Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Contributions from autobiographical memory to lexical representation McKinnon, Audrey Alison 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1997-0037.pdf [ 3.72MB ]
JSON: 831-1.0087593.json
JSON-LD: 831-1.0087593-ld.json
RDF/XML (Pretty): 831-1.0087593-rdf.xml
RDF/JSON: 831-1.0087593-rdf.json
Turtle: 831-1.0087593-turtle.txt
N-Triples: 831-1.0087593-rdf-ntriples.txt
Original Record: 831-1.0087593-source.json
Full Text

Full Text

CONTRIBUTIONS FROM AUTOBIOGRAPHICAL M E M O R Y TO LEXICAL REPRESENTATION by AUDREY ALISON MCKTNNON B.A., The University of Ottawa, 1994 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E DEGREE OF MASTER OF SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES The School of Audiology and Speech Sciences We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA DECEMBER 1996 © Audrey Alison McKinnon, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be a(lowed without my written permission. Department of IVAALCAQQU 9 The University of British Columbia Vancouver, Canada Date l L 50- (1H -A C J L A DE-6 (2/88) 11 ABSTRACT The purpose of this study was to investigate the nature of the interaction between autobiographical memory and lexical representation in semantic memory. To evaluate whether improved lexical access could be due to a reduction of cognitive load, the effects of discourse type on lexical specificity were also examined. The data were obtained from conversations with a subject diagnosed with probable Alzheimer's disease. Measures of informativeness using the Correct Information Unit analysis (Nicholas & Brookshire, 1993) and lexical specificity using a noun specificity analysis, noun classification analysis and unique adjective analysis were used to analyze texts representing recent and remote, personal and public memories, and across discourse types (a procedure, a fairytale, events, descriptions). The results showed that the personal memory texts had the greatest informativeness and lexical specificity across memory types. The remote-recent memory distinction varied, but the remote memories generally led to better lexical specificity than the recent ones. Finally, discourse type did not have an effect on lexical specificity, but did not have an effect on informativeness. The conclusion was that there are connections between semantic memory and autobiographical that lead to improved access and to use of lexically specific words. i i i T A B L E OF CONTENTS Abstract ii Table of Contents iii List of Tables v Acknowledgements vi CHAPTER ONE INTRODUCTION 1 Reason for Study 1 Introduction to Literature Review 1 Literature Review 4 Semantic Memory and Lexical Retrieval 5 The Nature of the Semantic Memory Impairment in A D 8 Episodic Memory and Autobiographical Memory 11 Discourse Characteristics 16 Schematic Structure 18 Discourse Characteristics in A D 22 Summary 23 Hypotheses 25 CHAPTER TWO: METHOD 2 6 The Subject 26 Procedures: Selection of Conversational Topics 28 Data Collection 29 Audiotape Transcription 30 Analysis: Transcript Analysis 31 Evaluation of Noun Specificity 31 Noun Specificity Survey 32 Classes of Nouns 33 Adjective Uniqueness 34 Analyses 35 CHAPTER THREE: RESULTS 3 6 Correct Unit Information Analysis 36 Lexical Specificity Across Memory Dimensions: 38 Noun Specificity Analysis 38 Noun Classification Analysis 40 Comparison of the Noun Specificity and Noun Classification Analyses 41 Descriptive Adjectives 42 Summary of Noun and Adjective Findings in Terms of Memory Dimensions 42 Lexical Specificity Across Discourse Types 43 A Comparison of Effects of Memory Dimensions and Discourse Types on Informativeness and Lexical Specifcity: 44 Memory Dimensions 44 Discourse Types 45 CHAPTER FOUR: DISCUSSION 46 CHAPTER FIVE: CONCLUSION 53 BIBLIOGRAPHY AND REFERENCES CITED 55 i v Appendix 1 Assessment Results for K D 62 Appendix 2 Noun Specificity Survey 63 Appendix 3 Nouns with a Standard Deviation Less than 1.0 66 Appendix 4 Descriptive Adjective List for Each Topic 68 LIST OF TABLES Table Page 2.1: Decontextualized Topics 28 2.2: Contextulaized Topics 29 3.1: Total Number of Utterances, Words, CIUs, and Percent CIUs in Transcripts 37 3.2: Measure of Informativeness by Topic 38 3.3: Noun Specificity Percentages for the Memory Dimensions 39 3.4: Percentage of Specific Nouns from the Noun Classification 39 3.5: Percentage of Unique Adjectives 42 3.6: Consistency of Ordering of Memory Dimensions with the Predicted Gradient Across Topics 42 3.7: Ordering of Discourse Types by Measure of Lexical Specificity 43 3.8: Informativeness and Lexical Specificity Means Related to Memory Dimensions 44 3.9: Informativeness and Lexical Specificity Means for Discourse Types 45 ACKNOWLEDGMENTS Many thanks to K D and her caregivers for their commitment to this project. Thanks to K D for sharing her stories, interesting viewpoints, and her tea and cookies. Tremendous gratitude and thanks to my husband, Stuart McKinnon, for his unwavering belief, encouragement, support, and fabulous meals. Finally, thanks go out to all the people who guided me in the completion of this project. 1 CHAPTER ONE PNTRODUCTION Reason for Study The impetus for this study arose as a follow-up to a case study completed by Palm and Purves (1996). In their study, they described an elderly woman, K D , who was diagnosed with probable Alzheimer's disease. K D requested language therapy to help her cope with her difficulties with word-retrieval and memory loss, which impaired her ability to communicate effectively. Palm (1996) developed a theory-based intervention plan targeting the discourse problems experienced by K D . In her discussions, Palm (1996) suggested the need for further investigation into the question of whether or not the processing of an autobiographical memory could result in improved lexical access of those items specifically associated with the memory. In this study, an attempt will be made to answer the question of what the nature of the interaction between autobiographical memory and lexical representation in semantic memory is. Introduction to Literature Review Memory and communication deficits are two critical features associated with Alzheimer's disease (AD). The memory deficits can manifest themselves as difficulties with orientation to time, place and person, the ability to learn new information, and the capacity to retain new material. Memory impairment can result in language comprehension problems and pragmatic deficits such as poor topic maintenance and discourse incoherence (Kempler, 1991). Initially, the communication deficit is often expressed as difficulties with word-finding. This is especially apparent in reduced ability to retrieve substantive words, which leads to the appearance of semantic paraphasias and circumlocutions (Kempler, 1991; Huff, 1990). It has been clinically observed, however, that the word-finding difficulties experienced by many individuals with A D are not uniform across all topics of conversation in which they engage. For example, some topics of conversation are characterized by the presence of a large number 2 of vague terms and nonspecific words, while other topics are more informative and lexically specific. The increased use of general lexemes reflects a lack of specificity, which reduces the informativeness of discourse (Palm, 1996). A major question arising from this observation is: What are potential factors contributing to lexical access and subsequent lexical retrieval? Several bodies of literature will be examined in an attempt to answer the above question. First, a basic understanding of how concepts and their lexical items are represented in the brain is needed. For this we look to the field of study concerned with semantic memory. Semantic memory contains a permanent representation of our knowledge of concepts and words and their meanings (Chertkow & Bub, 1990). Within our semantic memory is knowledge that is overlearned and decontextualized to the setting in which learning took place. We will look at the arrangement of the semantic memory system and the types of connections that exist within it. Next will be a discussion regarding the nature of lexical processes such as access and retrieval. We will look at theories that explain how concepts activate words or words activate concepts with the end result of lexical retrieval. An overview of logogen theory will help explain one particular search process postulated for lexical retrieval. Compromise of conceptual knowledge in the semantic system is evident from the outset of A D in the form of word-finding difficulties. The nature of this semantic impairment has been studied by various researchers. Certain evidence has led some researchers to postulate that the semantic impairment in A D reflects a loss of information or a disruption in the integrity of the semantic memory system. Other evidence has been interpreted to demonstrate a problem with access processes into semantic memory, while all the component parts remain intact. As will be seen, it is still unresolved as to whether the linguistic and memory deficits in A D discourse reflect a loss of semantic knowledge or a breakdown in the mechanism's accessing that knowledge (Palm & Purves, 1996). The associations and connections contained within semantic memory may not be sufficient to explain the observation of improved word-retrieval experienced by individuals with A D as a function of the topic of discussion. Clinically, topics that seem to enjoy enhanced 3 lexical retrieval are those which are personally and/or emotionally relevant to the individual. The system in which these memories are processed is autobiographical memory, a subset of episodic memory. These systems contain knowledge contextualized to time, place and personal experience. What is the nature of these memory systems as they apply to normal and A D populations? One finding is that memories with personal relevance and high emotional import are better remembered by both normals and A D subjects. Another is that events and experiences associated with a certain period in a person's lifespan seem to be better recalled than memories from other life periods. These will be explored in greater detail to determine whether there exists evidence of a relationship between autobiographical memory and lexical representation in semantic memory. Because lexical retrieval most often occurs in the context of discourse, it is necessary to review relevant discourse properties as they relate to adults with intact language and cognition and how they are affected by A D . Discourse properties provide us with organizational frameworks for discourse planning and production. They are cognitive knowledge structures which regulate how we construct a particular discourse such as, for example, a narrative or the recounting of an event. Pragmatic deficits affecting discourse in A D include inappropriate initiation of new topics (Kempler & Zelinski, 1994), frequent exophoric reference, general referencing difficulties, production of inappropriate or irrelevant details, and failure to convey important information (Palm & Purves, 1996). These are all factors implicated in reducing discourse informativeness. Other aspects of the pragmatic domain are relatively well-preserved, especially in social interactions, like conversations. For example, individuals with A D retain the ability to take appropriate conversational turns and often produce socially ritualized parts of conversation at the right time and with the right affect (Kempler & Zelinski, 1994). A review of the literature as outlined above leads to the identification of possible pertinent variables associated with lexical retrieval of substantive words. These are discussed in the research questions. The thesis is organized in the following way: Chapter one provides a 4 literature review and the research questions; Chapter two contains the method of the study, including subject information, data collection, orthographic transcription procedures, and the analyses protocols; Chapter three comprises the results of the study, and in chapter four the results are discussed with reference to the research questions and the relevant literature. Literature Review This section provides a review of research on memory and discourse as they pertain to individuals with A D . Memory is not a unitary system. It comprises several component parts, each of which perform particular tasks. Relevant to this project are semantic memory and episodic memory, with the latter further analyzed to include an autobiographical memory component. Semantic memory contains decontextualized knowledge. This includes memories for facts about objects, people, events, rules, and scripts. Knowledge in semantic memory is consciously accessible, modifiable and open to manipulation (Nebes, 1989, p.377; Conway, 1990, p.3; van der Linden et al., 1992, p.484). Episodic memory's role was described by Tulving (1983) to be the processing of information tied to a specific episode or event that has been personally experienced and can be identified as having happened at a particular time and place (i.e., it is contextualized). Episodic memory processes information in a spatial-temporal context and acts as an autobiographical record of the unique events in an individual's experiences (Nebes, 1989; Bayles & Kaszniak, 1987; Tulving, 1983). For people with A D , impairments of episodic memory can affect the ability to form new memories and the ability to recall memories formed before A D onset (Osimani & Freedman, 1991). Autobiographical memory is the part of episodic memory concerned with record-keeping of facts of an individual's personal past (Conway, 1990). It stores biographical information about one's life experiences (Robinson, 1992; Conway, 1990). In most cases, these are episodic memories, but some of what we know about ourselves is also stored as semantic knowledge. For example, we know certain facts about our past without being able to place and date when the 5 specific memory was acquired (Morris & Gruneberg, 1994). Here we see the notion of an interrelationship between autobiographical memory and semantic memory. Autobiographical memory represents the long-term recollection of general features of an event, personal interpretations of complex events, and some recall of a few specific details (Conway, 1990; van der Linden et al., 1991). The interpretations are based on actual occurrences and cognitive integration of events. They also include sensory and perceptual features as well as information about current thoughts, wishes, and motivations, all of which can be closely related to other memories. Conway (1990) observed that memories of emotionally and personally significant events were represented in greater detail than those of unremarkable routine events. After that brief introduction, let us now take a more detailed look at these different memory systems. Semantic Memory and Lexical Retrieval In 1969, Collins and Quillian described the hierarchical-network model as a system of semantic memory organization. This model had a significant influence on subsequent accounts of many patterns of memory and language disorders seen in populations with nonfocal brain damage such as the dementias (Chang, 1986). According to the hierarchical-network model, knowledge is stored in long-term memory as an associative or semantic network. The network has nodes corresponding to word concepts (e.g., "bird", "feather"), and links or arcs corresponding to conceptual relations that connect the concepts (e.g., "is a..." describes a subset-superset relation; "has, can, is" describe attribute relations). The links specify the relationship between concepts and show properties of a node/noun concept (e.g., 'lias whiskers", "purrs" to the node/noun "cat") (Morris & Gruneberg, 1994). Noun concepts are arranged hierarchically, with superordinate concepts linked only with their most direct subordinates (Chang, 1986, p.207). Then in 1975, Collins and Loftus refined the hierarchical-network model. Their new version was called the spreading-activation model. In this model, there are a larger variety of links between concepts. The links have different criterialities or numbers indicating how 6 important they are to the meanings of the concepts they connect. As well, there are different accessibilities, strengths and travel times between the links. The more properties that two concepts have in common, the more possibilities there are for linking two nodes and the more related in memory the two concepts will generally be (Morris & Gruneberg, 1994; Chang, 1986) . According to the spreading-activation model, paths are found in the semantic network through the quasineurological process of activation or priming. Activation starts from presented words associated with concepts, then fans out in a decreasing gradient along a link (Morris & Gruneberg, 1994; Chang, 1986). In other words, a node in the network is activated as an individual is engaged in processing incoming information and when retrieving information during a task (e.g., in conversation). Activation from that node automatically spreads out through the associative network to corresponding nodes in semantic memory. This activation is maintained over a period of time. The result is a brief increase in the accessibility of related concepts. The additional increase can have a facilitative effect on subsequent lexical processes as the related concepts are easier to access. Nodes that are highly associated with a concept have a stronger activation than those with weaker associations (Morris & Gruneberg, 1994; Nebes, 1989; Smith, Butters & Granholm, 1988). A concept, as it has been discussed above, depicts general classes or categories of objects and events. Concepts are stored in a hierarchy of categories (Bayles & Kaszniak, 1987) . The hierarchical organization consists of a superordinate category (e.g., "animal"), with more specific attributes (e.g., "has four legs") further down, followed by lower order or subordinate categories that are more specific in their reference (e.g., "dog", "tiger") (Huppert, 1994). Concepts have interconnecting nodes with other concepts in a variety of relationships. These include, but are not limited to, membership in a common category (e.g., knife-spoon), functional relationships (e.g., knife-cut), and property relationships (e.g., knife-sharp) (Morris & Gruneberg, 1994; Nebes, 1989). 7 With a basic understanding of conceptual representation in semantic memory in place, let us now look at the retrieval of lexical items from semantic memory. At least two different attempts have been made to describe the search process through the lexicon. In a serial search model, lexical information taken from a stimulus is sequentially compared with an individual's internal lexicon until an appropriate item is selected. For this system to work, each item in the lexicon would need to be searched, a process that would take a long time to implement and could not account for the speed at which lexical access occurs. Parallel or direct access theories, on the other hand, assume that incoming information can be simultaneously compared with all the relevant information in the lexicon (Gordon, 1983). The logogen theory, in particular, is a good example of a parallel search model. Logogen theory was introduced by Morton in 1969 as a model of access in both language comprehension and production (Levelt, 1989). Morton characterized the mental lexicon and the lexical items within it as consisting of a collection of logogens, where a logogen is described as an information-collecting unit that activates lexical items (Caplan, 1992). In speech production, a logogen retrieves its information from a semantically active cognitive system. Each logogen is sensitive to specific information and is activated whenever the number of features feeding into it exceeds some threshold value. This is a semantic task that results in lexical selection of one appropriate word. The effect of incomplete activation of a logogen would be an inability to adequately retrieve a lexical item. On the surface, this could appear as a word-finding problem or a tip-of-the-tongue experience. A logogen's activation level remains high for some time after activation has occurred (Caplan, 1992). Similar to concepts in semantic memory, it is likely that connections exist between different logogens, with the activation of one increasing the thresholds of those associated with it, hence making them more readily available for subsequent lexical selection as needed. Just as there are connections between concepts and their lexemes in semantic memory, there are different connections among the lexemes in the mental lexicon. There are intrinsic relations between a word and the superordinate category to which it belongs (e.g., "dog" to 8 "animal"), a word and another that is at the same level (e.g., "cat" to "dog"), or a word and a near synonym (e.g., "close" to "near") (Levelt, 1989). These sets of meaning-related words are called semantic fields. There can also be morphological connections between lexical entries with the same morphological stem, and phonological and syntactic connections. The connections between these lexical entries may be either direct or mediated through spreading activation in semantic memory. Levelt (1989) states that associative relations are joined according to the frequency of their co-occurence in the language (e.g., "war" and "death"). However, there are also complex conceptual relations that link these words together (e.g., wars generally result in the death of many people). The relations between the lexical entries in semantic memory are crucial to the idea of spreading activation. As will be discussed later, lexical associations within semantic memory may be affected by input from different memory systems. The Nature of the Semantic Memory Impairment in A D Certain characteristics of language in A D have led to questions regarding the nature of the semantic memory impairment. There is, for example, an apparent breakdown of generative (where a subject is asked to name as many things as he/she can belonging to a particular category, e.g., animals, fruit) and confrontation naming (i.e., the subject is shown a picture or object and asked to name it). At the sentence level, there is an increased number of circumlocutions, an overuse of indefinite pronouns, demonstratives (e.g., "here", "there") and exophoric reference (e.g., "this"), an absence of content words, and inappropriate use of empty words (e.g., "thing", "stuff) and phrases (e.g., "something like that") (Nebes, 1989; Salmon, Heindel & Butters, 1991; Kempler & Zelinski, 1994; Osimani & Freedman, 1995). At the discourse level, these characteristics lead to decreased informativeness and increased incoherence, verbosity, tangentiality and ideational perseveration (Ulatowska & Chapman, 1991; Kempler & Zelinski, 1994; Osimani & Friedman, 1995). Some researchers (Bayles & Kazniak, 1987) have suggested that the source of this impairment is semantic memory. Based 9 on what is known about semantic memory, researchers devised different tests to tap into its function. Included are tests of category-word generation, associative priming, confrontation naming, completion of highly structured sentences, judgment about the sequence of procedural tasks, generation of ideas in response to questions, and knowledge of vocabulary and word associations (Huppert, 1994; Salmon, Heindel & Butters, 1991; Nebes, 1990; Swihart & Pirozzolo, 1988). The results from these tests have led to conflicting descriptions regarding the nature of the semantic memory impairment. Some accounts postulate that the impairment is caused by a loss of semantic knowledge (i.e., dissolution of the semantic network). Others characterize the problem as one of accessibility to semantic knowledge. In the accessibility view, semantic network is intact, but the processes allowing access into it are no longer functioning optimally. Both groups rely on the model of spreading activation through the semantic memory network to make and test their hypotheses. Evidence pointing to a permanent deterioration of concepts in semantic knowledge comes from findings of consistent responses in semantic tasks. Chertkow and Bub (1990) studied the integrity of semantic memory in ten A D subjects through tests of item specificity (i.e., confrontation naming), cueing effects, and verbal fluency. They found an item-to-item correspondence between loss of name production and comprehension which they tied to a storage deficit. For verbal fluency, their A D subjects had a marked impairment in generating semantic category exemplars (e.g., animals, fruits, tools). Semantic cues were found to be ineffective for picture naming. Finally, superordinate information was found to be preserved, but detailed information was not available. There was a high consistency of naming errors over time as well as over tasks. That is, when a subject was unable to name an item from a picture, semantic cueing was not helpful and the item was almost never offered as a category exemplar on the verbal fluency test. Warrington (1975) also found that A D subjects were unable to provide specific attributive and associative information, while their superordinate information was relatively well preserved. This was taken as an indication that destruction of 10 semantic memory followed a stable route through its hierarchical arrangement. A study by Huff, Corkin and Growden (1986) found that superordinate knowledge was preserved for their A D subjects. They also found stable item-to-item correspondences in naming impairment. That is, if an item was not named in one task it was not subsequently named in another. Support for impaired access to semantic memory comes from inconsistency of subjects' performance on semantic tasks and error analyses. The inconsistency suggests that the semantic memory network is intact, while the activation between conceptual knowledge and lexical representation is impaired. Most lexical substitutions made by A D individuals at the single-word level are semantically related to a presented stimulus, such as the name of its category or another member of the same category (Bayles, Boone, Tomoeda, Slauson & Kaszniak, 1989; Bayles, 1985). Other incorrect responses include providing the function of the object or the context in which it is found (Nebes, 1989). Flicker, Ferris, Crook, & Bartus's (1987) A D subjects had difficulty describing the use of an object; however, when asked to pick from a choice of items those objects useful in a particular chore, they were successful. Huppert (1994) found that errors made in cueing tasks were related to the same semantic field as the cue. Deficits with the lexical search process, rather than a problem with the semantic association, were implicated by Nebes (1985,1989) for impairments on associative tasks. This was based on the observation of preserved priming effects when the semantic prime was partially automatic and did not require the same amount of attentional resources and volitional control as object naming. Chertkow and Bub (1990) also found that semantic priming was effective for their A D subjects. They attributed this finding to several possibilities: different naming and lexical decision tasks were used for cueing and priming procedures, priming uses on-line measures of reaction time while cueing only measures success or failure in carrying out a naming task. They did not take the priming effect as evidence of an intact semantic memory system for their A D subjects. There have also been similarities found in the performance of A D subjects and normal controls which have furthered the impaired access arguments. For both groups, words with 11 lower frequency were more difficult to name, increasing abstractness led to increased problems with providing definitions, and similar positive effects from priming in naming experiments were found (Nebes, 1989). Proponents of the impaired access view consider that processes making connections between conceptual and lexical representations are at fault. The impairment may be one of impaired activation of the connections, impaired ability to use semantic information, or impaired lexical search processes (Nebes, 1989). As Palm and Purves (1996) point out, no one explanation has provided an adequate account of the nature of the semantic impairment. The evidence is conflicting and there are conflicting interpretations of the same evidence. They also discuss the further possibility of differences between individuals with A D , so that each explanation could apply to different sub-groups, and finally that "both problems [may] occur, either simultaneously or sequentially" (Palm & Purves, 1996, p. 157). For example, phonemic cueing was found to be more beneficial for subjects with mild A D than those in the more advanced stages of the disease (Neils, Brennan, Cole, Boiler & Gerdeman, 1988). The suggestion made by Neils et al. was that in the course of A D , deficits of retrieval of a word's lexical representation may precede those of semantic representation. Without knowing the extent that these processes are distinct and represent different cognitive mechanisms, it is not possible to resolve the question of the nature of the semantic impairment. The relevance of studying different accounts for word-finding deficits in A D is to emphasize the importance of clarifying, as much as possible, the nature of the deficit for each individual with A D . Differences in people with A D may limit the generalizability of findings in my study. Episodic Memory and Autobiographical Memory Snowden, Griffiths and Neary (1994) were interested in the role that personal experience played in meaning preservation and access to semantic memory. These researchers noticed that vocabulary was relatively spared for subjects with semantic dementia when engaged in conversations triggering concepts stored in autobiographical memory (i.e., 12 personally experienced events). They gave subjects tasks related to understanding peoples' names, place names and object identification. Their subjects performed consistently better for current, personally relevant names and objects than for nonpersonal ones. This effect was not seen for the Alzheimer control group whose members had a profound memory impairment for daily events, some difficulties with verbal expression, but no clear semantic deficit. The importance of this finding is the implication of an existing link between autobiographical memory and semantic memory. In fact, Snowden et al. (1994) concluded that "there is a continual interaction between one's autobiography and semantics" (p.283) where autobiographical experience plays a role in the maintenance of meaning. Their findings, then, point to the existence of an interrelationship between episodic and semantic knowledge which has not been previously discussed in this way. From the above observations, it is clear that the role of episodic memory, and more specifically autobiographical memory, in areas of lexical representation and selection needs to be further explored. A subset of episodic memory is autobiographical memory. Within autobiographical memory, then, evidence points to a three-layered structure in which knowledge is represented: (1) "lifetime periods represent knowledge about thematically distinct periods in a person's life" spanning periods of years and decades; (2) "generalized events represent thematic knowledge", but the knowledge is localized to shorter time spans and general events; (3) finally, "event specific knowledge represents sensory-perceptual knowledge spanning periods of seconds, minutes, or hours" (Conway and Rubin, 1993, p.104). Autobiographical memory, then, is organized in a hierarchical fashion. The highest level in the hierarchy contains knowledge about the chronology of one's life. It consists of extended timelines that represent thematic knowledge (e.g., goals, people, events) relating to a period. The periods are temporally marked with beginnings and endings that may overlap. The themes can link together whole sets of memories, sometimes across an individual's lifespan (Conway & Rubin, 1993). Autobiographical memory also has an index of representations that summarizes events in shorter time periods within a lifetime. These are general events. At the lowest level of the 13 hierarchy, specific autobiographical memories are indexed directly by representations of event specific knowledge. There are potentially multiple levels of descriptions of any event that are stored in the hierarchical structure (Conway, 1990). Temporal patterning is one way to organize experience when reproducing an autobiographical memory (Fromholt & Larsen, 1992). Goals may be another central feature of organization. An event can be seen as action directed toward a goal such as finishing high school to get into university. Goals can be scaffolded and link diverse events over extended periods of time (Robinson, 1992). Emotions also play a role in memory formation. When we process concepts associated with an emotional experience (e.g., "love") and not with actual memory retrieval, it may be that autobiographical memory becomes highly available, leading to spontaneous and involuntary retrieval of memories (Conway, 1990). Autobiographical memory, then, involves issues relating to the self, personally relevant goals, and, ultimately, personal meanings (Conway & Rubin, 1993). Fromholt and Larsen (1992) looked at how memories are distributed over the lifetime. They interviewed both normal and A D populations using a free narrative method to elicit their data. They found that the two populations had a similar distribution of memories with a "pronounced early peak and a smaller recency increase" (p.417). This peak was termed a "reminiscence effect or bump," which occurred around early adulthood (between 15 and 25 years old) for people over 40; It corresponded to personally important, transitional events in the individual's life, especially for those memories that were vivid (Fromholt & Larsen, 1992). The reminiscence bump most likely reflects a critical period in an individual's development related to the emergence of a stable and enduring self-concept (Conway & Rubin, 1993). Fromholt and Larsen (1992) concluded that the pattern of remembering for the A D population was not the result of a sparing of early memories, because it was similar to the pattern of remembering produced in normals. Sagar, Sullivan, and Corkin (1991) also looked at the distribution of memories retrieved. They presented their normal and A D subjects with noun cues and asked them to relate an autobiographical memory to each cue. In this study, the 14 majority of memories retrieved were from the subjects' most recent 20-year period, with a secondary peak corresponding to early adulthood. These findings are opposite from those of Fromholt and Larsen (1992) and may reflect the effect of different task presentation on memory retrieval. In conjunction with personal memory retrieval, researchers have looked at how public memories are distributed. Public memories are complex memory structures formed from knowledge of on-going news stories, and augmented with new information as relevant events unfold (Brown, 1990). They are held within semantic memory and contain our knowledge for the faces and names of famous people and public events. Interestingly, it has been found that public memories of historical events cluster thematically and the same way that personal memories do (Brown, 1990). People maintain many temporally overlapping public and personal periods in memory. Brown (1990) postulated the existence of contextual links directly connecting facts derived from news stories to facts about the personal context in which they were acquired. This finding has been supported by other researchers (Fromholt & Larsen, 1992; van der Linden et al., 1992) who concluded it likely that autobiographical memory and semantic memory intersect with each other, based on the observation that many public events are associated with a personal context, such as where the individual heard about the event and how it affected him/her. But do public memories, which have different characteristics from autobiographical memories, show the same pattern of remembering as described above? In fact, Kopelman (1989) and Fromholt and Larsen (1992) both found a similar distribution (i.e., reminiscence peak) for public historical memories as that seen with autobiographical memories for a group of normal and A D subjects. The public memories also had a pronounced early peak and a smaller recency increase. The A D subjects recalled proportionately more information for remote public events which occurred during late adolescence and early adulthood. They showed a gentle temporal gradient in remote memory loss, with older memories being better recalled than more recent ones. There are several possible explanations which have been proposed for the 15 existence of this effect. First, the memory traces for remote events were laid down before the onset of the disease. This, however, does not explain the effect seen for the normal control group. Second, the remote events may have been subject to more rehearsal and, as a result, been more salient. Third, the trace of an episode loses its associations with time and context, and assumes the character of semantic memory which would suggest that the autobiographical memory system is not being tapped at all in the retrieval of these memories (Huppert, 1994; Kopelman, 1989; Sagar et al., 1988). According to this interpretation, semantic memory, would need to be relatively spared. However, it is more probable that the explanation lies in the view that because one's political, social or generational identity is being formed during this period (Conway & Rubin, 1993), the memory is encoded in a stronger, more elaborate fashion. Thus, more links or associations are formed for concepts related to these memories. . As suggested earlier, if the autobiographical memory system is connected to the semantic memory system, stronger links for spreading activation in semantic memory and/or a greater number of features for pushing a logogen's threshold are also available for lexical access. Additional research has shown the importance of autobiographical memory for lexical retrieval for people with A D . For example, Conway (1990) found that an event characterized by emotional experience or with a lot of personal import was remembered more clearly and persistently. Personal significance has been found to be important in recall of events by both normals and A D subjects. Further support for the notion that experiences stored in autobiographical memory may have a special status comes from a study by Keenan and Baillet (1980). Their A D subjects were first asked to judge a list of traits as characteristic of themselves or a close friend. Then the subjects evaluated whether an adjective described a person they knew well versus someone who was not as well known. Their A D subjects were quicker at making evaluations about adjectives they had earlier judged to be characteristic of themselves or a close friend. Keenan and Baillet (1980) also found that highly self-relevant information was processed more quickly and retained better than factual knowledge with low personal import. The researchers suggested that self-referring factual knowledge is represented 16 in a different manner from nonreferring factual knowledge. The question being posed in this study concerns the nature of this different representation. One answer may be that autobiographical memories are represented lexically. Support for this view comes from research with bilingual A D subjects. Hyltenstam and Stroud's (1989) subject G M had German as a first language and at the age of 45 acquired Swedish which he used primarily at work, when speaking to customers and for business correspondence. When G M discussed experiences in early childhood and early adulthood he spoke only in German. The researchers concluded that GM's long-term memories were discussed in the language in which they were encoded. In other words, the personal memories were lexically represented in German when they were initially encoded and they were retrieved in that language as well. Discourse Characteristics Up to this point, the representation of language has been described. However, immediate context also has relevance for lexical retrieval. Evidence from lexical semantic research has demonstrated that context plays a role in lexical usage. One notable example is termed imitation phenomenon. This is the observation of speaker preference for recently used (either by self or conversational partner) words even when the words are semantically non-discriminative (Levelt, 1992). This cannot be explained solely on the basis of semantic reasons for word selection. In fact, it could be that a recently produced word has a high activation level in the semantic network. Thus, episodic memory is not necessarily implicated. Another possibility is that the environmental context in which a particular discourse is occurring also helps to maintain higher activation levels, making already used words readily available and, therefore, easier to use. Empty speech in discourse has been noted often by conversational partners of people with A D . Nicholas, Obler, Albert and Helm-Estabrooks (1985) attempted to characterize the composition of empty speech in A D . Their subjects gave descriptions of a picture stimulus. The descriptions were analyzed for words or phrases that detracted from or did not contribute 17 to a coherent description. The A D subjects produced significantly fewer content elements than did normal controls. They also had more deictic terms (e.g., "this", "here"), semantic paraphasias (e.g., "chair" for "stool"), pronouns without antecedents, repetitions, and indefinite terms (e.g., "thing", "stuff). The researchers also examined whether there existed a correlation between scores on the Boston Naming Test and empty speech measures, and between Boston Naming Test scores and thematic content measures. For the A D subjects, there was a significant correlation between naming deficit and production of indefinite terms and reduced thematic content elements. Nicholas et al. (1985) concluded that, overall, their results did not support a naming deficit underlying discourse emptiness. Their results do show that there is a discrepancy between informativeness and lexical specificity. The one measure of lexical specificity included in their list of factors affecting discourse informativeness was indefinite terms. It had a significant correlation with naming score on the Boston Naming Test, which gives a measure of lexical retrieval. The semantic representations governing discourse have been investigated to a lesser degree than the concept or word level for disruption in A D (Palm & Purves, 1996). Discourse is made up of conceptual units that enable us to process and understand information in our environment. It also has an internal organizational framework independent of content which facilitates memory processes during production and comprehension processes (Ska & Guenard, 1993). At least three different levels of structure are thought to exist in discourse. First is "schematic" or superstructure. This is the highest level of discourse which is responsible for the flow of discourse topics. Next, is the macrostructure. At this level, utterance sequences sharing a common topic are grouped together in a cohesive, coherent manner and represent the core concepts being discussed. Finally, the microstructure is the level where sentences are related to each other locally. It is at the microstructure level that reference continuity is maintained (Caplan, 1992). These levels of organization are represented in semantic memory, but the content that is retrieved may have connections or associations with knowledge in other memory systems. 18 Schematic Structure In this next section, I will discuss schematic structure in more detail. Psychological researchers have talked about the construct of schema theory or what Mandler (1984) suggests we refer to as a schema framework as opposed to theory "since the principles subsumed under this view of the mind consist of very general beliefs about how this form of organization works" (p.l). Schemata are higher level, complex, organized mental representations. They function as ideational scaffolding in the organization and interpretation of experience (e.g., they are concerned with generic knowledge of specific domains or classes of events) (Hess, 1990; Brown and Yule, 1983). A schema, then, is general knowledge a person possesses regarding a particular domain or topic. Schematic knowledge is thought to facilitate retention of information because individuals attempt to understand events in terms of an organizational framework which allows for information processing, storage and, eventually, retrieval (Hess, 1990). Different schemata exist with which we organize our experiences. They are related to the topic or type of knowledge being encountered. I will describe four different schemata related to the topics of conversation that K D 1 was asked to discuss in this study. They include: event schemata, script schemata (procedural knowledge), person schemata (descriptions of people), and narrative schemata. Event schemata are hierarchically organized units that describe generalized knowledge about an event sequence. The schemata consist of basic components called event units which specify an event's properties: the action taking place, the identity/role of participants, the location of the action, and one or two salient details. Some of the components are directly linked with corresponding concepts in semantic memory which serve as retrieval indices (Brown, 1990). Also contained within the event schema is information about what will or what is likely to happen in a given situation and often the order in which events will take place. The hierarchy consists of general event classes that contain more specific events within them 1 K D is the subject with Alzheimer's disease who was interviewed for this study. 19 (e.g., going on a trip has nested within it a particular vacation the individual went on). The content of these schemata is contextualized to the spatial and temporal context in which they were formed. Of importance is the notion that events in this schemata are related temporally through causal (obligatory), enabling, or arbitrary connections (Mandler, 1984). Associations exist between elements at any given level of the hierarchy, with activation of one component resulting in activation of other components at that level and ultimately the entire structure (Hess, 1990). This view is very similar to that of spreading activation theory. A script schema is a subset of event schemata. Scripts are stereotypical knowledge structures or generic information people have acquired about common routines and/or activities. A restaurant script is one example. It usually consists of the following series of scenes: entering a restaurant, ordering, eating, paying and leaving (Mandler, 1984). Although the content of the script depends on a particular occasion, the procedure remains relatively stable. The main difference between an event schema and scriptal knowledge is that the latter is strongly tied to a specific activity or procedure and has sequential ordering constraints. The components of the script schema may be connected through temporal or enabling (e.g., getting your meal allows you to eat) relations (Hess, 1990; Mandler, 1984). Script schemata are more highly decontextualized knowledge structures than event schemata, since the many experiences which led to the formation of the framework are no longer available for reference. To distinguish the two more clearly, a script schema would tell you what steps are necessary in a certain situation such as going to the bank. The event schema would organize your experiences of what actually occurred during a particular visit to the bank. Thus, the event schema is personally tied to an experience. In a person description task, the knowledge structure employed is a person schema. This schematic structure contains knowledge about traits and stereotypical descriptions. Trait information is thought to be represented according to expectations about characteristics and behaviours associated with a particular trait. Conceivably, if a person demonstrates a particular trait (e.g., kind, strong), his/her name will be linked through an associative connection with the 20 trait and other traits he/she is perceived as possessing. Assuming this is true, then, it is not surprising that people process more extensively and have better recall for highly descriptive information about an individual than less descriptive information (Hess, 1990). As described earlier, Keenan and Baillet (1980) looked at memory for personally and socially significant events. They found that subjects made faster decisions about adjectives they had previously judged as referring to themselves or close friends. They assumed that "when a specific evaluation (i.e., an adjective) is already stored with a person schema, its storage is so intimately connected to the particular information (i.e., an event) from which it was derived that access of the evaluation automatically causes this supporting information to also be activated" (p.660). Finally, when telling a story a narrative/story schema is used, especially for a well-known fairytale such as Cinderella. The content of a fairytale can be said to exist in semantic memory since, in most cases, the story is retold and heard so often it takes on a lexical representation in memory. A narrative/story is described by Mandler (1984) as "a mental structure consisting of sets of expectations about the way in which stories proceed" (p. 18). The schema is said to play a role in facilitating memory for stories by creating an organization for knowledge (Stein & Glenn, 1982). Methods of quantifying this schema have been studied by different researchers. One proposed method is the story grammar.2 It is a rule-based system developed by Stein and Glenn (1982) to describe the underlying structure of stories (Mandler, 1984). These researchers believe that the story grammar contains "schematic knowledge that is used to guide encoding and retrieval" (p.256). Information within this schema is connected through causal and temporal relations (Stein & Glenn, 1982). 2 Brown and Yule (1980) discuss the fact that "for the proponents of story-grammars, there exists socio-culturally determined story-schema, which has a fixed conventional structure containing a fixed set of elements" (p.247). The main elements in Stein and Glenn's (1982) story grammar are setting, initiating event, internal response, attempt (to obtain a goal), consequence or resolution, and reaction or evaluation (p.258). 21 Palm and Purves (1996) suggest the possibility "that access to [the] knowledge structures [discussed above] facilitates language processing by reducing demands on the overall pool of cognitive resources (i.e., organizational demands are lessened) and allowing more resources to be directed to language processes such as, for example, lexical searches" (p. 159). If this is the case, then the facilitative effect of knowledge structures should be observable across different discourse types. In a study relevant to this claim, Wirth (1991) gave different narrative tasks to an aphasic subject. She found that the type of discourse appeared to have an effect on the total number of words and utterances produced by her subject. Narratives had the most words produced as compared to procedural and picture tasks. Her aphasic subject's production of complete sentences also varied with discourse type (i.e., procedural discourse had 90% complete sentences, fairytales had 80% complete sentences, memorable experiences had 70% complete sentences). Supporting evidence for the claim of a facilitative effect of knowledge structures is, however, difficult to evaluate. Several researchers who have looked at discourse production by older adults have found that different text types, methodological aspects and participant variability resulted in differences in research findings. Other factors influencing discourse production have been found to vary with the context and the conditions of discourse production (Ska & Joanette, 1996). Ska and Guenard (1993) observed that for their normal and A D subjects, performance and production of schema components were influenced by the nature of the narrative. For example, a sequence of pictures depicting a car accident constrained the story structure and resulted in more schema components being generated. Tfie telling of Little Red Riding Hood without any visual support resulted in fewer schema components and was more likely to be left incomplete by both normal and A D subjects. It appears that visual support may be another factor affecting discourse. However, it should be noted that sequential-ordering errors were still made with the picture sequence task. 22 Discourse Characteristics in A D Only a few studies have looked at whether the representation of superstructure and schematic knowledge in A D is intact or compromised in some way similar to other aspects of semantic memory. In a study of world knowledge, Weingarter et al. (1983) asked A D subjects to generate a script for an everyday situation such as eating in a restaurant. Their A D subjects generated fewer activities appropriate to a given situation and were less aware of the proper order for actions making up a complex behaviour. The serial ordering problems they exhibited may have indicated a loss or impairment in using temporal sequence information providing structure to such situations (Nebes, 1989). Using three different tasks to elicit narratives, Ska and Guenard (1993) performed a narrative analysis looking at the presence of story grammar components in the stories told by A D subjects. Their A D subjects provided fewer schema components, made sequential-order errors and produced more irrelevant propositions. Results indicated that A D story production was comparable to performance on script production. There were fewer schema components, more sequential-order errors and more irrelevant propositions. Ska and Guenard (1993) found that for the most constrained task presented, a sequence of seven pictures depicting a car accident, both normal and A D subjects generated more schema components. It was also noted that more irrelevant information was produced in less constrained tasks (i.e., Little Red Riding Hood with no visual support and a single picture of a bank robbery scene); There were, however, sequential-order errors even with the picture-sequence story. A D subjects were unable to make simultaneous use of both content and sequencing information. This can be taken as evidence that increases in a task's cognitive complexity led to decreases in communication competence. The results of the above studies are consistent with the conclusion that impairment exists in superstructure and schematic knowledge for people with A D . In a study of A D subjects' ability to produce narrative and procedural discourse, Ulatowska, Allard, Donnell, Bristow, Haynes, Flower, and North (1988) found that superstructure knowledge was 23 preserved for their subjects. The researchers felt, however, that this preservation of structural knowledge was a reflection of the relative simplicity of the presented tasks. It should be noted that similar tasks were used in studies by Ska and Guenard (1993) and Cardebat, Demonet and Doyon (1993), both of which found schematic knowledge to be disrupted. In the Cardebat, Demonet and Doyon (1993) study, A D subjects treated each picture in a picture-sequence as a closed narrative and found it difficult to build hierarchical relations between the elements in the pictures. From these research findings, it appears that for certain A D subjects there was some compromise of superstructural knowledge. It also appears that the nature of the discourse task has an effect on narrative production performance for A D subjects. The results from Ulatowska et al.'s (1988) study suggest that not all A D subjects have the same compromises in schematic knowledge. The import of this, in both clinical and research settings, is that A D patients need to be individually assessed for intact superstructure knowledge. Summary From the above discussions it is apparent that in studying lexical representation, two relevant factors need to be explored further: memory and discourse type. First, we are specifically interested in the nature of the interaction between episodic memory— and more specifically autobiographical memory— and lexical representation in semantic memory. Personal experience has been shown to have a facilitative role on memory recall in free recall and cue word tasks for both normal and A D subjects. As seen in the Keenan and Baillet (1980) study, lexical decisions about adjectives were faster for those that were self-referring or related to a close friend. It has also been suggested by researchers like Snowden, Griffiths and Neary (1994) that personal experience is relevant to the maintenance of meaning. This notion has additionally come up in studies of public memories where it was found that people encoded personal experiences such as, for example, where they were or what they were doing when they heard about an important public event (e.g., the assassination of John F. Kennedy). 24 Within autobiographical memory, there appears to be a gradient of strength, with certain remote memories being better preserved than more recent ones. Fromholt and Larsen (1992) had the same finding using a free recall task as well as finding a reminiscence peak associated with memories from 15 to 25 years of age for people over 40. A similar pattern was reported in the context of public memories (Kopelman, 1992; Fromholt & Larsen, 1992). Using lexical specificity as a measure, if it is true that an interaction exists between episodic memory and lexical representation, within a spreading activation framework, then memories that are more "strongly" represented in autobiographical memory should be associated with better lexical specificity. However, an alternative account is also possible. It may be that fewer cognitive resources are necessary for the recall of "stronger" memories. That is, if a memory is elaborately represented, then fewer resources would be necessary for the recall of that memory, making more resources available for other language production tasks such as lexical retrieval. This account can be explored by measuring lexical specificity across discourse types. Discourse structure is said to provide a cognitive framework for the organization of discourse tasks. This knowledge framework is used in tasks of both language comprehension and production, and is hypothesized to facilitate language production. As Palm and Purves (1996) asked in their study, if this knowledge is not impaired, does access to it facilitate discourse production by lessening organizational demands on an already compromised pool of cognitive resources? If the answer is yes, then we should see lexical specificity differentially represented in different discourse types. An additional measure is necessary to ensure that difference between discourse types are sufficiently robust to affect production. Informativeness is the amount of information conveyed relative to the overall amount of language produced (Ulatowska et al., 1988). The study by Nicholas et al. (1985) suggested that informativeness and lexical specificity, though correlated, measure different aspects of language. Hence, informativeness is the alternative 25 measure which will be used in evaluating the effects of varying different memory dimensions and of discourse types. Hypotheses If lexical specificity is closely tied to the predicted memory gradient in autobiographical memory, then we could postulate that there is a connection between autobiographical memory and lexical representation in semantic memory. Memory gradients suggest that personal memories are stronger than public memories, and that remote memories are stronger than recent ones. The following is hypothesized: 1 a. Reports of personal memories will be more lexically specific than reports of public memories. b. Reports of remote memories will be more lexically specific than reports of recent memories. 2 a . Reports of personal memories will be more informative than reports of public memories. 2 b . Reports of remote memories will be more informative than reports of recent memories. Alternatively, the "stronger" autobiographical memory representations may free cognitive resources for language processing and production tasks. If this is the case, then the use of discourse structure to help organize language production, which is postulated to make more cognitive resources available, should also result in better lexical specificity and informativeness. The following is, then, hypothesized: 3. More structured texts will be more lexically specific than less structured texts. 4. More structured texts will be more informative than less structured texts. 26 CHAPTER TWO METHOD The Subject K D is a 76 year old woman who was a professor of literature in eastern Canada. She moved back to Vancouver approximately 13 years ago. She is single and does not have any children. Her closest family live on Vancouver Island. After being independent for most of her life, K D now lives with friends (a late middle-aged couple). Throughout her life, K D has been socially active in her church community. Her social circle is not as extensive as before, but she does keep in touch with a few good friends and family on a regular basis. K D received diagnosis of probable Alzheimer's Disease1 approximately five years ago. She was very upset by this diagnosis at the time, but no longer seems to remember it. She refers to her difficulties in terms of word-finding and memory problems which interfere with her ability to communicate effectively. K D received therapy from a speech-language pathologist over a period of 18 months after her diagnosis. Therapy focused on "direct intervention to facilitate lexical retrieval and indirect intervention to maximize communicative effectiveness with caregivers and close friends...with a broader goal to reduce the impact of (KD's) disability on her independence and social involvement" (Palm & Purves, 1996, p.20). Her caregiver was involved with the therapy sessions to learn more about KD's past in order to successfully mediate when KD's memory further declines, and to learn which cues and prompts are most helpful to K D . There is no history of neurological impairment or other significant health issues. K D has a hearing loss in her left ear and was fitted for a hearing aid about 11 years ago. She was 1 Based on the criteria outlined by the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA). 27 never observed to use the hearing aid during any of the taping sessions. She does use reading glasses. K D continues to carry out most activities of daily living (e.g., dressing, self-care, ironing) independently. She reads the newspaper daily and keeps up with current affairs. Although she continues to initiate conversations, frequent breakdowns occur. K D is aware of these breakdowns and gets very frustrated at such times. Her communication partner needs to carry the burden of repairing communication breakdowns by asking clarifying questions and "filling in the blanks" whenever possible. At the time of the study, KD's abilities were generally consistent with moderate A D . Results from her most recent assessment are given in Appendix 1. In their interpretation of the assessment results for word-finding abilities, Palm and Purves (1996) concluded that K D was "unable to retrieve words for which she nevertheless had relatively intact conceptual and lexical-semantic representation" (p.162). They cite as evidence the following: (1) given a multiple choice version of the Boston Naming Test (BNT), K D was able to correctly identify an item she was initially unable to name from amongst a list of semantic distractors, (2) synonym and antonym multiple choice recognition tasks were performed with excellence, (3) access to a lexical item could be facilitated through phonemic cueing, (4) K D often gestured the functional use of an item during confrontation naming and definition tasks when unable to verbalize her thoughts, (5) K D provided relevant semantic details when unable to fully describe a lexical item, and (6) she had inconsistent errors for specific items on two separate presentations of the BNT (Palm & Purves, 1996). KD's language abilities were more compromised than her episodic memory abilities, a profile that contrasts with typical A D findings. Her "ability to refer to many events from her past" was taken as evidence that she had "relatively spared autobiographical memory" (p.162). For these reasons, K D was judged to be an appropriate subject for this study of memory and factors that play a role in lexical retrieval. 28 Procedures Selection of Conversational Topics The discourse topics were developed prior to data collection. They were chosen from a pool of topics K D frequently initiated in therapy sessions. They can be classified as representing either decontextualized knowledge or contextualized knowledge. Knowledge that is decontextualized, such as discourse structure knowledge, falls within the domain of semantic memory. The four discourse tasks related to this were telling a fairytale, explaining a procedural task, telling about an event, and giving a description of a person (see Table 2.1). Contextualized knowledge is held within episodic memory and includes both temporal and spatial aspects of an individual's personal experience. The contextual tasks differed according to two dimensions: personal versus public knowledge, and remote versus recent memories. The topics were characterized as personal-remote, personal-recent, public-remote, or public-recent (see Table 2.2). These topics belonged to one of two discourse types: events or descriptions, which differ in that the former have distinct temporal-sequential properties and a defined physical context of occurrence. The latter rely on the use of descriptors, without a well-defined schema guiding the manner in which the information is stored or retrieved. Table 2.1: Decontextulaized Topics Discourse Type Topic Procedure Narrative Event Description Going to a Restaurant Fairytale: Cinderella First Teaching Position Assassination of JFK Trip to Eastern Canada Quebec Referendum Father Winston Churchill Jim Jean Chretien 29 Table 2.2: Contextualized Topics Personal Memory Public Memory Remote Memory Events Description First Teaching Position Father Assassination of JFK Winston Churchill Recent Memory Events Description Trip to Eastern Canada Jim Quebec Referendum Jean Chretien Data Collection K D was informed that the purpose of the study was to find out which kinds of conversations would be easiest for her to engage in with regard to her word-finding and memory difficulties. It was decided that her primary caregiver not be present for the interviews because K D relied on her to supply words and frequently turned conversations over to her to finish. The corpus was gathered by audiotaping topic-directed conversations with K D . It was collected on three different Friday afternoons, within a six-week period, in the dining room of KD's home. K D provided tea and cookies. The conversational topics were led by the interviewer, who asked open-ended questions to initiate a particular topic. The direction of each topic was determined by K D , with redirection back to the topic as requested and/or required. The interviewer provided feedback through the use of conversational continuers (e.g., "mhm") and questions for clarification. When K D was having excessive trouble with a topic (i.e., being completely off-topic or having considerable word-finding difficulty), the interviewer introduced a topic change. A topic was closed when K D made a concluding statement or otherwise signaled an end. All topics except one (Quebec Referendum) were presented a second time on different days to ensure that KD's best performance was elicited. 30 Audiotape Transcription The tape recordings were made on a Marantz recorder with the subject wearing a cordless microphone on her lapel. There were approximately four-and-a-half hours of audiotaped interview. Each interview was orthographically transcribed by the interviewer. To ensure transcription reliability, the interviewer reviewed the audiotapes and transcripts two months later. Contextual notes such as another person entering the room, reference to the environment (e.g., "the sun's out"), and gestures and nonverbal behaviours made by K D were included. For completeness, unintelligible words or phrases were marked with a question mark and, where possible, transcribed phonetically. The main unit of transcription was the utterance. This was based on Shewan's (1988) definition of an utterance. It was defined as "a complete thought, usually expressed in a connected grouping of words, which is separated from other utterances on the basis of content, intonation contour, and/or pausing" (p. 124). The following criteria (from Shewan) were also taken into consideration: 1. A change in content signaled the end of an utterance. 2. Falling intonation contour signaled the end of an utterance. Rising intonation signaled the end of an utterance if it was a question. 3. Pauses longer than 2.0 seconds signaled the end of an utterance. 4. Tag questions or tag sentences were not segmented as separate utterances, e.g. "You'll do it, won't you?" (1 utterance) 5. Parenthetical remarks that were complete thoughts were segmented as separate utterances, e.g. "I guess that's what they were thinking." (1 utterance) 6. Sentence starters and enders (e.g. "okay," "and," "you know") were not segmented as separate utterances. 31 Analysis Transcript Analysis The transcripts were separated according to conversational topic. Then the Correct Information Unit (CIU) analysis, developed by Nicholas and Brookshire (1993), was used to evaluate the informativeness of the conversations. This is "a standardized rule-based system for quantifying the informativeness of connected speech" (p.339). It was designed "to evaluate the accuracy, relevance, and informativeness of the words produced by an individual" (p.344). Nicholas and Brookshire (1993) reported that the ratio of correct information units to words (percent CIUs) was found to be a more stable measure in repeated descriptions by aphasic and nonaphasic subjects than either number of CIUs or number of words. The CIU analysis was developed for use with adults with aphasia, but was found to be equally applicable for use with people with A D (Schmidt, 1996). The number of utterances, words, CIUs, and percent CIUs were calculated for each transcript (see Chapter 3, Table 3.1). Transcript analysis was validated by the interviewer's supervisor, who has considerable experience with the CIU method. Evaluation of Noun Specificity Concepts were described as having a hierarchical organization in semantic memory with superordinate or more general categories at the top and more specific subordinate exemplars further below. Macnamara (1984) supported the psychological reality of this portrayal of semantic memory. He reported that when mothers spoke to their children they automatically used more generic nouns (superordinate level) and when speaking to other adults they used more specific exemplars (e.g., given a picture of a daisy, they would call it a "flower" when talking to a child and "daisy" when talking with an adult). Thus, the mothers demonstrated an intuitive sense about the specificity of nouns. This observation supports Lyons's (1977) claim 32 that the specificity of nouns can be ranked in a scalar fashion from most to least specific in meaning. Noun Specificity Survey Based on the above observations, a survey was developed, in which adult, native-English speakers were asked to rate the specificity of meaning of given nouns. The nouns used in the survey were obtained from the transcript analysis described above. The survey required that each subject (a total of 15) judge the specificity of each noun on a scale of 1 to 5, with 1 being most general in meaning and 5 most specific. Some nouns occurred in more than one transcript, and some appeared in both a plural and a root word form. For the purpose of the survey only the root word was used and, to avoid redundancy, each noun appeared only once on the final survey. The resulting survey consisted of a total of 125 nouns from the 10 topics (19 transcripts in total). Simple written directions and examples of words which would be rated as either 1 or 5 were provided. No additional directions were given other than the written ones. Nouns, directions, and examples are given in Appendix 2. The results of all the surveys were compiled. From this a mean "specificity" rating and standard deviation were calculated for each noun (see Appendix 2). Nouns with a standard deviation of 1.0 or greater were not scored in the final analysis since this reflected a greater variability in their scoring than appeared for the majority of nouns scored. Hence, only those nouns with a standard deviation of less than 1.0 were considered to be reliably rated. All nouns, however, counted toward the total noun count for a particular transcript. The number of nouns with a standard deviation less than 1.0 was 91 (see Appendix 3). The results from the survey showed a definitive pattern of clustering for each noun. No bimodal distributions were present, demonstrating that people do have an intuitive sense about how specific a noun is in its meaning. Using a standard deviation cutoff of 1.0 also 33 helped to minimize the effects of words with more inconsistencies in their ratings. The noun survey was used in the following way: 1. Each noun with a standard deviation less than 1.0 was assigned the value of its mean "specificity" rating as obtained from the noun survey for each transcript. For each transcript, the percentage of nouns with a rating of 3.02 or more was determined. 2. As each topic had two transcripts (except for the Quebec Referendum), a mean percentage of nouns with a rating of 3.0 or more for each topic was determined. 3. The topics were then ranked from highest to lowest in specificity based on the mean percentage of nouns with a specificity rating of 3.0 or more. Classes of Nouns One of the primary drawbacks of the survey approach was that nouns were taken out of a conversational context and presented as single words. The raters did not have the benefit of context when making their specificity judgments. Because the nouns produced by K D were in a conversational context and as this was not addressed by the single word paradigm of the noun specificity survey, a second method of determining noun and topic specificity was employed. Nouns can be described by how specific, generic and definite they are (Huddleston, 1984). A specific noun or noun phrase has as its reference some particular object or person that can be easily identified by a communication partner. This has also been termed singular definite reference and it includes proper nouns, proper names and nouns which are limited by a prepositional phrase (Lyons, 1977). A generic noun has as its referent a particular class of objects. This includes category names such as "fruit." There are some generic words that can be described as both a subordinate noun and a superordinate noun. For example, "The leopard has a dark-spotted yellowish-fawn coat" can be a general statement about animals belonging to 2 Even though 3 can be considered a neutral rating, it was included since only one noun received a rating of 5.0. 34 the species leopard, but "The leopard was growling menacingly in its cage" can only be used to make a statement about some one individual leopard (Huddleston, 1984). Finally, defining information distinguishes the intended referent from all others. In this final category we will focus on indefinite pronouns (e.g., "everything," "somebody") which do not make a distinction in person, case or gender. The indefinite class was also used for all nouns that did not have a referent and did not make a general statement about a class or category. One important difference between this method and the single-word rating procedure described above is that "the interpretation [of the nouns] depends on both the form and the meaning of a sentence, not just the noun phrase itself (Huddleston, 1984, p. 255). The three noun classes were ordered from highest to lowest in specificity in the following way: specific, generic, and indefinite pronoun. The nouns in each transcript were classified into one of the three groups as appropriate. Once again the percentage of nouns represented in the highly specific list was calculated for each transcript and a mean score of the two transcripts for each topic (except the Quebec Referendum) was determined. The topics were then ranked from highest to lowest based on their mean score. Adjective Uniqueness Can other word categories show word retrieval effects as a function of the topic being discussed? Recall that the contextualized discourse tasks consisted of retelling of events and the description of people. The first is a task requiring the use of nouns to set the spatial framework and introduce common knowledge, such as the people involved and the setting, to the listener. The second task heavily relies on the use of adjectives to successfully meet the discourse expectations of the listener. Unlike nouns, adjectives are noncategorical in nature. They cannot be ranked in order of how specific they are in relation to each other. They can, however, be judged according to how unique they are to a particular topic. To judge the uniqueness of the adjective used for a particular topic the following steps were taken: 35 1. Adjectives used as quantifiers were not included in the analysis, since the main task of an adjective is as a descriptor. 2. For each topic with 2 transcripts, the descriptive adjectives were collapsed together. Repetitions of descriptive adjectives from the 2 transcripts were only listed once for each topic. 3. There was a total of 10 topics. Each adjective was given a value based on the number of topics that it was used for. For example, if an adjective was only used in one topic, then it got a rank of 1/10 which was highly specific or unique for that one topic. If it was used in 5 topics, then it received a rating of 5/10. There were a total of 82 descriptive adjectives (see Appendix 4) obtained from the transcripts. Of them, 47 received a value of 1/10. This accounted for 57.3% (47/82) of the adjectives produced. Since more than half of the adjectives were in this category, it was decided that the percentage of adjectives only appearing once for each topic would be calculated for the uniqueness analysis. The topics were then ranked from highest to lowest based on the percentage of adjectives receiving a rating of 1/10. Analyses The informativeness and lexical specificity analyses of each text were used to determine whether there differences among the texts according to memory gradients of recency-remoteness and personal-public. This will demonstrate whether there is an interaction between autobiographical memory and lexical representation or whether the effects of improved lexical specificity are a result of a reduction in cognitive processing loads. 36 CHAPTER THREE RESULTS Correct Information Unit Analysis The results from the Correct Information Unit (CIU) analysis are presented in Table 3.1. The purpose of the CIU analysis was to determine what proportion of KD's output was informative (i.e., accurate, intelligible, non-redundant and relevant relative to the total number of words produced for each topic). The measure used was percent CIUs, calculated by dividing the total number of CIUs by the total number of words for each transcript. In Table 3.2, the mean number of words and CIUs, and mean percent CIUs are presented by topic type. In terms of the effect of memory dimension on informativeness, recounting of personal memories was consistently more informative (across both events amd descriptions) than recounting of public memories. The remote-recent memory dimension was less consistent. For both personal and public event memories, recounting of recent memories was more informative than recounting of remote memories. Across descriptions of personal and public figures, however, the order of the remote-recent memories was not consistent. In other words, for three out of four possible comparisons, recent memories were more informatively recounted than remote memories. In order to determine the effect of discourse type on informativeness, mean percent CIUs for the most informative recounting of event and description topics (personal-recent in both cases) were compared with results for the narrative and procedural tasks. From most to least informative, these discourse types ranked as follows: 1. Procedure 2. Event 3. Narrative 4. Description 37 Table 3.1: Total Number of Utterances, Words, CIUs and Percent of CIUs in Transcripts T r a n s c r i p t s T o t a l N u m b e r o f U t t e r a n c e s T o t a l N u m b e r o f W o r d s T o t a l N u m b e r o f C I U s P e r c e n t C I U s ( t o t a l C l U s / t o t a l ^ j w q r d s ) ^ C o n t e x t u a l i z e d T o p i c s : Events: #1 First Teaching 69 434 145 33.4% Position #2 First Teaching 65 399 215 53.9% Job Position #1 Assassination 69 474 105 22.2% of JFK #2 Assassination 105 600 166 27.7% of JFK #1 Trip to Eastern 199 1270 631 49.7% Canada #2 Trip to Eastern 76 664 279 42.0% Canada #1 Quebec 76 486 207 42.6% Referendum Descriptions: #1 Father 32 476 143 30.0% #2 Father 85 539 179 33.2% #1 Winston 68 426 67 15.7% Churchill #2 Winston 21 147 56 38.1% Churchill #1 Jim 65 436 128 29.4% #2 Jim 56 362 147 40.6% #1 Jean Chretien 48 374 66 17.6% #2 Jean Chretien 42 371 107 28.8% D e c o n t e x t u a -l i z e d T o p i c s : #1 Going to a 44 332 120 37.3% Restaurant #2 Going to a 21 213 142 66.7% Restaurant #1 Telling a 73 413 179 43.3% Fairytale #2 Telling a 76 632 251 39.7% 38 Table 3.2^ Measure of Informativeness by Topic T o p i c C o n t e x t u a -l i z e d E v e n t s : M e a n N u m b e r o f J V o r d s m Personal-Recent (967) Public-Remote (537) Public-Recent (486) Personal-Remote (416.5) M e a n N u m b e r o f C I U s Personal-Recent (453.5) Public-Recent (207) Personal-Remote (180) Public-Remote (135.5) M e a n P e r c e n t C I U s Personal-Recent (45.9%) Personal-Remote (43.7%) Public-Recent (42.6%) Public-Remote (25.0%) C o n t e x t u a -l i z e d D e s c r i p t i o n s : Personal-Remote (507.5) Personal-Remote (161) Personal-Recent (399) Public-Recent (372.5) Public-Remote (286.5) Personal-Recent (137.5) Public-Recent (86.5) Public-Remote (61.5) Personal-Recent (35.0%) Personal-Remote (31.6%) Public-Remote (26.9%) Public-Recent (23.2%) D e c o n t e x t u a -l i z e d T o p i c s : Narrative (522.5) Narrative (215) * Please note: Bracketed numbers indicate the mean for each topic. Procedure (52%) Lexical Specificity Across Memory Dimensions In this section, results relating to the memory dimensions (personal-public, remote-recent for events and descriptions) will be discussed in relation to the different measures of lexical specificity. Noun Specificity Analysis The results of the noun specificity analysis are summarized in Table 3.3. The percentage of contextualized nouns with a mean score of 3.0 or higher is given for each topic. Both the event and description topics have the same ordering of memory dimensions. From highest to lowest in lexical specificity the order obtained was: 1. Personal-Remote 2. Personal-Recent 3. Public-Remote 39 4. Public-Recent As predicted, recounting of memories associated with personally relevant events and people yielded a higher proportion of specific nouns than did recounting of public memories. As predicted, remote memories resulted in greater production of specific nouns than their recent counterparts. This finding was consistent across events and descriptions for both personal and public memories. Table 3.3: Noun Specificity Percentages For the Memory Dimensions Topics Tjpc Vc of Specific Nouns Contextuaiized Events First Teaching Position Personal-Remote 55.5% Trip to Eastern Canada Personal-Recent 48.4% Assassination of JFK Public-Remote 38.7% Quebec Referendum Public-Recent 17.6% Contextuaiized Descriptions Father Jim Winston Churchill Jean Chretien Personal-Remote 48.5% Personal-Recent 44.1 % Public-Remote 12.5% Public-Recent 12.5% Table 3.4: Percentage of Specific Nouns From Noun Classification Analysis Topics Types % of Specific Nouns Contextuaiized Events First Teaching Position Personal-Remote 45.3% Trip to Eastern Canada Personal-Recent 30.5% Assassination of JFK Public-Remote 11.9% Quebec Referendum Public-Recent 00.0% Contextuaiized Descriptions Father Jim Winston Churchill Jean Chretien Personal-Remote 37.6% Personal-Recent 41.6% Public-Remote 28.0% Public-Recent 13.9% 40 Noiin Classification Analysis The results of the noun classification analysis are summarized in Table 3.4 with the percentage of specific nouns for each topic given. For the event memories, the order of the topics from highest to lowest proportion of specific nouns was as follows: 1. Personal-Remote 2. Personal-Recent 3. Public-Remote 4. Public-Recent For the description memories, a different order was obtained: 1. Personal-Remote 2. Personal-Recent 3. Public-Recent 4. Public-Remote As predicted, contextualized memories associated with personally relevant events and people had a higher proportion of specific nouns than did public memories. The results of the remote memories compared to the recent memories was not as clear as it was in the noun specificity analysis. The remote over recent prediction was upheld for three of the four topics. Comparison of the Noun Specificity and Noun Classification Analyses In comparing the two methods of analyzing noun specificity, the percentages themselves were not compared because the criteria producing them were different. Rather, the relative order of the topics was of importance. A similar ordering of results was produced with both methods. The personal memories allowed for consistently better lexical retrieval of specific nouns than the public memories. The remote-recent memory ordering was consistent in all but one case. 41 Descriptive Adjectives The results from the adjective analysis are summarized in Table 3.5. It shows the percentage of unique adjectives (i.e., those that received a score of 1/10) for each topic. For the event memories the order of the topics from highest to lowest for unique adjectives was as follows: 1. Personal-Remote 2. Personal-Recent 3. Public-Recent 4. Public-Remote The personal event memories had better production of unique adjectives than the public memories. As predicted, the personal-remote memory had the greatest proportion of unique adjectives for the event topics. The remote-recent distinction was reversed for the public memories. The public-recent memory had a greater proportion of unique adjectives than the public-remote memory. The ordering of the adjectives for the description topics was slightly different from that obtained for the event topics. The order was as follows: 1. Personal-Remote 2. Personal-Recent 3. Public-Remote 4. Public-Recent As predicted, K D produced a greater percentage of unique adjectives for the personal description memories than she did for the public ones. Consistent with the predictions, the remote memories elicited more unique adjectives than their recent counterparts, and the personal-remote memories had the greatest production of unique adjectives. 42 Table 3 . 5 : Percentage of Topics Contextualized Events First Teaching Position Trip to Eastern Canada Assassination of J F K Quebec Referendum Contextualized Descriptions Father Jim Winston Churchill Jean Chretien ue Adjectives Types'"" Personal-Remote Personal-Recent Public-Remote Public-Recent Personal-Remote Personal-Recent Public-Remote Public-Recent Percentage of Unique Adjectives (score 1/10) 28.4% 15.0% 6.30% 14.3% 43.1% 25.4% 22.5% 16.7% Summary of Noun and Adjective Findings in Terms of Memory Dimensions The ordering of the contextualized topics according to their memory dimensions and type of analysis performed is summarized in Table 3 . 6 . The personal-public dimension was stable in ail cases (i.e., the personally relevant topics resulted in better retrieval and production of specific nouns and unique adjectives). The remote-recent dimension, however, wavered slightly. The overwhelming conclusion was that even when different parts of speech were analyzed and different analysis methods were used, essentially the same results are obtained. Table 3 . 6 : Consistency of Ordering of Memory Dimensions with Predicted Gradient Across Topics Memory Dimension Analysis Type Noun Specificity Noun Classification Unique Adjective P e r s b h a l / P u b l i c 4 / 4 4 / 4 4 / 4 Remote/Recent 4/4 3/4 3/4 43 Lexical Specificity Across Discourse Types In the following section, results relating to the different discourse types (event, description, narrative and procedure) will be presented. To analyze the data for lexical specificity the highest scoring topic for each discourse type was used. The order obtained is shown in Table 3.7. The scores for the topics using the noun analyses are all within 10% of each other, whereas across memory dimensions the range of scores was about 48%. The adjective scores are within 16% of each other, whereas across the memory dimensions the range of scores was 37%. The scores from the noun specificity survey and the noun classification analysis are all relatively close to one another. For the noun specificity survey, the spread in scores shows the procedural task with the highest percentage of specificity and the descriptions with the lowest. For the noun classification analysis, events have the highest percentage of specific lexemes and the procedural task has the least. From the adjective analysis, we see a greater spread between the scores. Interestingly, but not surprisingly, the description task has the greatest percentage of unique adjectives. The narrative has the fewest unique adjectives. Based on the results obtained, it is not possible to make a conclusive statement regarding the role of superstructure or schematic organization in lexical retrieval for the tasks presented. Hence, it does not appear that discourse type is as relevant a factor as memory type for lexical retrieval. Table 3.7: Ordering Discourse Types by Measure of Lexical Specificity Noun Specificity Survey Noun Classification Analysis Unique Adjectives Procedure ( ^ 9 ^ E v e n t s (45.3%) Descriptiohs (43.1%) Events (55.5%) Narrative (42.1%) Procedure (35.0%) Narrative (55.3%) Descriptions (41.6%) Events (28.4%) Descriptions (48.5%) Procedure (35.4%) Narrative (27.2%) 44 A Comparison of Effects of Memory Dimensions and Discourse Types on Informativeness and Lexical Specificity Up to this point we have discussed informativeness and lexical specificity separately. Comparing the two measures may reveal whether there is an interaction between autobiographical memory and lexical representation or whether the freeing up of cognitive resources is the explanation for enhanced lexical retrieval in different discourse tasks. Memory Dimensions Mean informativeness scores for memory dimensions were derived by averaging the percent CIUs obtained for each transcript with the same memory dimension (e.g., all the personal-remote memories, all the public-recent memories, etc.). They are shown in Table 3.8 along with the mean lexical specificity scores. The mean lexical specificity scores were determined by averaging the results of the specificity scores, for each memory dimension, obtained for each of the three analysis methods employed. These averaging procedures were justified on the grounds that there was sufficient consistency of results across analyses to suggest that these measures were different instances of the same phenomenon. The results show that the personal memories are more informative and lexically specific than the public ones. For the remote-recent dimension there is an interesting dissociation for the two measures of informativeness and lexical specificity. The recent topics are more informative, yet less specific than the remote ones. Possible reasons for this will be presented in the discussion. Table 3.8: Informativeness and Lexical Specificity Means Related to Memory Dimensions Memory Dimensions Mean Informativeness Mean Lexical Specificity PersohaURemote '37.6% "4371% " Personal-Recent 40.4% 34.2% Public-Remote 25.9% 20.0% Public-Recent 29.7% .12.5% 45 Discourse Types To study the effects of discourse type, mean informativeness scores for the event and description topics were obtained by averaging the transcripts with the highest percent CIUs for each topic. The average of the mean percent CIUs for the two narrative and procedural transcripts was calculated to obtain the mean informativeness results. Mean lexical specificity scores were derived by using the topic with the highest proportion of specific words for each discourse type. As can be seen, mean informativeness measures show some differences for discourse type. The procedural task is the most informative and the description task is the least informative, as would be expected based on the differences in their internal structures. When mean lexical specificity is calculated, the differences among the discourse types is no longer observed. That is, no one discourse type demonstrates an advantage for lexical specificity. Table 3.9: Informativeness and Lexical Specificity Means For Discourse Types Discourse Type Mean Informati\cnvss Mean Lexical Specificity Procedure 52.0% 43.3% Event 43.5% 43.1% Narrative 41.5% 41.5% Description 35.2% _ 44.4% 46 CHAPTER FOUR DISCUSSION The purpose of this study was to investigate the nature of the interaction between autobiographical memory and lexical representation in semantic memory. Within autobiographical memory a gradient of strength of memories has been described. According to the gradient, personal memories are said to be better represented than public ones and remote memories are better represented than recent ones. If there is an interaction between autobiographical memory and lexical representation, then this gradient should be reflected in measures of lexical specificity. A secondary purpose of the study was to evaluate whether differences in lexical specificity result from decreases in cognitive load. Possible caveats need to be considered when interpreting the results from this study. First, the personal-recent event memory (Trip to Eastern Canada) triggered past memories of an earlier holiday to eastern Canada which contributed to the CIU count, and calculation of informativeness, noun specificity and adjective uniqueness. Thus, there is a confound in the interpretation of results regarding the remote-recent memory distinction. Second, there is a confound concerning public memories. As mentioned earlier, a few researchers (Brown, 1990; Fromholt & Larsen, 1992; van der Linden et al., 1992) have found that public memories are connected to personal interpretations and experiences (e.g., peoples' opinions regarding political issues). It may be that facts concerning a particular public event or personality are also strongly represented in autobiographical memory. It is difficult, then, to isolate how much of a public recall task involves information strictly from semantic memory and not that from autobiographical memory. Third, compared to Nicholas and Brookshire's (1993) subjects, KD's informativeness scores (i.e., percent CIUs) varied considerably. The mean difference in the percent CIUs for Nicholas and Brookshire's aphasic subjects between sessions (averaged across elicitation stimuli) was 3%, with scores ranging from 0% to 6%. Because Nicholas and Brookshire (1993) measured between-session scores across all stimuli it is not possible to 47 compare the variability between stimuli in their data with KD's results. However, in comparing topic by topic across two sessions, it appears that there is greater variability in KD's scores than for Nicholas and Brookshire's subjects. The overall informativeness of each of KD's transcripts ranged between 14% and 67%. The differences between sessions in percent CIUs for her topics ranged from 3.1 % to 29%, with a mean difference of 11.6%. The ranges of differences for K D are appreciably greater than those for Nicholas and Brookshire's subjects. This greater range may reflect KD's fluctuating access to specific memories, differences in attention, or other factors which could influence overall informativeness. Thus, these results need to be interpreted cautiously. The two measures used in this study were informativeness and lexical specificity. Informativeness refers to the amount of information conveyed relative to the overall amount of language produced. Factors implicated in reducing informativeness include repetitions, revisions, use of fillers, indefinite terms and reference, perseveration, and attentional deficits. To measure KD's informativeness in the conversational tasks presented, the CIU analysis was used. Using the noun specificity, noun classification, and the unique adjective analyses, the proportion of specific lexical items produced by K D was determined. Specificity refers to how closely a referent is named so as to leave no doubt to its identification; for example, "African violet" is more specific than "flower" because it refers to a particular type of flowering plant. The data in this project addressed the effects of three different dimensions on informativeness and lexical specificity: personal versus public memories, remote versus recent memories, and discourse type. I will discuss mean informativeness and mean lexical specificity results in terms of the hypotheses (repeated here from Chapter One): l a . Reports of personal memories will be more lexically specific than reports of public memories. lb. Reports of remote memories will be more lexically specific than reports of recent memories. 48 2a. Reports of personal memories will be more informative than reports of public memories. 2b. Reports of remote memories will be more informative than reports of recent memories. 3. More structured texts will be more lexically specific than less structured texts. 4. More structured texts will be more informative than less structured texts. First, let us begin with the personal-public memory distinction. The results from the mean informativeness calculation revealed that the personal memories were more informative than the public ones. All the lexical specificity measures also demonstrated that the personal memories consistently had the highest proportion of specific lexemes produced. This finding supports hypothesis la and 2a, and is compatible with the proposal that, according to memory gradients, personal memories are more strongly represented than public ones in autobiographical memory. The question that needs to be asked from the above finding is: How is it that memories which are stored in the autobiographical memory system result in better access to specific lexical items in semantic memory, while those which presumably have a stronger representation in semantic memory (i.e., public memories) did not show this effect? Perhaps the answer lies in the logogen model. Recall that for the logogen model the number of features or specific information feeding into a logogen must exceed some threshold value before lexical selection can occur. I am suggesting that one possible feature available to a word's logogen is information from autobiographical memory. Hence, when a logogen for an item in subordinate levels of semantic memory is associated with an autobiographical representation, access to these levels is possible because the additional information from autobiographical memory exceeds the threshold allowing lexical selection to occur. This process may explain the results of the overall strength of personal memories in the retrieval of both informative and specific lexemes from lexical representation. This suggests, then, that there are connections between autobiographical memory and lexical representation. In fact, Hyltenstam and Stroud (1989), 49 on the basis of a single bilingual subject, and Dronkers, Koss, Friedland and Wertz (1986) have proposed that personal experiences are encoded lexically. Unlike the personal-public memory distinction, the remote-recent memory distinction differed for measures of informativeness and lexical specificity. Whereas the mean informativeness showed that the recent topics were more informative than the remote ones (a finding that does not support hypothesis 2b), the mean lexical specificity results demonstrated that, in most instances, the remote topics had the most specific lexemes, in accordance with hypothesis 2a. Similarly, the relative strength of remote versus recent memory for normal and A D individuals is not reported consistently in the literature. Sagar et al. (1991) found that given noun cue words, their normal and A D subjects retrieved or recalled memories from the most recent 20 years and had a smaller increase for remote memories. The lexical specificity results correspond more closely to results from Fromholt and Larsen (1992). Using a free narrative recall method, they found that remote memories from a person's early adulthood were recalled better than more recent ones by people over the age of forty. Conversely, they found a smaller recency increase. In this thesis, the methodology used could be seen to be a mixture of those used by the above researchers. The topics chosen for this study were those that K D had often initiated in therapy, prior to the study. For the study, however, she was asked to discuss these topics, cued by an open-ended question related to a particular topic. If we assume that the remote-recent memory effect is reliable for our two measures, how might we account for this dissociation between informativeness and specificity? It may be that remote memories are rehearsed verbally many times leading to them being lexically encoded with specific lexemes activated by co-occurring lexemes. During retrieval processes, then, these specific items are easily, almost automatically, recovered for use by an individual. Recent memories, on the other hand, have not had the benefit of frequent repetition or the ensuing associations between lexical representation; they may not be fully lexically encoded. Their higher informativeness results may be reflecting a greater number of attempts to describe the memories without the benefit of producing specific lexemes. An example that illustrates 50 this is the use of cicumlocutionary speech; it is possible to be informative when using circumlocutions, but not specific at the same time. There is the possibility that the lexical specificity advantage seen for these "stronger" memories could be the result of freeing up of cognitive resources for language production. To evaluate this effect, lexical specificity and informativeness across discourse types were examined. Discourse types consist of knowledge structures that are postulated to (differentially) facilitate language processing by reducing organizational demands placed on cognitive processes. More resources can then be diverted to other language processing tasks, such as lexical selection and retrieval. All discourse types have an inherent structure. However, they do not all have the same composition. Event topics are temporally organized, description topics are organized according to trait and stereotype information, procedures have a scriptal arrangement, and narratives have specific story components according to which they are planned. The results from the mean informativeness analysis revealed that the procedural task was the most informative discourse type. It was followed by event and narrative discourse. Description discourse was the least informative discourse type. Procedural discourse was the most constrained of the discourse types presented to K D . It has clearly outlined temporal and sequential guidelines. Descriptions, however, are the least structured of the discourse types. As other researchers (Wirth, 1991) have already shown, performance on different discourse tasks varies according to the discourse type. The fact that the descriptions had the lowest informativeness scores can be explained, then, by their lack of available structure to help K D organize her discourse. That these informativeness results were demonstrated by K D is consistent with hypothesis 4 and the conclusion that at least some aspect of her superstructural or schematic knowledge was intact, although this was never formally assessed. Measures of mean lexical specificity, however, revealed that all four discourse types (events, descriptions, procedures, and narratives) were comparable in the proportion of specific lexemes produced. That is, no one discourse type demonstrated a lexical specificity 51 advantage as was predicted by hypothesis 3. This very strong finding of a dissociation between informativeness and lexical specificity scores leads to the conclusion that there is a connection between autobiographical memory and lexical representation. Recall the proposal that lexical specificity could reflect a reduction in cognitive load due to certain autobiographical memories being more strongly represented, and thus are more easily retrieved from memory. However, the mean lexical specificity results are less consistent with this conclusion. Hence, it is less likely that more cognitive resources are available for selection of specific lexemes. Given the limitations of the data from this study other possible reasons for the differences in mean informativeness and mean lexical specificity scores must be considered. They include: (1) overall informativeness scores for the same topic were highly variable; (2) the topics chosen may not have been the most effective for looking at lexical specificity; (3) more data may be necessary to ensure reliability of the measures and the results obtained (for example, only one topic each was used for both the decontextualized tasks (procedure and narrative) whereas the contextuaiized tasks had four topics each). Overall, the results of this study strongly demonstrate that personal significance was a major contributor to improved lexical access and retrieval for KD. Regardless of the analysis method used, the personal topics always had the greatest proportion of specific lexical items. Based on these findings, the conclusion can be made that there does exist an interaction between autobiographical memory and lexical representation. This corresponds with views by Keenan and Baillet (1980) and Conway (1990) that memories encoded with autobiographical content are represented in a more elaborate and meaningful way that allows for more successful retrieval. It is also congruent with conclusions by Snowden et al. (1995) that there are associations between different memory systems, in particular autobiographical memory and semantic memory. It fits with the notion held by semantic network theorists that the stronger connections are between concepts and lexical items in the network, the easier it becomes for them to be activated and eventually retrieved. Along the same lines, when associations are formed between semantic memory and autobiographical memory they have a strong 52 representation due to an elaborate encoding process that occurs in autobiographical memory. Finally, as shown in the bilingual data (Hyltenstam & Stroud, 1989), some aspect of autobiographical memory storage requires lexical encoding to take place for all of the above to be present. Based on the results of the present study, I will make the claim that one important factor contributing to the lexical processes of access and retrieval is the semantic memory lexical representation connections formed with concepts in autobiographical memory. Also, autobiographical memory representations may facilitate improved lexical access from semantic memory. Although this study was limited to the retrieval of nouns and adjectives, it is predicted that verbs and adverbs would show a similar pattern of improved access if they were originally encoded as a part of an autobiographical memory. 53 CONCLUSION In conclusion, the following results were obtained for the hypotheses. Hypotheis la predicted that personal memories would result in better lexical specificity than public memories. This was upheld in the results of this thesis. Hypothesis lb predicted that personal memories would result in better informativeness than public memories. This was also supported by the results of this study. Hypothesis 2a predicted that remote memories would result in better lexical specificity than the recent memories. The results of the analyses revealed that remote memories were overall more lexically specific, but the results were inconsistent across analyses. Hypothesis 2b predicted that remote memories would result in better informativeness than the recent memories. This hypothesis was not supported by the results. Hypothesis 3 predicted that more structured texts would be more lexically specific than less structured texts. This hypothesis was not supported by the results of the study. Finally, hypothesis 4 predicted that more structured texts would be more informative than less structured tasks. This was upheld in the results of the thesis. Results were interpreted as support for the notion that autobiographical memory contributes to lexical representation. Clinically, suggestions have been made to family members of individuals with A D to discuss topics that are personally relevant to the affected family member. Reasons have included observations that the person with A D has more success when engaged in these topics, feelings of self-worth are increased, and there is a reduction in emotional stress (Palm & Purves, 1996). The results from this study show that an additional reason for making this suggestion is that there exist strong connections between autobiographical memory and lexical representation which facilitate lexical retrieval of substantive words and improve overall discourse and conversation. The clinical implication is to continue encouraging significant others to engage the person with A D in conversations that are personally and emotionally significant. 54 The above implication, however, would have to reconcile itself with the claim that within the heterogeneous group of people with A D , there exist more homogenous subgroups. Within each subgroup the course of the disease is similar. For example, some individuals show more profound memory losses compared to their language abilities, and others have language skills that are more impaired than their memory. The course of each person's illness, then, would determine the effectiveness of the above therapy suggestions. Further research questions are suggested by the need to know more about the memory gradient that exists for remote and recent memories. These findings were not consistent in this study or in other studies that have been done. It could be that using different methods of retrieving memories has a differential effect on which memories are retrieved. Also, learning more about the various subgroups of A D patients would be beneficial for clinical interventions and strategies provided to families and caregivers, as well as providing more clues into the working of memory systems. 55 BIBLIOGRAPHY AND REFERENCES CITED Alba, J. & Hasher, L. 1983. Is memory schematic? Psychological Bulletin. 93 (2), 203-231. Appell, J., Kertesz, A. , & Fisman, M . 1982. A study of language functioning in Alzheimer patients. Brain and Language. 17. 73-91. Baddeley, A. 1984. Memory theory and memory therapy. In B. Wilson & N. Moffat (Eds.), Clinical Management of Memory Problems. San Diego, CA: Singular Publishing Group. 1-31. Bakerink, R.A. 1988. Memory in Alzheimer's Disease. Unpublished Master of Thesis. University of British Columbia, Vancouver, Canada. Bayles, K. 1994. Management of neurogenic communication disorders associated with dementia. In R. Chapey (Ed.), Language Intervention Strategies in Adult Aphasia: 3rd Edition. Baltimore, MD: Williams & Wilkins. 535-545. Bayles, K.A. 1985. Communication in dementia. In H. Ulatowska (Ed.), The Aging Brain: Communication in the Elderly. San Diego, CA: College-Hill Press. 157-173. Bayles, K .A. , Boone, D.R., Tomoeda, C.K., Slauson, T.J., & Kaszniak, A.W. 1989. Differentiating Alzheimer's patients from the normal elderly and stroke patients with aphasia. Journal of Speech and Hearing Disorders. 54, 74-87. Bayles, K. and Kaszniak, A. 1987. Communication and Cognition in Normal Aging and Dementia. Boston: College-Hill Press. Blennow, K., Wallin, A. , & Gottfries, C. 1994. Clinical subgroups of Alzheimer disease. In V.O.B. Emery & T.E. Oxman (Eds.), Dementia: Presentations. Differential Diagnosis, and Nosology. Baltimore, MD: The Johns Hopkins University Press. 95-107. Brown, G. and Yule, G. 1983. Discourse Analysis. Cambridge: Cambridge University Press. Brown, N.R. 1990. Organization of public events in long-term memory. Journal of Experimental Psychology: General. 119 (3), 297-314. Brown, R. 1958. Words and Things: An Introduction to Language. New York: The Free Press. Caplan, D. 1992. Language: Structure. Processing, and Disorders. Cambridge, M A : The MIT Press. Cardebat, D., Demonet, J-F., & Doyon, B. 1993. Narrative discourse in dementia. In H.H. Brownell & Y. Joanette (Eds.), Narrative Discourse in Neurologically Impaired and Normal Aging Adults. San Diego, CA: Singular. 317-332. Chang, T . M . 1986. Semantic memory: Facts and models. Psychological Bulletin. 99 (2), 199-220. Cherfkow, H. , & Bub, D. 1990. Semantic memory loss in dementia of Alzheimer's type: What do various measures measure? Brain. 113. 397-417. 56 Chertkow, H. , & Bub, D. 1990. Semantic memory loss in Alzheimer-Type Dementia. In M.F. Schwartz (Ed.), Modular Deficits in Alzheimer-Type Dementia. Cambridge, M A : The MIT Press. 207-244. Collins, A. , & Loftus, E. 1975. A spreading-activation theory of semantic processing. Psychological Review. 82 (6), 407-428. Conway, M.A. 1990. Autobiographical Memory: An Introduction. Philadelphia: Open University Press. Conway, M.A. , & Rubin, D.C. 1994. The structure of autobiographical memory. In A.F. Collins, S E . Gathercole, M.A. Conway, & P.E. Morris (Eds.), Theories of Memory. Hillsdale, NJ: Lawrence Erlbaum Associates. 103-137. Craik, F.I.M., & Byrd, M . 1982. Aging and cognitive deficits: The role of attentional resources. In F.I.M. Craik & S. Trehub (Eds.), Advances in the Study of Communication and Affect (Volume 8): Aging and Cognitive Processes. New York: Plenum Press. 191-208. Cummings, J.L. 1985. Dementia: Neuropathological correlates of intellectual deterioration in the elderly. In H. Ulatowska (Ed.), The Aging Brain: Communication in the Elderly. San Diego, CA: College-Hill Press. 53-67. Cummings, J.L., & Benson, D.F. 1983. Dementia: A Clinical Approach. Boston, M A : Butterworth. Dronkers, N.F., Koss, E., Friedland, R.P., & Wertz, R.T. 1986. "Differential" language impairment and language mixing in a polyglot with probable Alzheimer's disease. . Paper presented at The International Neuropsychological Society meeting. Ehrlich, J.S. 1994. Studies of discourse production in adults with Alzheimer's disease. In R.L Bloom, L.K. Obler, S. DeSanti & J.S. Ehrlich (Eds), Discourse Analysis and Applications: Studies in Adult Clinical Populations. Hillsdale, NJ: Lawerence Erlbaum Associates. Publishers, 149-160. Flicker, C , Ferris, S., Crook, T., & Bartus, R.T. 1987. Implications of memory and language dysfunction in the naming deficit of senile dementia. Brain and Language. 31, 187-200. Fromholt, P., & Larsen, S.F. 1992. Autobiographical memory and life-history narratives in aging and dementia (Alzheimer's type). In M.A. Conway, D.C. Rubin, H . Spinnler, & W.A. Wagenaar (Eds.), Theoretical Perspectives on Autobiographical Memory. Dordrecht: Kluwer Academic Publishers. 413-426. Garrett, M.F. 1980. Levels of processing in sentence production. In B. Butterworth (Ed.), Language Production: Speech and Talk (Volume 1). London: Academic Press Inc. Ltd. 177-220. Garrett, M.F. 1982. Production of speech: Observations from normal and pathological language use. In A.W. Ellis (Ed.), Normality and Pathology in Cognitive Functions. London: Academic Press Inc. Ltd., 19-76. 57 Garrett, M.F. 1984. The organization of processing structure for language production: Applications to aphasic speech. In D. Caplan, A.R. Lecours, & A. Smith (Eds.), Biological Perspectives on Language. MIT Press. 172-193. Gordon, B. 1983. Lexical access and lexical decision: Mechanisms of frequency sensitivity. Journal of Verbal Learning and Verbal Behaviour. 22, 24-44. Graf, P. & Schacter, D.L. 1985. Implicit and explicit memory for new associations in normal and amnesic subjects. Journal of Experimental Psychology: Learning. Memory, and Cognition. 11 (3), 501-518. Heindel, W.C., Salmon, D.P., & Butters, N. 1993. Cognitve approaches to the memory disorders of demented patients. In P.B. Sutker, & H.E. Adams (Eds.), Comprehensive Handbook of Psychopathology: Second Edition. New York: Plenum Press. Hess, T . M . 1990. Aging and schematic influences on memory. In T . M . Hess (Ed), Aging and Cognition: Knowledge Organization and Utilization. New York: Elsevier Science Publishers B.V. 93-160. Huddleston, R. 1984. Introduction to the Grammar of English. Cambridge: Cambridge University Press. Huff, F.J. 1990. The disorder of naming in Alzheimer's disease. In L.L. Light & D . M . Burke (Eds.), Language. Memory, and Aging. Cambridge: University Press, 209-220. Huff, F.J., Corkin, S., & Growdon, J.H. 1986. Semantic impairment and anomia in Alzheimer's disease. Brain and Language. 28, 235-249. Huppert, F.A. 1994. Memory function in dementia and normal aging-dimension or dichotomy? In F.A. Huppert, C. Brayne & D.W. O'Connor (Eds.), Dementia and Normal Aging. Cambridge: Cambridge University Press. 291-330. Hyltenstam, K. & Stroud, C. 1989. Bilingualism in Alzheimer's dementia: Two case studies. In K. Hylenstam & L. Obler (Eds.). Bilingualism Across the Lifespan: Aspects of Acquisition. Maturity, and Loss. Cambridge: Cambridge University Press. 202-226. Keenan, J.M., & Baillet, S.D. 1980. Memory for personally and socially significant events. In R.S. Nickerson (Ed.), Attention and Performance Vm. Hillsdale, NJ: Lawrence Erlbaum Associates. 651-669. Kempler, D. 1995. Language changes in dementia of the Alzheimer's type. In R. Lubinski (Ed.), Dementia and Communication. San Diego, CA: Singular Publishing Group. 98-114. Kempler, D. & Zelinski, E . M . 1994. Language in dementia and normal aging. In F.A. Huppert, C. Brayne & D.W. O'Connor (Eds.), Dementia and Normal Aging. Cambridge: Cambridge University Press. 331-365. Kertesz, A. 1994. Language deterioration in dementia. In V.O.B. Emery & T.E. Oxman (Eds.), Dementia: Presentations. Differential Diagnosis, and Nosology. Baltimore, MD: The Johns Hopkins University Press. 123-138. 58 Kopelman, M . 1992. Autobiographical memory in clinical research and practice. In M.A. Conway, D.C. Rubin, H . Spinnler & W.A. Wagenaar (Eds.), Theoretical Perspectives on Autobiographical Memory. Dordrecht: Kluwer. 427-450. Kopelman, M . 1989. Remote and autobiographical memory, temporal context memory and frontal atrophy in Korsakoff and Alzheimer patients. Neuropsychologia. 27 (4), 437-460. Kopelman, M . 1984. The psychopharmacology of human memory disorders. In B. Wilson & N. Moffat (Eds.), Clinical Management of Memory Problems. California: Singular. 189-215. Levelt, W.J.M. 1992. Accessing words in speech production: Stages, processes and representations. Cognition. 42, 1-22. Levelt, W.J.M. 1989. Speaking: From Intention to Articulation. Cambridge, M A : The MIT Press. Lyons, J. 1977. Semantics 1. Cambridge: Cambridge University Press. Macnamara, J. 1984. Names For Things: A Study of Human Learning. Cambridge, M A : MIT Press. Mandler, J. 1984. Stories. Scripts, and Scenes: Aspects of Schema Theory. Hillsdale, NJ: Lawerence Erlbaum Associates. Martin, A. 1990. Neuropsychology of Alzheimer's disease: The case for subgroups. In M.F. Schwartz (Ed.), Modular Deficits in Alzheimer-Type Dementia. Cambridge, MA: The MIT Press. 143-175. Martin, A. , & Fedio, P. 1983. Word production and comprehension in Alzheimer's disease: The breakdown of semantic knowledge. Brain and Language. 19. 124-141. McArthur, T., & McArthur, F. (Eds.), 1992. The Oxford Companion to the English Language. Oxford: Oxford University Press. McKhann, G., Drachman, D., Folstein, M . , Katzman, R., Price, D., & Stadlan, E . M . 1984. Clinical diagnosis of Alzheimer's disease: Report of the NINCDS-ADRDA work group under the auspices of Department of Health and Human Sources Task Force on Alzheimer's disease. Neurology. 34 (7), 939-944. Molloy, D.W., & Lubinski, R. 1995. Dementia: Impact and clinical perspectives. In R. Lubinski (Ed.), Dementia and Communication. San Diego, CA: Singular. 2-21. Morris, P.E., & Gruneberg, M . M . 1994. The major aspects of memory. In P.E. Morris & M . Gruneberg (Eds.), Theoretical Aspects of Memory: Second Edition. New York: Routledge. 29-49. Nebes, R.D. 1989. Semantic memory in Alzheimer's disease. Psychological Bulletin. 106 (3), 377-394. Nebes, R.D. 1985. Preservation of semantic structure in dementia. In H. Ulatowska (Ed.), The Aging Brain: Communication in the Elderly. San Diego CA: College-Hill Press. 109-122. 59 Nebes, R.D, & Brady, C.B. 1990. Preserved organization of semantic attributes in Alzheimer's disease. Psychology and Aging. 5 (4), 574-579. Nebes, R.D., Brady, C.B., & Jackson, S.T. 1989. The effect of semantic and syntactic structure on verbal memory in Alzheimer's disease. Brain and Language. 36, 301-313. Neely, J.H. 1977. Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading activation and limited-capacity attention. Journal of Experimental Psychology General. 106 (3), 226-254. Neils, J., Brennan, M . M . , Cole, M . , Boiler, F., & Gerdeman, B. 1988. The use of phonemic cueing with Alzheimer's disease patients. Neuropsychologia. 26 (2), 351-354. Nicolas, L.E., & Brookshire, R.H. 1993. A system for quantifying the informativeness and efficiency of the connected speech of adults with aphasia. Journal of Speech and Hearing Research. 36, 338-350. Nicolas, 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. 405-410. Osimani, A., & Freedman, M . 1995. Functional anatomy. In R. Lubinski (Ed.), Dementia and Communication. San Diego, CA: Singular. 22-46. Palm, S. 1995. A Case Study in Alzheimer's Disease and Language Impairment. Unpublished Graduating Essay in the Faculty of Graduate Studies: The School of Audiology and Speech Sciences, University of British Columbia, Vancouver, Canada. Palm, S., & Purves, B. 1996. Management of a word-finding deficit in discourse: A case example. Journal of Speech-Language Pathology and Audiology. 20 (2), 155-166. Patterson, K., & Hodges, J.R. 1995. Disorders of semantic memory. In A . D . Baddeley, B.A. Wilson & F.N. Watts (Eds). Handbook of Memory Disorders. London: John Wiley & Sons. 167-186. Ripich, D.N., & Terrell, B.Y. 1988. Patterns of discourse cohesion and coherence in Alzheimer's disease. Journal of Speech and Hearing Disorders. 53. 8-15. Robinson, J.A. 1992. Autobiographical memory. In M . Gruneberg & P. Morris (Eds.), Aspects of Memory Second Edition Volume I: The Practical Aspects. New York: Methuen. 223-251. Sagar, H.J., Cohen, N.J., Sullivan, E.V., Corkin, S., & Growdon, J.H. 1988. Remote memory function in Alzheimer's disease and Parkinson's disease. Brain. 111. 185-206. Sagar, H.J., Sullivan, E.V. & Corkin, S. 1991. Autobiographical memory in normal aging and dementia. Behavioural-Neurology. 4_ (4), 235-248. Salmon, D.P., Heindel, W.C., & Butters, N. 1995. Patterns of cognitive impairment in Alzheimer's disease and other dementing disorders. In R. Lubinski (Ed.), Dementia and Communication. San Diego, CA: Singular. 37-46. 60 Schmidt, J. 1996. Quantitative and qualitative memory changes in Alzheimer's patients. Unpublished Doctoral Thesis. Psychologische Institut der Alberta Ludwigs-Universitat, Freiburg, Germany. Shewan, C . M . 1988. The Shewan Spontaneous Language Analysis (SSLA) system for aphasic adults: description, reliability, and validity. Journal of Communication Disorders, 21 (2), 103-138. Ska, B., & Guenard, D. 1993. Narrative schema in dementia of the Alzheimer's type. In H.H. Brownell & Y. Joanette (Eds.), Narrative Discourse in Neurologically Impaired and Normal Aging Adults. San Diego, C A : Singular. 299-316. Ska, B., & Joanette, Y. 1996. Discourse in older adults: Influence of text, task and participant characteristics. Journal of Speech-Language Pathology and Audiology. 20 (2), 101-108. Smith, S., Butters, N., & Granholm, E. 1988. Activation of semantic relations in Alzheimer's and Huntingdon's disease. In H.A. Whitaker (Ed.). Neuropsychological Studies of Non-Focal Brain Damage. New York: Springer-Verlag. 265-285. Snowden, J., Griffiths, H. , & Neary, D. 1994. Semantic dementia: Autobiographical contribution to preservation of meaning. Cognitive Neuropsychology. 11 (3), 265-288. Stein, N.L., & Glenn, C.G. 1982. Children's concept of time: The development of a story schema. In W.J. Friedman (Ed.), The Developmental Psychology of Time. London: Academic. 255-282. Swihart, A . A . & Pirozzolo, F.J. 1988. The neuropsychology of aging and dementia: Clinical issues. In H.A. Whitaker (Ed). Neuropsychological Studies of Non-Focal Brain Damage. New York: Springer-Verlag, 1-60. Tulsving, E. 1983. Elements of Episodic Memory. Oxford: Oxford University Press. Ulatowska, H . (Ed.). 1985. The Aging Brain: Communication in the Elderly. San Diego, CA: College-Hill Press. Ulatowska, H.K., Allard, L., & Chapman, S. 1990. Narrative and procedural discourse in aphasia. In Y. Joanette & H.H. Brownell (Eds). Discourse Ability and Brain Damage: Theoretical and Empirical Perspectives. New York: Springer-Verlag. 180-198. Ulatowska, H.K., Allard, L., Donnell, A. , Bristow, J., Haynes, S.M., Flower, A. , & North, A.J. 1988. Discourse performance in subjects with dementia of the Alzheimer type. In H.A. Whitaker (Ed), Neuropsychological Studies of Non-Focal Brain Damage. New York: Springer-Verlag. 108-131. Ulatowska, H.K., & Chapman, S.B. 1995. Discourse studies. In R. Lubinski (Ed.), Dementia and Communication. San Diego: Singular. 115-132. Van der linden, M . , de Partz, M.P., Schils, J.P.,& Seron, X. 1992. Semantic and autobiographical memory: Neuropsychological dissociations? In M.A. Conway, D.C. Rubin, H. Spinnler, & W.A.Wagenaar (Eds.), Theoretical Perspectives on Autobiographical Memory. Dordretch: Kluwer, 473-492. 61 Warrington, E.K. 1975. The selective impairment of semantic memory. Quarterly Journal of Experimental Psychology. 27. 635-657. Weingarter, H. , Grafman, J.M Boutelle, W., Kaye, W., & Martin, P.R. 1983. Forms of memory failure. Science. 22 (4608), 380-382. Weingarter, H. , Kaye, W., Smallberg, S.A., Ebert, M.H. , Gillin, J.C., & Sitaram, N. 1981. Memory failures in progressive idiopathic dementia. Journal of Abnormal Psychology. 90 (3), 187-196. Wirth, D. 1991. Discourse Facilitation of Language Production by a Nonfluent Aphasic Adult. Unpublished Master of Science Thesis in the Faculty of Graduate Studies: The School of Audiology and Speech Sciences, University of British Columbia, Vancouver, Canada. Ylvisaker, M . , and Szekeres, S.F. 1994. Communication disorders associated with closed head injury. In R. Chapey (Ed.), Language Intervention Strategies in Adult Aphasia: Third Edition. Baltimore: Williams & Wilkins. 546-568. 62 Appendix 1: Assessment Results for K D 1 Arizona Battery for Communication Disorders if Dementia: Construct Scores Mental Status Episodic Memory Linguistic Expression Linguistic Comprehension Visuospatial Construction 3.0 3.5 2.6 2.6 4.0 Note. Construct scores are standardized to a 5 point scale, with upper values discriminating performance by normal elderly, and lower values discriminating performance by patients with dementia. Modified Boston Naming Test Confrontation Naming (Time 1) Confrontation Naming (Time 2) Multiple Choice Version 13 15 27 Note. Maximum score = 30. Confrontation naming tests used even-numbered items. Multiple choice version used odd-numbered items with semantic distractors. 1 By permission of authors (Palm and Purves, 1996, p.161). 63 Appendix 2: Noun Specificity Survey Instructions: Some words are specific in their meaning and some are more general. Please rank the following words on a scale of 1 to 5 in terms of how specific or informative you think they are, 1 being "very general" and 5 being "very specific". For example, 1 2 3 4 5 1. Stuff 2. Helicopter NOUNS Mean Standard Deviation l . Ida 4.7 0.8 3.1 0.9 3.trouble 2.6 1.0 4.time 2.3 0.7 3.7 0.8 6.plane 3.4 0.8 7. arrangements 2.2 1.2 3.1 1.0 9.year 3.1 1.1 3.1 0.6 11 .minister 3.6 1.0 12.people 1.9 0.7 3.4 0.8 14.kinds of things 1.1 0.3 15.notes 2.3 1.0 16.way 1.7 0.7 17. stage 2.9 0.6 18.Nova Scotia 5.0 0 19. wife 4.0 0.8 3.1 0.7 21. Margaret 4.7 0.5 22.summer 3.5 0.7 2 3. week 3.5 0.9 24.friends 3.2 0.8 25. Alzheimer's 4.7 0.8 26.setup 2.1 0.6 27.end 2.7 0.8 3.1 0.8 2 9 . M c G i l l 4.7 0.6 30.child 3.0 0.5 31.husband 4.1 0.7 3.4 0.7 33.a thought 1.8 1.1 1.4 0.5 3 5.downstairs 3.1 1.0 3 6.Daisy 4.7 0.6 3 7.troubles 2.1 0.6 2.1 0.5 3 9. Montreal 4.9 0.3 40.beginning 2.7 1.0 41.father 3.9 0.7 42.mother 3.9 0.7 43 .home 3.3 0.8 44. situation 1.8 0.9 45 .electricity 3.9 0.8 46.thing 1.0 0 4.2 0.9 side 2.7 0.8 49.any thing 1.1 0.4 50.point 2.2 0.8 3.1 0.9 2.6 0.8 5 3. area 1.9 0.7 54.night 3.5 1.0 55. set of other pants 3.9 0.9 56.minutes 2.9 1.0 57. plate of food 2.9 1.0 5 8. war 3.2 0.8 59.bed 3.9 0.8 60.fella 2.4 0.6 61.idea 2.3 1.0 6 2. water 3.7 1.0 63.story 2.7 0.5 64.girl 3.7 0.8 65. work 2.1 0.5 66.woman 3.5 1.0 67.daughter 3.7 1.0 68.reason 2.3 0.9 69.every thing 1.2 0.4 70.shoe 3.4 0.9 71 .person 2.7 1.0 72.nothing 2.1 1.2 73.prince 3.7 0.6 74.side 2.3 0.8 75.moment 2.5 1.0 76.foot 3.6 0.9 7 7. slipper 3.8 0.6 78.ending 2.9 1.0 79.misery 3.1 1.6 80.babies 3.4 0.7 81.everybody 1.7 0.6 82.roofs 3.1 1.0 83.feeling 1.7 0.7 84.part 1.5 0.7 85.Washington 4.3 0.6 86.models 2.8 0.4 87.killing 3.2 0.9 8 8. Jackie 4.5 0.8 89.boy 3.5 0.8 3.6 0.8 91 .twinkle 3.5 1.1 92.eye 3.8 1.0 2.6 1.1 94.wisdom 2.7 1.1 95.requirements 2.2 1.0 96.edge 2.9 0.9 97.Uncle Harry 4.8 0.6 9 8.pattern 2.6 0.8 99.somebody 1.4 0.6 lOO.pipe 3.5 0.9 101.enjoyment 2.5 0.8 102.fuss 2.5 1.0 103.a smoke 3.7 0.7 2.3 1.0 105.mouth 4.1 0.8 106.possibility 2.1 1.0 107.guy 2.6 0.8 108.soup 3.5 1.1 109.cup of coffee 4.5 0.5 110.animals 2.2 1.1 111 .omelettes 3.7 1.0 2.5 0.9 113.set of table(s) 3.2 1.3 114.milk 3.3 0.8 115.tea 3.9 0.9 3.1 0.9 117.funniness 2.3 1.3 118.times 1.9 0.9 119.religion 2.5 0.7 120.something 1.2 0.4 121.B.C. 4.3 0.6 122. sisters 4.1 0.7 12 3. some where 1.2 0.4 124.whatever 1.0 0 2.7 1.1 Appendix 3: Nouns with a Standard Deviation Less Than 1.0 NOUNS Mean Standard Deviation l . Ida 4.7 0.8 3.1 0.9 3.time 2.3 0.7 3.7 0.8 5.plane 3.4 0.8 3.1 0.6 7.people 1.9 0.7 3.4 0.8 9.kinds of things 1.1 0.3 10.way 1.7 0.7 ,11. stage 2.9 0.6 12.Nova Scotia 5.0 0 13 .wife 4.0 0.8 3.1 0.7 15. Margaret 4.7 0.5 16. summer 3.5 0.7 17. week 3.5 0.9 18.friends 3.2 0.8 19. Alzheimer's 4.7 0.8 20. setup 2.1 0.6 2 L e n d 2.7 0.8 3.1 0.8 2 3 . M c G i l l 4.7 0.6 24.child 3.0 0.5 25 .husband 4.1 0.7 3.4 0.7 1.4 0.5 28.Daisy 4.7 0.6 29.troubles 2.1 0.6 2.1 0.5 31 .Montreal 4.9 0.3 3 2.father 3.9 0.7 3 3.mother 3.9 0.7 34.home 3.3 0.8 3 5. situation 1.8 0.9 36.electricity 3.9 0.8 37.thing 1.0 0 4.2 0.9 3 side 2.7 0.8 40.anything 1.1 0.4 41.point 2.2 0.8 3.1 0.9 43 .clothing 2.6 0.8 44.area 1.9 0.7 45. set of other pants 3.9 0.9 46. war 3.2 0.8 47.bed 3.9 0.8 4 8 . T V 4.2 0.9 49.fella 2.4 0.6 5 O.story 2.7 0.5 51.girl 3.7 0.8 2.1 0.5 5 3.reason 2.3 0.9 54.everything 1.2 0.4 5 5. shoe 3.4 0.9 56.prince 3.7 0.6 57.side 2.3 0.8 5 8.foot 3.6 0.9 5 9. slipper 3.8 0.6 60.telephone 4.3 0.7 61 .babies 3.4 0.7 62.everybody 1.7 0.6 63.feeling 1.7 0.7 64.part 1.5 0.7 65 .Washington 4.3 0.6 66.models 2.8 0.4 67.kill ing 3.2 0.9 6 8. Jackie 4.5 0.8 69.boy 3.5 0.8 3.6 0.8 71.edge 2.9 0.9 72.Uncle Harry 4.8 0.6 7 3.pattern 2.6 0.8 74.somebody 1.4 0.6 75.pipe 3.5 0.9 76.enjoyment 2.5 0.8 77.a smoke 3.7 0.7 78.mouth 4.1 0.8 79.guy 2.6 0.8 80.cup of coffee 4.5 0.5 2.5 0.9 82.milk 3.3 0.8 8 3.tea 3.9 0.9 3.1 0.9 85.times 1.9 0.9 86.something 1.2 0.4 87 .B .C. 4.3 0.6 8 8. sisters 4.1 0.7 8 9. some where 1.2 0.4 90.whatever 1.0 0 91 .religion 2.5 0.7 68 u C <u <u . 3 « 3 9 6U O .5 o ts "S H J" t« to ._ <u © I fa H BH I3 c CO t: o a W) c .« B •5^ ,3 CJ > O cj '1 i5 1 p cs O- 2 L. cs ed H W U & i -25 > o I * ad •a o e CJ 5 £ 4 u c a s o o 3 "8 I. CJ to 3 CJ > O CJ '•a o -8 •a CJ > O B 3 .a T3 -8 B B S * *> la 2° « B * * O CJ U « a J3 cj _> c X CJ DC 2 *" H fa U '"8 6 o o c 3 M l 8> 3 O 3 CJ <N CJ O u C CJ T3 C o I <^  CN T3 .3 CN 3 I T3 .1 CN I 6 9 Quebec Referen-dum difficult | comfortable | long | ordinary | First Teaching Position Assassina-tion of JFK strong comfortable huge Trip to Eastern Canada pleasant strong awful long bad Jean Chretien strong able honest awful Father pleasant strong warm looking Jim •a pleasant strong g. £ comfortable huge alone long bad la 1 nice looking terrific Winston Churchill wrong strong able sensible valuable honest difficult awful ts g large quick Going to a Restaurant sour •a s 0} easy wrong attractive pleasant Telling a Fairytale: Cinderella oo C N O s C N 0 c n c n C N co CO co •<t co i n c o VO co r -co 00 co O s co 0 5 C N c o >n VO 00 O s •* 0 i n i n C N • n CO • n i n >n <n Quebec Referen-dum far 1 ridiculous | bothersome | First Teaching Position funny special easy 43 W • c troublesome Assassina-tion of JFK poor terrible clear Trip to Eastern Canada far special sick baby gorgeous fun terrible Jean Chretien poor funny different particular aware Father fine upset highest strongest better crooked not straight normal 1 Jim Winston Churchill Going to a Restaurant Telling a Fairytale: Cinderella </-> 00 in ON o vo vo CN VO CO vo vo IT) vo vo vo r-VO oo vo ON VO o r- r-CN r-co >/-> r~ vo' r- 00 r-Ov o 00 oo CN oo J 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items