Agatha Christie was a master storyteller who weaved together seemingly disparate clues to tell a compelling mystery. Now, scientists are trying to sleuth the details of her mental health using linguistics analyses. A new study, which looks at the works of six renowned authors over their decades’ long writing careers, found that Christie likely suffered from some form of dementia, but not Alzheimer’s.
In addition to Agatha Christie, the study looks at the writings of Iris Murdoch, Gerard Reve, and Hugo Claus, who were each diagnosed with Alzheimer’s disease, as well as P.D. James and Harry Mulisch, who served as the healthy controls. The research, led by Marjolein van Velzen of the University of Groningen in the Netherlands, extends and refines past linguistics analyses of prolific authors.
For example, a 2005 study examined the writings of Iris Murdoch, a British author whose books were consistently high quality across nearly 50 years until the end of her life, when the quality plummeted. After the publication of her poorly received last novel, she was diagnosed with Alzheimer’s disease, which was confirmed postmortem. Researchers conducted a linguistics analysis of her works and found distinct patterns of vocabulary use that they attributed to Alzheimer’s disease, including a decrease in so-called “lexical diversity” in her final works.
In the new work, published in Cortex, van Velzen and colleagues take a similar approach to this and other such studies but use a different data modeling technique to analyze the linguistic factors that indicate cognitive decline. She hopes that her work will contribute to a better understanding of dementia, specifically how it affects people’s communications skills.
Van Velzen spoke with CNS about this new study, including what they found about Agatha Christie.
CNS: Would you please define “lexical diversity”? What does it indicate about an author’s mental health?
van Velzen: Lexical diversity is the number of words we use in conversations or in writing. It’s something like our “active vocabulary.” that is the number of words we can think of while speaking/writing. As we age, we expand our vocabulary by learning new words.
For instance, doing the study of Dutch-writing author Hugo Claus, I, a Dutch native speaker, learned the word fluwijn for marten [a small weasel-like mammal]. Never having heard of a fluwijn, I looked it up in a dictionary. Doing that, I expanded my passive vocabulary: I now understand what it means. By using this word here, I’m showing that it’s not only in my passive but also also in my active vocabulary. And if I want to talk at length about martens without repeating myself, I’ll alternate the two terms — and that means I’ve increased my lexical diversity.
CNS: How does lexical diversity typically change in a healthy adult over time?
van Velzen: Well, lexical diversity tends to increase up to the age of, say, 65. A 65-year-old adult has a larger vocabulary than a 30-years old adult, and this 65-year-old vocabulary may either remain at that level, which would be normal, or may start to decrease, for instance due to a neurological condition.
CNS: How did you choose which authors to study?
van Velzen: We are interested in authors with a literary career of at least four decades, preferably five. Inspired by Peter Garrard’s study of Iris Murdoch, we looked at authors who had been diagnosed with, and/or treated for, Alzheimer’s Disease. What we saw so far in the work of these authors was a common pattern for several parameters, lexical diversity being one of the stronger indicators: As the authors’ cognitive powers declined, their vocabulary — lexical diversity — became smaller. We also looked at some healthy authors [P.D. James, Harry Mulisch] with a long career: We wanted to know whether the patterns we found were age-related or disease-related.
CNS: Your study looked at noun-pronoun ratio specifically. Why is this important?
vanVelzen: Noun-pronoun ratio is important because a noun is far more specific than a pronoun. Nouns are so-called open-class words, which means we can replace one by an other at will, and we can even invent our own nouns and add them to our vocabulary. Going back to the Dutch fluwijn example, I could use the better-known word for marten, alternating it with the low-frequency fluwijn. However, if I can’t think of either term, I’ll use “it” – and there’s no lexical diversity in “it.” That is because pronouns are closed-class words, which means this class can’t be expanded, it’s literally “closed.” “He,” “she,” “it”… we can’t add anything new there. So this is an important parameter for lexical diversity measurements.
CNS: How did you personally become interested in studying this topic?
vanVelzen: As a (neuro)linguist and neuroscientist, I’m interested in the phases that precede full-blown dementia. What happens during this stage, how do people function, what is the influence of what they learned earlier in life, the so-called cognitive reserve? How does this influence the communicative skills, which in my opinion are important in all domains of daily life?
CNS: Were you surprised by any of the results?
vanVelzen: We were surprised to see that so many of the parameters we studied showed similar patterns, whereas a few others are entirely different from what one might expect. Nouns and verbs, for instance, appear to have very different neural substrates – they seem to ‘”happen” in different parts of the brain, so that one category may be affected while the other remains intact.
CNS: Can you describe how your results for Agatha Christie differ from those in past linguistics studies? What do you think your study now suggests about Christie’s life?
vanVelzen: In 2011, the Toronto-based group led by Ian Lancashire published an in-depth analysis of the work of, among others, Agatha Christie. On the basis of their findings, they postulated that she may have been suffering from Alzheimer’s disease. Based on our own studies, however, we suspected that although there might indeed be indications of cognitive decline in Agatha Christie’s work, these pointed towards a different form of dementia. So, as a kind of neuroscientific sleuths, we started looking at the linguistic findings through a magnifying glass, modelling them, and comparing the findings with our own and others’ earlier and ongoing work, with her biographies, and with her autobiography. It’s early to say, but we think Agatha Christie may have been suffering from a different neurodegenerative condition, not from Alzheimer’s disease.
CNS: In general, what makes your linguistics analysis different from past ones?
vanVelzen: Our linguistics analysis is not all that different; rather, it is the way the patterns have been studied, calculated, and visualized. If, for instance for lexical diversity, you have a number of data points going up and down over the lifetime and you draw a line through all these points, you may end up with a flat, straight line: just the average. But if you model these data points correctly, you may end up with a surprising pattern. That is what we did in this paper, and we feel this is crucial for this type of studies.
CNS: What is the significance of your findings for daily life? Why study authors’ use of language?
vanVelzen: The significance of this work is that it aids in mapping the progress of neurodegenerative conditions in the brain. Very seldom do we get a chance to look at the development of one of the most basic components of somebody’s cognitive skills: communication. It is known that full-blown dementia has a prodromal [an early symptoms] stage of years or decades, and, through many neuroimaging studies all over the world, we are getting to know more about which networks in the brain are involved in language production. If we couple our studies of literary authors with and without disease with neuroimaging studies, which are inevitably far less longitudinal in nature, we contribute to creating a fuller picture of dementia.
CNS: What’s next for this work?
vanVelzen: We just finished a comprehensive study of the work of Hugo Claus, one of the authors modeled in our Cortex paper. We are working on more fine-grained studies of the work of Agatha Christie and other authors who show a decline in their literary output, comparing our findings with those of healthy aging authors. Next, and personally I’m very excited about this, we want to look at nouns and the various kinds of verbs: intransitive (e.g. to walk), transitive (e.g. to see something), and ditransitive (e.g. to give something to somebody). Using ditransitive verbs, we have to juggle three (pro)nouns in our head: A gives B to C. From our own and other studies so far, we know that there is something going on with verbs in neurodegenerative conditions. But so far, the kinds of verbs used in healthy aging and prodromal stages of dementia has not been quantified. Is there a shift away from ditransitives towards transitives and intransitives? This is one more parameter that might shed some light on the various forms of dementia.
-Lisa M.P. Munoz
The paper, “Data modelling in corpus linguistics: How low may we go?” by Marjolein H. van Velzen, Luca Nanetti, and Peter P. de Deyn, was published in Cortex in June 2014.