Content uploaded by Ian Lancashire
Author content
All content in this area was uploaded by Ian Lancashire
Content may be subject to copyright.
1
Vocabulary Changes in Agatha Christie’s Mysteries as an Indication of Dementia:
A Case Study
Ian Lancashire* and Graeme Hirst†
University of Toronto, *Department of English and †Department of Computer Science
Ian.Lancashire, Graeme.Hirst @utoronto.ca
Introduction
Alzheimer’s disease leads to changes in language production at all levels — lexical, syn-
tactic, and discourse — that are different to or markedly greater than those observed in
normal aging (Maxim & Bryan 1994). For example, whereas “the lexicon continues to
expand indefinitely until death or illness intervenes”, the “semantic and then
phonological output lexicon” becomes progressively “inaccessible” in Alzheimer’s dis-
ease (Maxim & Bryan 1994: 3, 24). And while in healthy aging, semantic retrieval speed
deteriorates and hence “the number of ‘indefinite’ words may increase” (Maxim & Bryan
1994: 46), as may the number of repeated phrases, Nicholas et al. (1985) demonstrated
that both indefinite words and repetitions occur significantly more often in the language
of Alzheimer’s patients than in that of healthy people of similar age and level of educa-
tion.
These facts suggest that an assessment of dementia might be based in part on an analysis
of a diachronic corpus of writing by the patient. Garrard et al. (2005) compared three
works by the British novelist Iris Murdoch, whose diagnosis of Alzheimer’s disease was
confirmed post mortem. Her final novel, which was written during her decline, had a
much smaller vocabulary than novels from her early and middle years. In addition, a
small sample suggested that her sentences were syntactically simpler.
Here, we analyze the vocabulary of the British mystery writer Agatha Christie, who, al-
though never diagnosed, was also believed to have suffered from dementia in her final
years even as she continued to write. Our analysis, on a much larger and more-
representative corpus than that which Garrard et al. used, concentrates on vocabulary-
richness measures and opens a project that will also look at syntactic and discourse-level
aspects of her texts.
Background
Agatha Christie (1890–1976), in a 53-year writing career, crafted about 85 novels and
plays. Bodley Head published her first novel, The Mysterious Affair at Styles, in 1920,
and Collins her last, Postern of Fate, in 1973. The public had bought 400 million copies
of her works by 1975 and 2 billion by 1990. Her typical detective novel gives readers a
Presented at the 19th Annual Rotman Research Institute Conference, Cognitive Aging: Research and Prac-
tice, 8–10 March 2009, Toronto,
2
problem to be solved. Conversational dialogue and spare narration carry the reader along.
She provides clues and diversions and maintains suspense until her sleuth unveils a usu-
ally elusive solution that ends the book. Her novels each contain between 55,000 and
75,000 words. Her custom, until the last books, was to work out the plot meticulously
beforehand in a notebook, and to write the last chapter—where her detective laid out the
solution—first. Her notes on Lord Edgeware Dies and Evil under the Sun “are almost
identical to the finished article” (Thompson 2007: 369).
Janet Morgan, a biographer trusted by the Christie family, says that after Elephants can
Remember, written when Christie was 81, “her powers really declined” (Morgan 1984:
370). When subsequently writing Postern of Fate, she reportedly found it “harder than
ever to concentrate”: this last book “nearly killed her” (Morgan 1984: 371). Her preoc-
cupation with old people and their memories in both Elephants can Remember and Pos-
tern of Fate reflects more on her personal circumstances than on crime, murderer, and
clues. Readers have complained about inconsistencies in character and plotting in both
these late works. Much of Postern digresses into Christie’s past memories and current
problems, and the murderer is an afterthought. Her agent directed her to editorial help,
and her husband Max and her secretary, Mrs Daphne Honeybone, “tidied it up”:
Christie’s daughter Rosalind then asked Collins “to press for no more books”. Morgan
concluded, “Physical and mental decline is sad” (pp. 371–72). By the time of Elephants
can Remember, Christie had aged considerably, having fallen and broken a hip (Thomp-
son 2007: 464, 473–74). Four years later, friends reported her thin and “frail”; she had
angry fits (in one she cut off all her hair) and did not always make sense in conversation
(Thompson 2007: 483). Although she was never assessed for dementia, her last novels
reveal an inability to create a crime solvable by clue-detection according to the rules of
the genre that she helped to create.
Material and Methods
Fourteen Christie novels written between ages 34 and 82 were digitized, and digitized
copies of her first two mysteries, The Mysterious Affair at Styles (age 28) and The Secret
Adversary (age 32), were taken from Project Gutenberg. After all punctuation, apostro-
phes, and hyphens were deleted, each text was divided into 10,000-word segments. The
segments were then analyzed with the software tools Concordance and the Text Analysis
Computing Tools (TACT).
We performed three analyses of the first 50,000 words of each novel:
1. Like Garrard et al. (2005), as a simple measure of vocabulary size and richness,
we counted the number of different words used.
2. As a second measure of vocabulary richness, we counted the number of different
maximal phrase-types (i.e., word n-grams) that were repeated. These are defined
by word-length and frequency. For example, if in a given text we saw 5 occur-
rences of “all sorts of things” and 7 occurrences of both “all sorts of” and “all
3
sorts”, we would count this as two repeated phrase-types, not three, because all
occurrences of “all sorts” are contained in the longer phrases.
3. We counted the number of occurrences of the vague, indefinite words “thing”,
“anything”, and “something”.
Results
Table 1 displays total counts for vocabulary size and repeating phrases, and the percent-
age of words that are indefinite nouns in the first 50,000 words of each novel.
Novel
Age at com-
position
Word-
types
Repeated
phrase-types
Indefinite
words (%)
Styles
28
5027
7623
0.27
Adversary
32
5576
7320
0.39
Ackroyd
34
4833
7905
0.48
Orient
43
4692
8184
0.30
Appointment
47
4985
8071
0.39
Curtain
50
5131
7846
0.43
Towards Zero
51
4941
7680
0.50
Announced
59
5181
7596
0.45
Destination
63
5442
7426
0.36
Ordeal
67
4440
8221
0.58
Clocks
72
4827
8014
0.61
Endless
76
4159
8559
0.78
Frankfurt*
79
5583
7418
0.71
Nemesis
80
4631
8103
0.66
Elephants
81
3762
8821
1.02
Postern
82
4275
8249
1.23
*A thriller (not a mystery) that was written with the help of book research.
Table 1. Counts for vocabulary (word-types) and repeating phrases, and per-
centages of indefinite nouns in the first 50,000 words of 16 Christie novels.
Vocabulary size. The richness of the vocabulary of Christie’s novels declines with her
age at composition. The three novels that she wrote in her 80s, Nemesis, Elephants, and
Postern, have a smaller vocabulary than any of the analyzed works written by her be-
tween ages 28 to 63. Word-types in the first 50,000 words of her novels fall by one-fifth
between ages 28–32 and 81–82. Elephants Can Remember, written when she was 81,
exhibits a staggering drop in vocabulary, almost 31%, compared with Destination Un-
known, written 18 years earlier. Some 15,000 words shorter than Nemesis and Postern of
Fate, which preceded and followed it, Elephants appears to register the onset of a pro-
found writing block. Possibly Christie’s broken hip, the year before, was a factor. A lin-
ear regression on the decline in vocabulary with age approaches significance [F(1,14) =
4
3.95, p = .066], and is highly significant when the outlier, Frankfurt, is removed (see dis-
cussion below) [F(1,13) = 9.80, p < .01].
Repeated phrases. The number of different repeating phrase-types in the first 50,000
words in Christie’s novels increases with age, again implying a decline in the lexical rich-
ness of her writing. The increase with age approaches significance [F(1,14) = 4.06, p
= .064], and again is highly significant when the outlier, Frankfurt, is removed [F(1,13) =
8.47, p < .015].
Indefinite words. Christie’s use of vague, indefinite “thing” words increases signifi-
cantly with age from 0.27% of her word-count in Styles (1920) to 1.23% in Postern
(1973) [F(1,14) = 22.6, p < .0005]. Frankfurt is not an outlier in these data, and exclud-
ing it makes very little difference to the analysis.
Discussion
Her family’s testimony about Christie’s otherwise undiagnosed physical and mental de-
cline offers an explanation for these data: encroaching dementia, as in the case of the
English novelist Iris Murdoch that Garrard et al. (2005) studied. Our analysis suggests,
in addition, that repeating phrases and, in particular, indefinite-term usage (not used by
Garrard et al.) are significant markers.
Outlier. Passenger to Frankfurt has the largest vocabulary of all the works we analyzed.
Unlike Christie’s other works, it is a thriller, not a detective mystery, conceived, written,
and researched in her early to mid 70s. Subtitled “An extravaganza”, it draws on books
by political thinkers that she requested of her publishers. On receiving her manuscript,
they were doubtful about bringing it out because it differed so much from her detective
fiction. Much of the vocabulary in Passenger to Frankfurt comes from her reliance on
these sources. We therefore exclude it as an outlier from our tests for vocabulary rich-
ness. Nonetheless, we observe that it was not an outlier with regard to indefinite words.
Ongoing work. We will next analyze Christie’s texts for known syntactic and discourse-
level characteristics of Alzheimer’s language. For comparison, we will also carry out
parallel analyses of the works of several writers who are not suspected of dementia in old
age.
Conclusion
While few present-day patients have a large online diachronic corpus available for analy-
sis, this will begin to change as more individuals begin to keep, if only by inertia, a life-
time archive of e-mail, blogs, professional documents, and the like. While the diversity
of topics and genres in such an archive brings methodological problems to the analysis
(as observed with literary genre in Passenger to Frankfurt), we can nonetheless foresee
the possibility of automated textual analysis as a part of the early diagnosis of Alz-
heimer’s disease and similar dementias.
5
References
Garrard, P., Maloney, L.M., Hodges, J. R., & Patterson, K. (2005). The effects of very
early Alzheimer’s disease on the characteristics of writing by a renowned author. Brain,
128.2, 250–60.
Maxim, J. & Bryan, K. (1994). Language of the Elderly: A Clinical Perspective. Lon-
don: Whurr.
Nicholas, M., Obler, L. K., Albert, M. L., Helm-Estabrooks, N. (1985). Empty speech in
Alzheimer’s disease and fluent aphasia. Journal of Speech and Hearing Research, 28:
405–10.
Morgan, J. (1984). Agatha Christie: A Biography. London: Collins.
Thompson, L. (2007). Agatha Christie: An English Mystery. London: Headline Review.
