's signs in Reagan's speech
« previous post | next post »
Lawrence Altman, "Parsing Ronald Reagan’s Words for Early Signs of Alzheimer’s", NYT 3/30/2015:
Even before Ronald Reagan became the oldest elected president, his mental state was a political issue. His adversaries often suggested his penchant for contradictory statements, forgetting names and seeming absent-mindedness could be linked to dementia.
In 1980, Mr. Reagan told me that he would resign the presidency if White House doctors found him mentally unfit. Years later, those doctors and key aides told me they had not detected any changes in his mental abilities while in office.
Now a clever new analysis has found that during his two terms in office, subtle changes in Mr. Reagan’s speaking patterns linked to the onset of dementia were apparent years before doctors diagnosed his Alzheimer’s disease in 1994.
The paper in question is Visar Berisha, Shuai Wang, Amy LaCross & Julie Liss, "Tracking discourse complexity preceding Alzheimer's disease diagnosis: a case study comparing the press conferences of presidents Ronald Reagan and George Herbert Walker Bush", Journal of Alzheimers Disease 2015:
Changes in some lexical features of language have been associated with the onset and progression of Alzheimer’s disease. Here we describe a method to extract key features from discourse transcripts, which we evaluated on non-scripted news conferences from President Ronald Reagan, who was diagnosed with Alzheimer’s disease in 1994, and President George Herbert Walker Bush, who has no known diagnosis of Alzheimer’s disease. Key word counts previously associated with cognitive decline in Alzheimer’s disease were extracted and regression analyses were conducted. President Reagan showed a significant reduction in the number of unique words over time and a significant increase in conversational fillers and non-specific nouns over time. There was no significant trend in these features for President Bush.
Some relevant past posts:
"Jackson's Dilemma and Alzheimer's", 12/2/2004
"Writing style and dementia", 12/3/2004
"Nun study update", 8/27/2009
"Literary Alzheimer's", 11/13/2009
"Authorial Alzheimer's again", 12/15/2009
In this context, it's really too bad that the Alzheimer's Disease Neuroimaging Initiative didn't include any linguistic material — like a picture description task, or answers to a set of scripted questions — in its design.
That was major mistake, in my opinion, since a speech task would have been a tiny, low-cost, low-impact addition to the massive expensive and time-consuming (though no doubt valuable) tests that were designed into the project. But it might turn out that a minute of appropriately designed speech, collected at each visit, would have turned out to be a better predictor of the future time course of symptoms than the brain imaging and body-fluids testing. And the cost of finding out would have been negligible.
Michael Tinkler said,
April 12, 2015 @ 7:06 pm
My mother is going through some of these vocabulary changes caused by cognitive change. She's 81, so this isn't particularly early (except in comparison to her own mother, who was sharp as a tack at 97 when she refused to eat or drink any more). I wonder now how much longer this has been going on than I have perceived. She is usually quick and ready in circumlocutions to avoid words and names she doesn't remember, so long as she has started the conversational topic. When I shift topics her difficulty in recalling nouns and names (never verbs?) is more noticeable.
GeorgeW said,
April 12, 2015 @ 7:42 pm
From the NYT article: "Earlier studies have shown that certain linguistic biomarkers change with disease progression. Spoken vocabulary size declines, for instance, and use of indefinite nouns increases."
Use of indefinite nouns increases? Hmm. Why would that be?
[(myl) I think that they mean "non-specific nouns" (e.g. "animal", "thing", "stuff", "place"), not "indefinite nouns" (e.g. "an elephant", "a pencil case", "some cocoa", "a peninsula"). And the reason would presumably be because the more specific nouns don't come to mind quickly enough.]
Mark Riley said,
April 12, 2015 @ 9:26 pm
This may well be true, but don't forget the practically universal media meme that Democratic presidents are philosopher kings while Republican presidents are crooks and/or morons. I am old enough to remember the ridicule of Pres Eisenhower for his wandering syntax – today he would be diagnosed with something other than his devious intelligence. All this makes me skeptical of this report's validity. As has long been known: You can often predict the results of a social scientist's research by knowing how he voted in the last election.
[(myl) Either you're not paying attention, but just lashing out reflexively; or else you're a troll, in the technical sense of the term.
(1) The authors of this study are clinical researchers, not social scientists.
(2) The point of their article is to compare two Republican presidents, Ronald Reagan and George Herbert Walker Bush, one of whom was later diagnosed with Alzheimer's, while the other was not.]
Rebecca said,
April 12, 2015 @ 10:01 pm
That take on non-specific nouns certainly matches my experience with my mom. She went through a period where it was clear she had a specific thing in mind, but would describe it rather than name it. At first it might be something somewhat technical, like a specific medicine, then it seemed to move on to ordinary words that she learned relatively late in life, like "cell phone" or "email", until finally she would often have to describe a very common word: I remember her struggling with both "key" and "lock" and end up saying something like "do you have my thing that you turn so the door won't open?" In some ways, it seemed cognitively more difficult to come up with the circumlocutions
Maria said,
April 13, 2015 @ 3:17 am
This is interesting, but Alzheimer is a specifical disease, since dementia is much more large. I think a lot of people, after 80 years old, have to deal with problems in lexical access, and therefore they show "a significant reduction in the number of unique words over time and a significant increase in conversational fillers and non-specific nouns over time". I have such exemples in my family (85-95 years old, great difficulty with specific nouns, but no other signs of Alzheimer's desease).
I think such a linguistic test could find a very great number of false hits. And this would be very anxiogenic for the persons who take the test…
Rubrick said,
April 13, 2015 @ 5:08 am
The findings are certainly plausible, but it seems odd to use only one other president — and a single-term one at that — as a control. An analysis of the last, say, ten presidents would make a far more convincing case that Reagan was actually an outlier.
[(myl) This is a good point. Ronald Reagan was an interesting target of opportunity, given his later diagnosis and the existence of lots of press conference recordings. But presidents are not a great data set in any case, for many reasons: N=10 is better than N=2, but not a lot better; they are not matched for age; the nature of each president's relationship to the press varies a lot across time (which is likely to influence fluency); etc.
What we really need is a large longitudinal study of ordinary people in (say) the 60-80 age bracket, using a simple but appropriately-designed task. This would be really easy and cheap — all that would be required would be a minute or two of audio recording taken at each regularly-scheduled checkup, pooled across several doctors/hospitals/clinics (with appropriate informed consent for data sharing), transcribed and aligned centrally after the fact, etc.
These days, this could be done with a smartphone app, which would automatically send the data to a central repository, along with the date and the doctor and patient IDs. The per-visit costs would be essentially zero, and the pooled analysis costs (i.e. transcription and alignment) would be a few dollars per patient-visit.
Then we'd have N=1000 or N=10,000 or N=100,000, and we could really see what's happening.
Given the Nun Study (as well as common-sense experience), this should have been an obvious thing to do twenty years ago (though then you'd have needed a PC application along with distribution of a standard microphone set-up, which maybe would be the right thing to do now anyhow).
Why hasn't it been done? As far as I can tell, it's because biomedical researchers (and biomedical research funders) only respect research methods that involve expensive equipment and/or tissue samples or bodily fluids. Maybe if we designed a multi-million-dollar Biomedical Audio Recorder, uh I mean Kilocycle-range Near-field Ambient Pressure Sensor (KNAPS)… Hmm, yes, starting with a microphone that works by acoustic perturbation of a spatial array of proton beams, sensed at a distance by Superconducting Quantum Interference Devices, and…]
maidhc said,
April 13, 2015 @ 5:35 am
I know people who knew Ronald Reagan. Apparently in person he was a very warm person who could put anybody at ease and make them feel at home.
Some of his statements, such as that he had been present at the liberation of the WWII concentration camps, or that German SS troops were victims of WWII, were probably intended in this light. They were just casual remarks to establish a friendly atmosphere. However he seemed to forget that, as President, his every public utterance was recorded and preserved for posterity.
I don't know whether this lack of awareness of context is characteristic of Alzheimer's or not. Also I'm a bit dubious about this study anyway. I don't recall Reagan's behavior changing noticeably over his term in office. Yes, he was non-specific. He was that way back in 1980 too. Of course I haven't done a study myself to back up my recollections.
I think that speculating about the mental state of historical characters is mostly useless. If we know what they did and said, we know how they affected the course of events.
Bloix said,
April 13, 2015 @ 5:45 am
Mark Riley – the reason to use Reagan and Bush for this is that presidents produce a very large volume of unscripted, extended speech that is readily available without charge.
BlueLoom said,
April 13, 2015 @ 7:19 am
Anyone who has walked down the Alzheimer's path with an aging parent knew well before (meaning several years before) the public announcement of Reagan's condition that something was very wrong. Presidents should not be the judge of their own fitness to serve. Some kind of objective panel (assuming anyone in our polarized society can be deemed objective) of neurology experts should be making this decision.
Jerry Friedman said,
April 13, 2015 @ 12:00 pm
BlueLoom: The article says Reagan left the decision up to his doctors, as I agree he should have.
Maria: Even if a study of non-specific nouns, conversational fillers, and vocabulary doesn't lead to a test for Alzheimers, I think it would be interesting and could well be useful.
I'm ashamed to say that I initially misunderstood MYL's "low-cost, low-impact". I'm so used to seeing "impact" with a positive meaning that I thought he meant "high-impact". But of course "impact" originally meant something harmful, like a collision, and Google tells me that "low-impact test" is used in medicine to mean a test that does little harm to the patient.
GH said,
April 13, 2015 @ 12:29 pm
@maidhc:
How would you design such a study? I would suggest that it might be something very much like the one that was actually done: looking for changes in his speech patterns, and comparing with another president as a control. And the fact is that they did find such changes.
That you don't recall having noticed any changes means next to nothing. Would you expect the average person to be good at intuitively estimating the number of unique words in someone's speech? Or to notice gradual, subtle changes in this and other speech patterns (particularly when the tendencies in question were already characteristic of the speaker)?
And even if you think you could do so in principle, consider the confounding factors: a lot of the time when you hear the president speak, it's from a script, and what you hear of unscripted remarks (as broadcast on news programs, for example) are very often selected sound bytes, which are most probably not representative.
J. W. Brewer said,
April 13, 2015 @ 1:12 pm
Interesting that Bush (sr.) did twice as many press conferences in only half as long in office, so was doing four times as many press-conferences/year. That makes the Bush dataset rich enough for the finding of lack-of-trend there to feel more reliable, but also makes one wonder if the sparser-by-comparison Reagan dataset might be thin enough to make the appearance of a trend less reliable? (Also note that GHW Bush was a little bit younger when he left office than Reagan was at his first inauguration, so there's no overlap at all there.)
J. W. Brewer said,
April 13, 2015 @ 1:29 pm
Another potential political subject for study – Margaret Thatcher generally withdrew from public speaking (scripted or un-) in 2002, after a series of small strokes (a perhaps less-stigmatizing explanation for communicative difficulty than Alzheimer's/dementia?), but her daughter claims to have seen visible evidence of memory loss/Alzheimer's several years prior to that. She had not been Prime Minister for quite some time, but there might still be some meaningful corpus of recordings/transcripts of unscripted remarks by her (in interviews and/or maybe speaking in the House of Lords) covering the decade-plus after she left Downing Street but before she ceased public speechifying that might or might not show a trend on these lexical markers.
Jonathan Mayhew said,
April 13, 2015 @ 3:25 pm
A family member in her 50s has "semantic dementia" (not Alzheimers). One of the noticeable features is the loss of specific nouns and proper names, but general nouns like "thing," "person" or "guy" are a mainstay of her vocabulary. She'll say "my son" but not know his name. It's clear the brain retains those more general categories in many of these conditions while losing very specific terms.
David Morris said,
April 13, 2015 @ 3:33 pm
They seem to be hedging their bets with Bush: "who has no known diagnosis of Alzheimer’s disease".
Haamu said,
April 13, 2015 @ 5:41 pm
Fibromyalgia "fog" has a similar linguistic manifestation, although perhaps more intermittent. My wife (early 50s) has the condition, so I'm quite familiar with the loss of vocabulary and circumlocutions. The symptoms come and go. She recently had something of a remission lasting a couple of years, during which she was almost always her familiar, brilliant self, but lately it's been coming back a lot more often.
Linguistic evidence seems like a promising avenue, but, like @Maria, I'd want us to be very careful about how we use it and how we distinguish possible causes.
quixote said,
April 13, 2015 @ 6:06 pm
I remember Reagan's time in office somewhat clearly, and in my recollection there was an obvious decline in speed of response and verbal ability. When he campaigned in 1980 he was your usual quick-on-his-feet politician. From his second term there's a clip, I think it's on youtube, of him talking with Queen Elizabeth at some state function. He's not only slow and bumbling, but what he says to the Queen is completely off in the ozone, just like every dementia patient I've ever seen. (Of course, I don't remember the specifics. But it's not Alzheimer's, I swear!)
The Queen, being the Queen, passes right over the blunder like a stately ship, and Reagan's handlers at his side quickly jump in and start talking.
I don't know when the clip was released, but I know I first saw it in the 2000s. I don't know if it was carefully not published in 1986 or whenever it was, or if the media colluded in protecting his image to that extent.
Chips Mackinolty said,
April 13, 2015 @ 6:13 pm
Um, just wondering. There is at least one former president of the US for which there are hundreds of hours of unscripted audio recordings. Has anyone thought to undertake this study over the years Nixon was in power?
Acilius said,
April 14, 2015 @ 8:37 am
@myl: I think you're coming down too hard on Mark Riley in your note on his comment above. It's way out of proportion with his remarks to respond with "Either you're not paying attention, but just lashing out reflexively; or else you're a troll, in the technical sense of the term."
Mr Riley does not reject the study ("This may well be true,") but cautions about the power of partisan bias in shaping researchers' perceptions of political leaders. That power that can distort studies in some very subtle ways, quite outside the conscious intent of the researcher. Frankly, the fact that this study was conducted by clinicians rather than political scientists or others in the social sciences which Mr Riley disparages gives me some pause, since people trained in the social sciences are usually taught to look keep an eye on their political biases and are unlikely to defend a result that suits those biases in a simplistic way.
For example, I know that my own bias against Reagan is strong enough that it would lead me to be very aggressive in searching for evidence to validate this thesis were I to be part of this research team. If I were presented with two ways of transcribing an utterance of Reagan's, one of which attributed words to him that fit the thesis and another which attributed words that did not, I would, if left to my own devices, choose the words that fit the thesis every time. And I wouldn't even realize I was doing it, that the other interpretation was possible.
I'm not saying that this research team has done this; I'm sure they haven't, in fact, as that would be a remarkably crude error to overlook. But there are any number of more subtle ways of achieving the same result, and until the study has survived challenges from other researchers I think that there is room for reminders like those Mr Riley offers.
[(myl) OK, maybe I should have been more diplomatic. But his objection was a reflex accusation of political bias, along lines clearly refuted by facts stated in the original post.
Your suggestion is a more subtle one, which requires reference to the published paper in order to verify that the authors did not have the opportunity to exercise any problematic judgment in preparing their data. They started from the "official archives of presidential transcripts", presumably from the presidential libraries, and then
To generate the corpus for analysis, we downloaded each transcript and omitted the prepared statement by the president and all questions/statements by other individuals. We filtered all annotations in brackets. Annotations refer to addendums to the transcripts by the editors to provide context (e.g., [Laughter]). Only answers to specific questions remained.
Based on my experience with such transcripts, they probably under-represent the disfluencies, which are usually partly edited by the normal conventions of professional transcription.
Skepticism about various sorts of political and cultural bias is always a good idea; but before muddying the waters with reflex accusations of bias, it's also a good idea to put a few tens of seconds into checking the facts of the case.]
Acilius said,
April 14, 2015 @ 5:37 pm
@myl: Sure, all well-taken. Not to belabor an already-settled point, but I feel I would be remiss if I did not quote a sentence from the abstract to which you linked in your response: "we discover that antisocial behavior is exacerbated when community feedback is overly harsh."
Jimbino said,
April 16, 2015 @ 4:41 pm
Funny, I always considered Reagan a careful speaker of English. He claimed no advance degree, but I never heard him utter solecisms like Clinton's "with Hillary and I" or Obamas's "The problem is is [sic] that."
Nor did he use "absolutely" for "yes" ad nauseam.