Brain and Cognition 59 (2005) 306–309 www.elsevier.com/locate/b&c

The effect of AlzheimerÕs disease and aging on conceptual combination Vanessa Taler a,b, Howard Chertkow a,c, Daniel Saumier a,c,* b

a Institut Universitaire de Ge´riatrie de Montre´al, Montre´al, Que., Canada H3W 1W5 ´ Departement de Sciences Biome´dicales, Universite´ de Montre´al, Montre´al, Que., Canada H3W 1W5 c Lady Davis Institute for Medical Research, Montre´al, Que., Canada H3W 1W5

Accepted 12 August 2004 Available online 12 March 2005

Abstract AlzheimerÕs disease (AD) subjects, healthy elderly, and young adults interpreted a series of novel noun–noun expressions composed of familiar object words. Subjects interpreted each item by selecting one of three possible definitions: a definition in which the referents of each noun were associated together in a particular context (e.g., rabbit cat: a cat that is raised by rabbits); a definition based on integrating the semantic attributes from each noun into a single referent (e.g., rabbit cat: a cat that has long ears and hops); and a semantic foil. The results show that the AD subjects selected significantly fewer integrational definitions and significantly more foils than healthy elderly subjects. Healthy elderly participants were also found to select significantly more foil definitions than the young adult subjects. These findings suggest that AD individuals have difficulty in integrating semantic features when interpreting novel noun–noun expressions and that both AD and healthy elderly individuals have morphosyntactic impairments related to the identification of the modifying and head nouns of noun–noun expressions. Ó 2005 Published by Elsevier Inc.

1. Introduction Impairments in semantic functioning in individuals with AlzheimerÕs disease (AD) have been well documented (for a review, see Whatmough & Chertkow, 2002). Their semantic impairments may be demonstrated in a variety of standardized tasks, including picture naming, verbal fluency (e.g., name as many animals as possible within a 1 min period), auditory probes (discrimination of true/false statements about objects), word-to-picture matching, and defining common object names. In contrast, healthy elderly individuals perform similarly to young adult subjects on such tasks, indicating that they have normal semantic functioning. While much effort has been devoted to investigating semantic memory functions in AD and elderly individuals, little

*

Corresponding author. E-mail address: [email protected] (D. Saumier).

0278-2626/$ - see front matter Ó 2005 Published by Elsevier Inc. doi:10.1016/j.bandc.2004.08.049

is known about how AD and normal aging affect the ability to form new semantic representations. The purpose of the present study is to explore the effects of AD and aging on the ability to form new conceptual representations of novel compound words. Novel noun–noun combinations require the integration of information about each of the composite nouns to achieve a coherent interpretation of the compound as a whole. Wisniewski (1996) posits that the nouns may be combined in one of two ways. First, the noun–noun expression may be interpreted by selecting an integrational interpretation, where a salient feature of the modifying noun (i.e., the first noun appearing in the noun–noun combination) is integrated into the representation of the head noun (i.e., the second noun of the noun–noun expression). For example, the combination ‘‘zebra horse’’ may be interpreted as a horse with stripes. On the other hand, an association process may be used whereby the referents of the modifying and head nouns are associated within the same context, but without feature integration. For example, the

V. Taler et al. / Brain and Cognition 59 (2005) 306–309

combination ‘‘zebra house’’ may be interpreted to mean a house in which zebras live. In the present study, AD subjects, healthy elderly individuals and young controls were required to match integrational and associative definitions to novel noun– noun expressions composed of familiar object words. Previous research suggests that the semantic errors of Alzheimer subjects occur because of mis-selection of a lexical entry due to featural competition among similar conceptual representations (e.g., responding ‘‘cat’’ to the stimulus lion; LaBerge, Balota, Storandt, & Smith, 1992), while associative errors (e.g., calling a pyramid a camel) occur rarely. It is therefore expected that AD subjects would be less likely to interpret novel noun–noun expressions in an integrational manner, while more frequently using associative definitions instead. We also tested whether AD individuals would show difficulties in correctly identifying the modifying and head nouns of noun–noun expressions. Finally, since there is substantial evidence indicating that semantic impairments do not occur as a result of normal aging, we expected healthy elderly adults would show similar patterns of performance when interpreting the noun–noun combinations.

2. Method 2.1. AD participants Ten AD patients were recruited from the Sir Mortimer D. Davis Jewish General Hospital Memory Clinic, a tertiary medical center referral clinic. Diagnosis of ‘‘probable AD’’ or ‘‘possible AD’’ was established according to diagnostic criteria for dementia according to the Diagnostic and Statistical Manual of Mental Disorders, Third Edition revised (DSM-III-R), and all subjects were diagnosed as having probable AD according to standard clinical criteria. Standard blood work and neuroimaging (CT or MRI) were also carried out, and the diagnosis was supported by abnormal performance on neuropsychological testing. Mean age and education of the AD participants were 81.9 (SD = 10.1) and 11.1 (SD = 11.1), respectively. Average Mini-Mental State Exam (MMSE) score was 23.8 (SD = 2.9). 2.2. Elderly and young adult controls Ten non-demented healthy and independently living elderly subjects (4 men and 6 women) served as agematched controls. Mean age and years of education were 77.2 (SD = 6.7) and 14.2 (SD = 2.0), respectively. There were no statistical differences between the AD and elderly control groups in terms of mean age [t (18) = 1.5, n.s.] or education [t (18) = 1.8, n.s.]. Ten young (5 men and 5 women; mean age 27, SD = 5.0) were also recruited for the study. The young subjects

307

were all first year undergraduate students. The healthy elderly and young adult subjects were native speakers of English, and none had a history of neurological or psychiatric disease. 2.3. Stimuli Twenty novel noun–noun compounds were created using familiar object words. An equal number of natural (e.g., rabbit cat) and non-natural (e.g., book bicycle) noun–noun compounds were constructed for the stimulus set. We avoided using combinations of natural and non-natural concepts (e.g., igloo cat) in the set because integrative definitions tend to be less associated with the meanings of such compounds (Wisniewski, 1996) and we wished to obtained adequate numbers of such definitions from among the subjects. The entire set of noun–noun expressions included the following: news magazine, coconut tiger, violin guitar, canary bat, deer horse, mosquito fly, balloon ball, car bus, igloo tent, rabbit cat, book bicycle, couch sax, cup sock, hat boat, mouse squirrel, mirror scarf, apple duck, banana frog, cactus potato, and robin termite. Three possible definitions were constructed for each noun–noun combinations, one of which was a semantic integration interpretation, one of which was an associative interpretation, and one of which was a semantic foil whereby the modifying noun was taken to be the head noun. For example, for the stimulus rabbit cat, the integrational interpretation was ‘‘a cat that has long ears and hops,’’ the associative interpretation was ‘‘a cat that hunts rabbits,’’ and the semantic foil was ‘‘a rabbit that hunts mice and purrs.’’ Participants were asked to decide which of the three possible definitions best represented the meaning of each noun–noun item. Each noun–noun combination was presented in written form, along with its corresponding three definitions. Stimuli were presented in written form. In the case of AD patients, they were also read aloud to ensure comprehension. After completing the experiment, the AD subjects were given a picture–naming matching test to determine whether they correctly recognized the meanings of the words making up noun–noun expressions in the conceptual combination test. In this test, the subject were required to match each name comprising the conceptual combination expressions (40 names in all) to one of three black-and-white line drawings, one of which correctly corresponded to the target name. The words and pictures were presented on a series of pages in a random order.

3. Results All AD subjects performed at ceiling on the word–picture matching test. The results of the conceptual combina-

308

V. Taler et al. / Brain and Cognition 59 (2005) 306–309

Fig. 1. Numbers of association, integration, and foil definitions selected by the AD, healthy elderly, and young participants.

tion test were analyzed in terms of the relative numbers of association, integrational, and foil definitions by the AD vs. healthy elderly groups, or by the healthy elderly vs. young control groups (see Fig. 1). The findings indicate that, while the AD subject selected similar numbers of association definitions (v2 (9) = 6.3, n.s.) as the healthy elderly subjects, they selected significantly fewer numbers of integrational definitions (v2 (9) = 31.3, p < .01) and significantly greater numbers of foil definitions (v2 (9) = 238.6, p < .01) as their elderly counterparts. Moreover, while the young healthy subjects selected similar numbers of association (v2 (9) = 11.4, n.s.) and integrational definitions (v2 (9) = 15.0, n.s.), the healthy elderly subjects selected significantly more foil definitions, than the young participants (v2 (9) = 35.0, p < .01).

4. Discussion As predicted, the results of this study reveal that AD subjects are less likely to use integrational definitions and more likely to select foil definitions when interpreting novel noun–noun compounds than the healthy elderly subjects. Moreover, the AD subjects performed perfectly on a control picture-matching task involving the individual words making up the noun–noun compounds. Perfect performance on a control picture–naming matching test ensured that the AD subjectsÕ knowledge of the individual words making up the noun–noun compounds was sufficiently intact to correctly identify their meanings. Unexpectedly, the healthy elderly subjects selected significantly more foil definitions when interpreting the noun–noun compounds than the young adult subjects. These findings suggest that AD individuals have a marked impairment in the capacity to integrate semantic features when interpreting novel noun–noun expressions. They also indicate that AD individuals have difficulties in correctly identifying the

modifying and head nouns of noun–noun expressions and healthy elderly individuals exhibit a similar but milder deficit. We also predicted that fewer integrational definition selections among the AD subjects would give rise to relatively greater selections of the alternative, but semantically plausible, associative definitions. Contrary to this expectation, there was a significant increase in the erroneous foil definition selections among these subjects when compared to the healthy elderly and young adult participants. However, foil definition selections were also higher in the healthy elderly subjects when compared to the young adults. These are intriguing results because they suggest that both AD and healthy elderly individuals have difficulty in syntactically parsing the modifying and head nouns of noun–noun compounds. Two plausible models have been put forward to account for conceptual combination processes (see Murphy, 2002; for a review). First, the concept specialization model (Cohen & Murphy, 1996) postulates that the concept of a head noun contains specific ‘‘empty slots’’ that can be filled by semantic properties of the modifier noun. Thus, the slot that is filled defines a relationship between the head noun and the modifying noun. Slots may relate to any property of the concept, such as habitat, color, shape, location, etc. For example, dog magazine refers to a magazine about dogs; thus, the ‘‘about’’ slot is filled by the modifier. Slot selection is predicated on the saliency of the features characterizing the modifier; the more salient a feature, the more likely it is that it will be selected to modify the concept. A second model, proposed by Wisniewski (1996), claims that conceptual combination may occur according to two interpretative processes. The first is integration, where, similarly to the previous model, the concept is defined as possessing a salient feature of the modifier. For example, a tiger lily is a lily which possesses the quality of being striped like a tiger. The second mechanism used for interpreting a noun–noun compound is through association, where the concept is taken to have some kind of contextual relation with the modifying noun. For example, a coffee table is a table where one drinks coffee. The selection between integration and association is influenced by the global shape similarity between the referents of the head and modifying nouns; the more similar they are in overall shape, the more likely a property of the modifying noun will be integrated into the concept of the head noun. For example, horse and cow are globally similar (they both have a similar configuration of head, legs, and torso), whereas horse and cup are not; thus, subjects would be more likely to select an integrative definition for horse cow than for horse cup. Congruently with these models, the difficulties that AD individuals have in processing integrational definitions of noun–noun concepts may arise as a result of a problem in either selecting the appropriate slot that captures the relation between a head and modifying noun,

V. Taler et al. / Brain and Cognition 59 (2005) 306–309

or identifying precise salient features that become part of the head nounÕs representation. Individuals with AD may also have difficulties in aligning the visual shape representations of head and modifying nouns prior to selecting relations and features that define the integrational noun–noun concept. Moreover, each of these problems may be related to difficulties in attentional control that would otherwise serve to support the selection, combination, or alignment of semantic and visual features when constructing noun–noun concepts online. Clearly, a detailed study examining the effects of each of these factors on conceptual combination performance in AD subjects is warranted.

309

References Cohen, B., & Murphy, G. L. (1996). Models of concepts. Cognitive Science, 8, 27–58. LaBerge, E., Balota, D., Storandt, M., & Smith, D. (1992). An analysis of confrontation naming errors in senile dementia of the Alzheimer types. Neuropsychology, 6(1), 77–95. Murphy, G. L. (2002). The big book of concepts. Cambridge, MA: MIT Press. Whatmough, C., & Chertkow, H. (2002). Category-specific recognition impairments in AlzheimerÕs disease. In E. Forde & G. Humphreys (Eds.), Category-specificity in mind and brain. Hove East Sussex: Psychology Press. Wisniewski, E. J. (1996). Construal and similarity in conceptual combination. Journal of Memory and Language, 35(3), 434–453.