ALZHE1MER DISEASE AND ASSOCIATED DISORDERS, 2(1):38-49,1 988 f' 1988 BY THE WESTERN GERIATRIC RESEARCH INSTITUTE

PROVERB AND IDIOM COMPREHENSION IN ALZHEIMER DISEASE

DANIEL KEMPLER, PH.D. " ', DIANA VAN L.~NCKER, PH.D.',

AND STEPHEN ReAD, M.D. '

Abstract. Twenty-nine patients diagnosed with Probable Alzheimer Disease were ad­ministered tests of word. familiar phrases (idioms and proverbs), and novel phrase compre­hension. From the early stage of the disease, patients performed worse at understanding familiar phrases than single words or novel phrases. The results uphold common observations that AD patients have difficulty interpreting abstract meanings. Cognitive variables respon­sible for poor idiom/proverb comprehension and the clinical implications ofthis new protocol are discussed.

Key Words. Language, Proverb. Idiom, Alzheimer

Language deficits appear early in the course of Alzheimer Disease (AD), and worsen along with other cognitive impairments (Bayles and Kaszniak, 1987; Cum­mings and Benson, 1983; Obler and Albert, 1981). Although onset and pattern of language disturbance varies across individuals (Wechsler, 1977), anomia is usually an early symptom, seen in poor performances in word-list generation and confrontation naming (Bayles and Tomoeda, 1983; Cummings et aI., 1985). By the mDderate stages, other aphasic symptoms appear, including paraphasias, im­paired narrative writing, and comprehension deficits (Appel et aI., 1982; Cum­mings et aI., 1985). It is now well established that control over meaning deteri­orates prior to ability to use grammatical forms, which is relatively preserved through the moderate stages of the disease (e.g. Bayles, 1979; Kempler, 1984; Kempler et aI., 1987; Schwartz et aI., 1979). Finally, reiterative disturbances (echolalia and palilalia) characterize the late stages (Cummings, 1982).

Pro'rerb and idiom interpretation, although not yet, to our knowledge, system­atically studied in this population, is a standard part of routine mental status testing (e.g. Strub and Black, 1985), and is often included in dementia screening evaluations. AD patients often produce concrete and otherwise inappropriate interpretations of proverbs such as "Rome wasn't built in a day" and idioms such as "loud tie." An ability to interpret these samples offigurative language is thought to reflect quality of thinking, concept formation, and abstract verbal reasoning (Lezak, 1983). Concrete interpretations are assumed to be symptomatic of loss of abstract ability, and are frequently observed in patients with widespread brain disease of any etiology. In an informal manner, an examiner asks the patient to interpret a proverb or idiom verbally (e.g. Wechsler, 1955), in writing, or in a verbal multiple choice format (Gorham, 1956). The patient responds, and the examiner determines whether the interpretation is abstract ("It takes time to do things well"), semiabstract ("Don't do things too fas!"), or concrete ("It took a long time to build Rome") (Strub and Black, 1985, page (31). If the patient produces one of these responses, the results are relatively easy to interpret.

: Scho,)l of Gerontology, University of Southern California, University Park, Los Angeles, CA 90089'() [91, USA.

: Neuropsychology Program. Department of Psychiatl"j and Biobehavioral Science, UCLA School of Medicine. Los Angeles, CA 90024, USA.

J Neuf.lbehavioral Unit, West Los Angeles Veterans Administration Medical Center, Los Angeles, CA 900;3. USA.

4 Correspondence should be sent to Dr. Daniel Kempler.

38

I

This type 0

plication. Veri into one cate! generally not c from AD pati responses to c paraphasias al because the pa or because the ported abstrac digms, are dif memory limita are available fi groups.

Recognizing and its import< and Novel La: comprehensior. tences. The pn terpretation ta! objective, quar deficient perf or the selected stir and idioms an whereas novel, tive grammatic performance of can be compan to gauge the c( proverb and idi

Three tasks were proverbs), and nOve (see Appendix). Sinl and to establish a overlearned e:<press depicted in a simple ph rases in length. w(

spoke C'8ch stimulus The subjects were el cases. a response wa

For the familiar p representation of 1 v­or related to the idi( literal foils (e.g. a m. included because litt

, We use the tenn eluding proverbs, idi sentences while otbe: as "phrases" througt

, . ,

lse were ad­rase compre­Ilderstanding observations .. bles respon­new protocol

ase (AD), and k, 1987; Cum­I:t and pattern ' 7), anomia is teneration and aI. , 1985). By raphasias, im­, 1982; Cum­

leaning deteri­vely preserved empler, 1984; " disturbances tI82). ,ledge, system­mental status

~ntia screening inappropriate d idioms such

uage is thought 'rbal reasoning Jomatic of loss despread brain • the patient to ~Iri ting, or in a Jonds, and the Ikes time to do ete ("It took a If the patient

interpret.

Los Angeles, CA

lice, UCLA School

m ter, Los Angeles,

KEMPLER ET Al.: IDIOM COMPREHENSION 39

This type of testing, however useful, often presents problems in practical ap­plication. Verbal responses, particularly from AD patients, often do not fit neatly into one category. For instance, comments and tangential remarks, which are generally not clear attempts at answering the question, are common response types from AD patients. It is often unclear how much to attribute their inadequate responses to comprehension deficits or to production problems, which include paraphasias and word finding difficulties. Responses can be judged to be poor because the patient chose the wrong word, did not respond to the question asked, or because the response is uninterpretable, and not necessarily because of a pur­ported abstraction deficit. Some formats, including verbal multiple choice para­digms, are difficult for AD patients because of poor attention and short-term memory limitations. In addition, for most formats, no reliable age-matched norms are available for comparison, nor are there comparative data from other clinical groups.

Recognizing the widespread use of proverb interpretation in clinical screening and its importance as a measure of cognitive intactness, we designed the Familiar and Novel Language Comprehension (F ANL-C) test to evaluate formally the comprehension of proverbs and idioms, in comparison to newly generated sen­tences. The protocol has several advantages over typical proverb and idiom in­terpretation tasks. Since only a picture pointing response is required, results are objective, quantifiable, and verbal production abilities are not involved. Thus, deficient performance can be interpreted more clearly, because the properties of the selected stimuli presented for a comprehension task can be specified: Proverbs and idioms are unitary, fixed expressions associated with complex meanings, whereas novel sentences are a sequence of words combined according to produc­tive grammatical rules. Another advantage of using a comprehension test is that performance of clinical groups can be compared to normal controls and the results can be compared directly with comprehension of non-idiomatic (novel) phrases to gauge the contribution of memory, attention, and grammatical abilities on proverb and idiom comprehension.

MATERIALS AND METHOD

Three tasks were designed to assess comprehension of single words, familiar pbrases3 (idioms and proverbs), and novel language. Ten nouns, 10 familiar stimuli, and 10 novel stimuli were selected (see Appendix). Single word comprehension was tested to ascertain that the subject could do the task and to establish a baseline. Familiar stimuli met the fonowing criteria: I) each was a common, overlearned expression having a stereotyped and non-literal meaning and 2) each stimulus could be depicted in a simple line drawing (see Fig. I). Novel phrases were then created to match the familiar phrases in length, word frequency, and grammatical structure (see Fig. 2). For all 3 tasks. the examiner spoke each stimulus aloud and provided the subject with a response sheet consisting of 4 line drawings. The subjects were encouraged 10 guess if they were reluctant 10 choose a single picture; in nearly all cases, a response was elicited.

For the familiar phrases, 3 foils (wrong answers) were designed. such that I contained a referential representation of I word in the stimulus (i.e. a concrete response). the second was a meaning opposite or related to the idiomatic meaning (a related response), and the third was an irrelevant choice. No literal foils (e.g. a man eating from a woman's hand for "she's got him eating out of her hand") were included because literal interpretations are, in one sense. correct. As the example illustrates, literal

S We use the term "familiar phrases" to refer generically to overlearned, holistic expressions, in· eluding proverbs, idioms, and social interaction formulae (e.g. greetings). Some of the items are full sentences while others are phrases, but all involve nooliteral interpretation, and all will be referred 10 as "phrases" throughout this paper.

40 ALZHEIMER DISEASE AND ASSOCIATED DISORDERS. VOL 2, NO. I. 1988

. II • @' ~

FIGURE I. Response array for the familiar phrase, "He's living high on the hog."

interpretations of ambiguous familiar phrases are often bizarre or less unlikely to occur than the figurath'e meaning. It has been shown that given the choice, listeners reject literal meanings in favor offigurative meanings, based on real world plausibility rather than ability to perceive abstract meanings (Gibbs, L980~ Van Lancker and Ca~ter. 1981), The inclusion ofa literal interpretation as a choice would make the results difficult to interpret, since it could not be unequivocally determined whether the litcni response was "right" or "wrong." Moreover, we were interested in patients' responses on familiar pbrases that do not have a picturable literal interpretation, which is the usual case (e.g. ·'She took a sudden turn for the worse" and "That's enough to drive a man to drink"). For the novel phrases, all 3 foils were grammatical misreadings such as adjective misassignment or subject-object reversal.

For each subtest, 2 practice items were given. Wrong answers on these items were corrected, correct answers were explained, and practice was repeated until tbe patient performed the task appropriately. The order of the tasks was always (1) words, (2) familiar phrases, and (3) novel phrases.

The meaning of familiar phrases is derived from associating the phrase as a whole with a single, often complex meaning. On the other hand, the meaning of a novel phrase relies on analysis of word meaning, word order, and grammatical relations. For the familiar phrases, this type of grammatical and single word analysis does not lead to a correct interpretation, but instead, results in a wrong answer. As mentioned above, the familiar and novel phrases were matched in length, grammatical structure, and word frequency . Therefore, the 2 phrase subte5ts provided a meaning/grammar contrast. That is, they differed only in whether the correct interpretation relied on decoding word meanings and syntactic relations (novels), or on inferring an abstract meaning where knowledge of individual words and grammatical relations was not useful for accurate comprehension (familiars).

Clinical subjects were 29 patients diagnosed with Probable Alzheimer Disease (10 female, 19 male) referred from the Neurobehavioral Unit of the West Los Angeles Veterans Administration Hospital, the UCSD Alzheimer Disease Research Center, and the UCLA Geriatric Outpatient Clinic. Consec­utive referrals were taken for the study; subjects were not pre-selected on the basis of language dysfunction or any other criteria. The evaluations were done by clinicians experienced in the evaluation of dementias. Each patient had physical and neurologic examinations, neuropsychological evaluation, laboratory tests (including urinalysis, CBC, thyroid function tests, serologic tests for syphilis, calcium, folate, SMAC panel, and vitamin Bl2 levels), EEG, EKG. chest X-ray. and CT of the head. The diagnostic evaluation documented a range of cognitive dysfunctions and a progressive history typical of AD, alongside an absence of focal motor, sensory, cerebellar, or cranial nerve defects. Diagnostic criteria v-ere in accord with the recent consensus (McKhann et aI., 1984), and in most cases were

--_ .... .... .. .

-1 ,

FIGURE 2.

more restrictive in patients with atypi, Angeles VA Medic

Written informe monolingual. nativ used SAE in their ' diagnosis of AD. A Mental Status Exal tunattly. reliable ir present that inform

Control subjects impairment (age ra both groups ranged and the groups did :

The protocol, normal subjects words. 97.6% co

In general, tho adequately; the) patients scored SO = 19.0; rang 49% correct; S[ Performance on

6 All (-test data al an overall 0.05 leVI significance. p must

(

Ile hog."

I) occur than the neanings in favor , .bstract meanings ~ition as a choice lermined whether ots' responses on

Jal cas< (e.g. "She "). For the novel or subject-object

correctcd, correct !Jsk appropriately. ,"ases. lole with a single, I analysis of word Ie of grammatical I'~ults in a wrong 19th, grammatical Jfamm8r contrast. I .~ word meanings dge of individual lars). l female, 19 male) .. tration Hospital, , t Clinic. Consec­basis of language I in the evaluation ogicaJ evaluation, syphilis, calcium, Df the head. The

Ive history typical I:fects. Diagnostic most cases were

KEMPLER ET AL: IDIOM COMPREHENSION 41

" o

~ /I . . \ \

1/\\

0 , /

r 0 -0-, , n

0

~~ 0 , .L1 ~ .", -r '.,-c 0 .) .,;- ',:>

flaURE 2. Response array for the novel phrase, "He's sitting deep in the bubbles."

more restrictive in requiring definite evidence of progressive cognitive dysfunction, and in excluding patients with atypical neurologic or psychiatric features. Five subjects were biopsied at the West Los Angeles V A Medical Center and all 5 were subsequently reclassified as Detinite Alzheimer Disease.

Written informed consent was obtained from primary caregivers in all cases. All subjects were monolingual, native speakers of Standard American English (SAE), who were schooled in SAE and used SAE in their home and work. None had any known speech or language pathology prior to the diagnosis of AD, Age ranged from 53 to 87 (mean age = 70.35; SD - 8.8). Perfonnance on the Mini Mental Status Ex.am (MMSE) (Foisiein et aI. , 1975) ranged from 2 to 28 (of a possible 30). Unfor­tunately. reliable information about age at onset of illness was not available in enough instances to present that information here.

Control subjects were a group of 43 normal older volunteers with no known history of cognitive impainnent (age range: 50-82; mean age = 66.0; SD - 9.8; 13 men and 30 women). Education for both groups ranged from 8th grade to postgraduate degree; most subjects had a high school education, and the groups did not differ in average educational levels. Testing was always completed in I session.

RESULTS

The protocols were designed to result in essentially perfect performance from normal subjects. The 43 older normal controls in this study scored 100% on single words, 97.6% correct on familiar phrases, and 99.7% correct on novel phrases.

In general, the AD patients were able to perform the word comprehension task adequately; they obtained a mean 96% correct (SO = 7.0; range 750/0-100%). The patients scored significantly better on the novel phrases task (mean 74% correct; SO = 19.0; range 40%-100%), than they did on the familiar phrase subtest (mean 49% correct; SO = 19.0; range 10%-100%; matched pairs t = 7.084; p < 0.01)'. Performance on single words was significantly better than performance on either

6 All t-test data are reported with the Bonferroni adjustment to correct for multiplicity of tests: for an overall 0.05 level of significance. each individual p must be less than 0.0125; for 0.01 level of significance. p must be less than 0.0015; for 0.00 I Icyel of significance, p must be Jess than 0.00025.

42

100

90

80

I- 70 u W 0:: 0::

60 a u I-:z: 50 w U 0:: W 40 a...

.30

20

10

AlZHEIMER DlSfASE AND ASSOCIATED DISORDERS, vOL 2. NO. I. 1988

•

MILD (n=91

MODERATE (n=lf)

Words

NOV6/ Phrl7ses

Forni/ior Phrl7se.s

SEVERE (n-9)

FIOuu3. Mean scores on words, novel phrases, and familiar phrases for 3 groups of AD patients. Group ! (mild) patients had a mean age of 65.6 (SD - 8.2). Mini Mental Status > 20 (mean = 23.7; SD = 3.1), and achieved mean percent correct scores on words, novel and familiar phrases of 98.9 (SO ~ l .3), 84.4 (SO ~ 13.3), and 54.4 (SO - 21.9) respe<:tively. Group 2 was moderately impaired (MMS 15-20; mean - 17.5; SO - 2.1), mean age of73.2 (SO ~ 8.2), and averaged 95.5 (SO - 7.2), 80.5 (SD ~ 15.9), and 52.7 (SO ~ 16.2) on the 3 subtests. Group 3 was more severely impaired (MMS ciS), :nean age 71.7 (SO - 8.9), and obtained mean scores of 92.8 (SD ~ 8.7), 55.0 (SD - 13.7),

and 38.9 (SO ~ 16.9) on the 3 language tasks.

familiar (/ = 12.75; P < 0.001) or novel phrases (/ = 7.1, P < 0.001). To compare performance on the 3 tasks at different stages of severity, the clinical group was divided into 3 subgroups according to their MMSE scores: mild (> 20; n = 9), moderate (15-20; n = 11), and severe « 15; n = 9)(see Fig. 3). A 2-factor, group x task Al'lOY A with repeated measures on task (words, familiar and novel phrases) revealed a significant main effect of group (F(2,26) = 7.834; P < 0.01), significant main effect of task (F(2,26) = 104.699; p < 0.0001), and a group x task interaction (F(4,52) = 2.75; p < 0.05). All 3 groups showed the same pattern, performing worst on familiar phrases and best on single words. This pattern held for even the mildly affected patients (MMSE > 20) who showed a substantial deficit in understanding familiar stimuli (mean = 54% correct; SO = 21.9), compared with novel ltimuli (mean 84% correct; SO = 13.3; matched pairs t = 4.811; P < 0.01). In fact. the discrepancy in performance on novel and familiar phrases generally held across the group: of the 29 AD subjects, 24 performed worse on the familiar phrases, 4 achieved identical scores on both tests, and only I performed better on the familiar phrases.

Correlational analyses comparing age, MMSE scores, and scores on the 3 lan­guage comprehension tasks yielded several significant correlations: MMSE scores

----_._-- ..... ...... ... ..... ... .

correlated sign . and words (r ' correlated (r = p < 0.01). Age, with scores on significantly yc performance al neither age nor measures. As IT

and worst on fi and younger) a1 groups were ot Similarly, no g.

Analysis of e Wrong answen choices of the: Ofa total of I~ choices and 8~ indicating a tef of severity.

Using a pict. differential diss Our results con understanding; mance interpre understanding c This pattern he alternative intel concrete respon meaning to intI significantly cor tasks (words an, functioning as n relevant psycho

Why do AD J

nonlinguistic ex visual pereeptua phrase subtest. phrases were m' matieal struclUr phrase subtest c, would be expect hension has bee Rochford. 1971' plained by diffie phrases. This is

°hrases

- Phrases

I i of AD patients. 20 (mean - 23.7; J' phrases of 98.9 terately impaired 95.5 (SD ~ 7.2),

I impaired (MMS ,5.0 (SD ~ 13.7),

). To compare cal group was (> 20; n = 9), letor, group x novel phrases) )I), significant 1sk interaction n, performing held for even

ntial deficit in ompared with II; p < 0.01). 'ases generally >n the familiar "formed better

i on the 3 lan­MMSE scores

KEMPLER ET AL.: IDIOM COMPREHENSION 43

correlated significantly with performance on novel stimuli (r = 0.618; p < 0.00 I) and words (r = 0.481; p < 0.0 I); novel phrases and words were significantly correlated (r = 0.507; P < 0.01); as were novel and familiar phrases (r = 0.506; p < 0.0 I). Age, overall, was not correlated with MMSE scores, nor did age correlate with scores on any of the 3 comprehension tasks. Although the mild group was significantly younger than either the moderate or severe group, the pattern of performance and level of significance were constant across groups, suggesting that neither age nor mental status affect the pattern of performance on these 3 language measures. As mentioned above, nearly all subjects performed best on single words, and worst on familiar phrases. When the patients were divided into younger (65 and younger) and older (>65) groups, no significant differences between the 2 age groups were observed in MMSE scores or performance on the language tasks. Similarly, no gender differences were observed on any measure.

Analysis of errors revealed a significant tendency to select concrete responses. Wrong answers on each of the 10 stimuli were pooled from all 29 subjects, and choices of the 3 kinds of foils (concrete, related, and irrelevant) were compared. Of a total of 145 errors, 81 % were concrete choices, compared with II % related choices and 8% irrelevant choices. This ratio was similar for all 3 subgroups, indicating a tendency towards concrete error responses for patients at all levels of severity.

DISCUSSION AND CONCLUSIONS

Using a picture pointing task in a comprehension protocol, we have shown differential dissolution of 3 features of auditory language comprehension in AD. Our results confirm suggestions that dementia patients have particular difficulty understanding abstract language, as observed by their particularly poor perfor­mance interpreting idioms and proverbs compared with their relatively better understanding of novel phrases and overall accurate interpretation of single words. This pattern held across mild to severe levels of impairment. When faced with alternative interpretations offamiliar phrases, AD patients overwhelmingly chose concrete responses, suggesting that they were using lexical (single word), referential meaning to interpret the phrases. Although comprehension of novel sentences significantly correlated with MMSE scores, neither of the other 2 comprehension tasks (words and familiar phrases) appeared strongly related to overall cognitive functioning as measured by the MMSE. In the remainder of this report we address relevant psycholinguistic issues and clinical implications of these findings.

Psycholinguistic Issues

Why do AD patients have trouble understanding familiar phrases? At least 3 non linguistic explanations must be addressed: attentional and memory deficits, visual perceptual deficits, and overall level of difficulty associated with the familiar phrase subtest. In the tasks reported here, the novel phrases and the familiar phrases were matched in length (as well as word frequency and superficial gram­matical structure). Therefore, signiflcantiy worse performance on the familiar phrase subtest cannot be attributed to a general memory or attention deficit, which would be expected to affect both tasks equally. Second, since poor word compre­hension has been attributed to visual perceptual deficits (Kirshner et aI., 1984; Rochford, 1971) it is conceivable that our results could be at least partially ex­plained by difficulty in interpreting the pictures which accompany the familiar phrases. This is unlikely, as the line drawings for both novel and familiar phrases

44 ALZHEIMER DISEASE AND ASSOCIATED DISORDERS, VOL. 2, NO. I. 1988

were of similar quality and complexity (see Figs. I and 2). Third, it is tempting to in',oke a difficulty metric. whereby familiar phrases could be seen as inherently more difficult to comprehend than novel phrases. If this were true, we would expect to see an effect of normal aging on the task, and a similar effect in all cases of brain injury, but we find neither. Normal controls performed nearly perfectly on b)th tasks, regardless of age, and age of the clinical subjects was not correlated with performance on the familiar phrase task (r = -0.075). Further, another brain injured population (left hemisphere damaged patients) demonstrated a reversed pattern, in that they understood familiar better than novel phrases (see below and Van Lancker and Kempler, 1987).

To understand what makes the familiar phrases psychologically different from the novel phrases, we must look at the processes involved in comprehension of the 2 sentence types. Psycholinguistic research with normal subjects indicates that, whereas the meaning of a novel sentence is derived from a sequential analysis of constituent parts, successful recognition of familiar phrases involves perception of an overall pattern (Gibbs, 1980; Horowitz and Manelis, 1973; Lieberman, 1963; Osgood and Hoosain, 1974; Simon, 1974). These studies have shown, for example, that normal subjects can identify and correctly interpret idiomatic phras­es more quickly than novel phrases (Gibbs, 1980; Swinney and Cutler, 1979). Moreover, normal subjects appear not to remember the individual words that make up idiomatic phrases, whereas they do remember words constituting com­parable novel phrases (Horowitz and Manelis, 1973; Osgood and Hoosain, 1974). Further evidence comes from digit span studies using linguistic units, showing that subjects retain in memory almost as many familiar phrases as words (5--9). These results indicate that familiar phrases are psychological "chunks" similar to word" and that they are perceived and understood as wholes, without analysis of constituent parts (e.g. individual words) taking place at all. The data suggest that people do not process idioms by first processing individual words and at­temp~ing to analyze their literal meaning, and only then deriving the idiomatic meaning. Precisely how grammatical analysis is bypassed is not well understood, but it is possible that, at least in part, single unitary intonation envelopes which accompany familiar but not literal phrases (Van Lancker et aI., 1981) serve as acoustic cues to initiate holistic rather than grammatical processing. Further, in order to understand a familiar phrase, one must be able to associate the overall entity with a unique meaning (Peters, 1983). It is likely that AD patients have diffic\:lty either with recognizing the overall pattern of familiar phrases and/or with accessing the associated complex meaning. Previous research with AD pa­tients lends support to both of these possibilities.

Regarding the first (overall perception ofa unitary phrase), there is reason from other observations to believe that AD patients are impaired in apprehending overa!l patterns. They are known to have difficulty perceiving and reproducing both nonsymbolic visual patterns such as geometric forms (Cummings and Ben­son, 1983) and symbolic, communicative gestures (Kempler, 1984, in press). In both these areas, it is notable that AD patients can often reproduce some of the details but not the overall Gestalt or pattern. These findings suggest that AD patier.ts may demonstrate a generalized impairment in what has been variously called holistic. integrative, global, or synthetic processing, and has been associated with right hemisphere specialization (Bever, 1975; Bogen. 1969a, b; Bradshaw and Nettleton, 1983; Bryden, 1982; Lezak. 1983). We would expect such a general deficit to affect perception and interpretation of familiar phrases, which also, hy definition, have holistic form and integrated meanings.

AD patients that AD patiel wrong answen mini shes the a more operatio: difference bet" a fixed form " sequential and known to hav. and Kaszniak, subtest is relate es. Since coml meanings. and and proverb in1 relative presen in this task, be lization of the~ 1980; Kempler speculative exp have difficulty f matching that I

Language pre associated with data presented damaged patie! specific diJferen and those with J

right handed pa origin, with the ed, native Engli between 10 and damaged and al type was diagno 8 patients had f anomie, and 2 , 63.4; average I' stand familiar. pattern opposite performed like to process and) (see also Benton Figure 4 shows hemisphere darr subtests. The sil reHect the neces pletion of the Ii left hemisphere associated with

it is tempting , as inherently lie, we would i.et in all cases ,:arJy perfectly not correlated another brain ted a reversed see below and

different from 'prehension of indicates that, tial analysis of "es perception .1 ; Lieberman, Ive shown, for iomatic phras­Cutler, 1979). 1111 words that stituting com­()osain, 1974). Imits, showing , words (5-9). lks" similar to thout analysis ,e data suggest words and at­ilie idiomatic

:11 understood, vel opes which 1981) serve as ug. Further, in lite the overall . patients have .hrases and! or I, wiili AD pa-

is reason from apprehending

ld reproducing lings and Ben­l, in press). In ce some of the 'sgest that AD been variously leen associated I, b; Bradshaw I such a general which also, by

KEMPLER ET AL.: IDIOM COMPREHENSION 45

AD patients are also known to have difficulty accessing meanings. The finding that AD patients selected significantly more concrete than related or irrelevant wrong answers lends empirical support to the popular belief that dementia di­minishes the ability to process abstract meanings. While this is true, we prefer a more operationally definable formulation of this notion. Recall that a primary difference between familiar and novel phrases is iliat familiars involve associating a fixed form with a complex and integrated meaning, while the novels require sequential and grammatical analysis to derive a meaning. Since AD patients are known to have particular trouble accessing meaning (e.g. Bayles, 1982; Bayles and Kaszniak, 1987), we suggest that the deficit observed on the familiar phrase subtest is related to the integrated, complex meanings associated with these phras­es. Since comprehension of familiar phrases requires ilie retrieval of complex meanings, and does not rely on syntactic abilities, it is understandable that idiom and proverb interpretation would be disproportionately compromised in AD. The relati ve preservation of grammatical abilities may actually defeat the AD subjects in this task, because correct interpretation of familiar phrases requires non-uti­lization of these very grammatical mechanisms which are best preserved (Gibbs, 1980; Kempler et ai. , 1987). In summary, we propose a tentative and necessarily speCUlative explanation for the proverb and idiom deficit, whereby AD patients have difficulty first perceiving the overall pattern ofthe familiar phrase and second, matching that phrase to its meaning.

Clinical Issues

Language profiles of AD patients have been compared with language profiles associated with focal aphasias (Appel et aI., 1982; Cummings et ai., 1985). The data presented here, taken together with comparisons of left and right brain damaged patients on this same task (Van Lancker and Kempler, 1987), indicate specific differences between the language abilities of patients with focal aphasias and iliose with AD. Van Lancker and Kempler studied 39 consecutively available, right handed patients with documented unilateral focal brain damage of vascular origin, with the same protocol and procedure described here. All were right hand­ed, native English speakers (5 women, 34 men; mean age 62.8; range 45-80) with between 10 and 19 years of education. Twenty-nine of the patients were left brain damaged and aphasic (mean age 62.3 ; average 63.5 months post onset). Aphasia type was diagnosed by a speech-language clinician using standardized instruments: 8 patients had fluent aphasia, 10 were non-fluent, 3 were globally aphasic, 5 were anomie, and 2 were mixed. Eleven subjects were right brain damaged (mean age 63.4; average 19.5 months post onset). The aphasic patients were able to under­stand familiar phrases significantly better than they understood novel phrases, a pattern opposite from that observed with AD patients. In contrast, the AD patients performed like the right hemisphere damaged patients insofar as they were able to process and recognize novel phrases significantly better than familiar phrases (see also Benton, 1968; Myers and Linebaugh, 1981 ; Winner and Gardner, 1977). Figure 4 shows the performance of 28 left hemisphere aphasic patients, II right hemisphere damaged patients and 29 AD patients on the familiar and novel phrase subtests. The similarity between AD patients and right hemisphere patients may reftect the necessity of a functionally intact right hemisphere for successful com­pletion of the familiar phrase task. The qualitatively different performances of left hemisphere damaged aphasic and AD patients suggests that the "aphasia" associated with AD is distinct from that associated with focal left hemisphere

46 AUHEIMER DISEASE AND ASSOCIATED DISORDERS. vOL. 2. NO. I. [988

100

90 ~ Novel Pl1ro.s~$

80 • Forni/lo r Phrustls

I- 70 u UJ a::: 60 a::: 0 u

50 I-:z UJ u 40 a::: UJ a..

30

20

10

0 LBO (n-26) RBD (n-ll) AD (n-29)

FIOURE 4. Mean scores on novel and familiar phrases for left brain damaged (LBO), right brain damaged (RBD), and Alzheimer (AD) patients.

lesions, despite earlier reports of the similarities (Appel et aI., 1982; Cummings et aL 1985).

Early identification of AD is crucial for developing and implementing treatments which will prevent and relieve the effects of the disease. Since much research on AD is plagued by problems of accurate diagnosis, it is notable that the accuracy of di~gnosis in our subjects can be expected to be high, Recent experience with biopsy (Harbaugh, personal communication) suggests that restrictive clinical cri­teria mch as those used in the present study yield an accuracy of diagnosis ap­proaching 95%, Our own data confirming that the 5 patients who underwent cortical biopsy (including 2 mildly affected patients with MMSE scores of 21 and 28) w~re accurately diagnosed as Alzheimer patients further validates our diag­nostic procedure, Although several teChniques have been proposed for early dif­ferential diagnosis (e,g,. larvik et aI., 1982), few behavioral tests have been suc­cessful in this regard. Language deficits have held particular promise, insofar as they have been present in mildly demented cases (Huff and Growdon, 1986) and they have been linked to clinical variables such as age at onset, family history, and rate of progression of AD (Filley et aI., 1986; Folstein and Breitner, 1982; Knesevich et al.. 1985). Although some patients present clear deficits in word­list generation early on in the course of the disease (Cummings et al., 1985), this symptom does not differentiate them from patients with aphasia due to focal left hemisphere damage, The Familiar and Novel Language Comprehension test de­scribed here seems to highlight an early-appearing, subtle intellectual defect which

may appear be phasias and p( patients from It to early cogniti psychological b

lllustrations wert assistance with test the Mahood Senior their patrons. This (NIA #P50-AGO-5 of the West L.A. ' Outpatient Clinic, I

Appel J, Kertesz A 17:73-91_ 198:

Bayles K: CommUJ 1979.

Bayles K: Langu3g Bayles K. Kaszniak:

Little, Brown a Bayles K, Tomoedl

1983. Benton AL: Differ! Bever TG: Cerebr

processes: holi! Bogen JE: The alhl

162. 1969 •. Bogen JE: The othf

219, 1969b. Bradshaw JL, Nettlf Bryden MP: Later.

1982. Cummings 1: Corti

Disease. New): Cummings 1. Bense Cummings J. Senso

35:394-397, 19 Filley eM, Kelly 1.

Disease. Arch It Falstein MF. Breitr

(ed.), Af=heime, 1982.

Falstein MF. Falstei state of patient!

Gibbs RW: Spilling 8:149-156,198

Gorham DR: A Pn Horowitz lM. Mant

296, 1973. HuffFl, Growdonll

Disease. Can J Jarvik LF. Matsuyal

marphonudear Kempler D: Syntac

talian. UCLA . . Kempler 0: Lexica:

29)

LBD), right brain

52; Cummings

ting treatments eh research on It the accuracy ' perience with ve clinical cri­r diagnosis ap­ho undeIWent :ores of 21 and ;!ates our diag­d for early dif­laVe been suc­lise, insofar as lon, 1986) and iamily history, rlreitner, 1982; ·ficits in word­aI., 1985), this lue to focal left ension test de­al defect which

KEMPLER ET AL.: IDIOM COMPREHENSION 47

may appear before other, more overt aphasic symptoms such as anomia, para­phasias and poor word-list generation. If the test continues to distinguish AD patients from those with aphasias due to focal lesions, and continues to be sensitive to early cognitive deficits, it may be beneficially added to diagnostic and neuro­psychological batteries.

ACKNOWLEDGMENTS

Illustrations were drawn by Susan Black. We thank Catherine O'Connor and Cathy Cornelius for assistance with testing and Wendy Mack for statistical consulting. We appreciate the cooperation of the Mahood Senior Center of Santa Monica for a1lowing us to recruit normal volunteers from among their patrons. This work was supported in part by The UCSD Alzheimer Disease Research Center {NIA I¥P50-AGO-5131 , Robert Katzman, P.I.}. We are grateful to Drs. J. Cummings and S. Y. Tsai of the West L.A. VA Medical Center, Drs. W. Ashford and J. Schaeffer of the UCLA Geriatric Outpatient Oinie, and The UCSD Alzheimer Disease Research Center for patient referrals.

REFERENCES

Appel J, Kertesz A, Fisman M: A study oflanguage functioning in Alzheimer patients. Brain Lang 17:73-91 , 1982.

Bayles K: Communication profiles in a geriatric population. Ph.D dissertation, University of Arizona, 1979.

Bayles K: Language function in senile dementia. Brain Lang 16:265-280, 1982. Bayles K, Kaszniak A: Communication and Cognition in Normal Aging and Dementia. Boston, MA:

Little, Brown and Co. , 1987. Bayles K, Tomoeda CK: Confrontation naming impairment in dementia. Brain Lang 19:98-114,

1983. Benton AL: Differential behavioral effects in frontal lobe disease. Neuropsychologia 6:53-60, 1968. Bever TG: Cerebral asymmetries in humans are due to the differentiation of two incompatible

processes: holistic and analytic. Ann NY Ac:ad Sci 263:251-262, 1975. Bogen IE: The other side of the brain II: an appositional mind. Bull Los Angeles Neurol Soc 3.4:135-

162. 1969a. Bogen JE: The other side of the brain II: an appositional mind. Bull Los Angeles Neural Soc 3.4:191-

219, 1969b. Bradshaw JL, Nettleton NC: Human Cerebral Asymmetry. Englewood Cliffs, Nl: Prentice-Hall, 1983. Bryden MP: LateralilY: Funclional Asymmet(l" in the Human Brain. New York: Academic Press,

1982. Cummings J: Cortical dementias. In DF Benson, D Blumer (eds.), Psychiatric Aspects of Neurologic

Disease. New York: Grune and Stratton, 1982, Pl'. 93-121. Cummings J. Benson DF: Dememia: A Clinical Approach. Boston: Butterwonhs, ]983. Cummings J, Benson OF, Hill MA, Read S: Aphasia in dementia of the Alzheimer type. Neuroiog)'

35:394-397, 1985. Filley CM, Kelly J, Heaton RK: Neuropsychologic features of early- and late-onset Alzheimer' s

Disease. Arch NeuroI43:574-576 , 1986. Folstein MF, Breitner JCS: Language disorder predicts familial Alzheimer's Disease. In S Corkin

(ed.), Alzheimer's Disease: A Report of Progress (Aging, Volume 19). New York: Raven Press. 1982.

Folstein MF, Folstein SE, McHugh PR: Mini Mental State: a practical method for grading the cognitive state of patients for the cJinician. J PS.I'ChiOlr Res 12:189-198, 1975.

Gibbs RW: Spilling the beans on understanding and memory for idioms in conversation. Mem Cognil 8:149-156.1980.

Gorham DR: A Proverbs Test for clinical and experimental use. Psycho! Rep 1:1-12, 1956. Horowitz LM. Manelis L: Recognition and cued recaU of idioms and phrases. J Exp Psych 100:291-

296. 1973. HuffFl , Growdon IH: Neurological abnormalities associated with severity of dementia of Alzheimer's

Disease. Can J Neurol Sci 13:403-405. 1986. Jarvik LF, Matsuyama SS, Kessler 10, Fu TK. Tsai SY, Oark EO: Philothennal response of poly­

morphonuclear leukocytes in dcmentia of the Alzheimer type. Neurobiol Aging 3:93-99, 1982. Kempler D: Syntactic and symbolic abilities in Alzheimer's Disease. Unpublished doctoral Clisscr­

lation, UCLA. 1984. Kempler D: Lexical and pantomime abilities in Alzheimer's Disease. Aphasiology, in press.

48 ALZHEIMER DISEASE AND ASSOCIATED DISORDERS, VOL 2, NO. I. 1988

Kem ~Ier D. Curtiss S, Jackson C: Syntactic preservation in Alzheimer's Disease. J Speech Hear Res )0:343-350, 1987.

Kirshne:- HS. Webb W, Kelly MP: The naming disorder of dementia, Neuropsycholog;a 22:23-30, 1984.

Knesevich JW, Torar FR, MorrisJC. Labarge E: Aphasia, family history, and the longitudinal course arsenite dementia orlhe Alzheimer type. Psychiatry Res 14:255- 263,1985.

Leza~: MD: Neuropsychological Assessmem. New York: Oxford University Press, 1983. Lieberman P: Some effects of semantic and grammatic context on the production and perception of

speech. Lallg Speech 6: 172-187. 1963. McKhar.n G. Drachman D. Folstein M. Katzman R. Price D, Stadlan EM: Clinical diagnosis of

.'.Izheimer's disease: report of the NINCDS-ADRDA Work Group under the auspices of De­partment of Health and Human Services Task Force on Alzheimer's Disease. Neurology 34:939-144. 1984.

Myers PS. Linebaugh CW: Comprehension of idiomatic expressions by right-hemisphcre-damaged dults. In Brookshire RH (ed.), Clinical Aphasiology Proceedings, 1981 , Minneapolis: BRK. Pub­I;,hers, pp. 256-261.

Obler L. Albert M: Language in the elderly aphasic and in the dementing patient. In MT Sarno (cd.), Acq,tired A.phasia. New York: Academic Press. 1981, pp. 385-399.

OsgOGd C. Hoosain R: Salience of the word as a unit in the perception of language. Perception and Psychoph),sic, 15: 168-192, 1974.

Peter. A: The Units of Language. Cambridge. England: The Cambridge University Press. 1983. Roch iord G: A study of naming errors in dysphasic and in demented patients. Neuropsychologia 9:

437-443. 1971. Schw"rtz M, Marin 0, Saffran E: Dissociations of language function in dementia: a case study. Brain

Lang 7:277-306, 1979. Simon HA: How big is a chunk? Sciellce 183:482-488, 1974. Strub RL, Black FW: The Mental Staws Examination in Neurology. Philadelphia: F. A. Davis

Company, 1985. Swinr.ey DA, Cutler A: The access and processing of idiomatic expressions. J Verbal Learning Verbal

Behavior 18:523-534, 1979. Van L!lncker D. Canter OJ: Idiomatic versus literal interpretations of ditropically ambiguous sen·

ttoces. J Speech Hear Res 46:64-69, 1981 . Van Ll ncker D. Canter GJ , Terbeek D: Disambiguation of ditropic sentences: acoustic and phonetic

cues. J Speech Hear Res 24:330-335, 1981. Van Lancker D, Kempler D: Comprehension of familiar phrases by left- but not by right-hemisphere

damaged patients. Brain Lang. 32:265--277. 1987. Wech~ler AF: Presenile dementia presenting as aphasia. J Neurol .Veurosurg Psychiatry 40:303-305,

19-77. Wechsler D: "f"echsler Adult Intelligence Scale. Manual. New York: Psychological Corporation, 1955. Winner E. Gardner H: The comprehension of metaphor in brain-damaged patients. Brain 100:719-

727, 1977.

APPENDIX

Single Words

1. cigarette 6. car 2. iro:! 7. hat 3. speDn 8. pen 4. umbrella 9. sink 5. ba.y 10. drum

Familiar Phrases

1. While [-he eat's away, the mice will play. 2. I'd like to give you a piece of my mind. 3. He' :; saving up for a rainy day. . 4. Tht truth, the whole .truth and nothing but

the truth. ' 5. She took a sudden tum for the worse.

---_ ........ .... .. ..•. .. ..... ... ........

6. He's turning over a new lea( 7. That's enough to drive a man to drink. S: She's got him eating out of her hand. 9. It's like talking to a brick wall.

10. He' s living high on the hog.

I. When the hapl swings.

2. The clown likes 3. He's taking her 4. The clown, the

or the girl.

I~{>(h Hear Res

rOgia 22:23-30,

titudinal course

'n Id perception of

.. a1 diagnosis of ~uspices of De­rtrology 34:939-

.phere-damaged mlis: BRK Pub-

MT Sarno (00.),

Perception and

Press, 1983. I'Opsychologia 9:

use study. Brain

la: F. A. Davis

u arning Verbal

9.mbigudus sen-

.tic and phonetic

!ght~hemisphere

,try 40:303-305,

tporation, 1955. Brain 100:719-

!if. lan to drink. her hand.

"all.

KEMPLER ET AL.: IDIOM COMPREHENSION

Novel Phrases

1. When the happy girl pushes, the angry boy swings.

2. The down likes to pull the girl on the wagon. 3. He's taking her for a little ride. 4. The clown, the small clown, is not in front

of the girl.

5. She'll push her brother to the ground. 6. He sits there looking away. 7. The happy giTI has a wagon to pull . 8. She's glad to be giving him a wagon ride. 9. She likes being in the wooden swing.

10. He's sitting deep in the bubbles.

49