Speech and language production in Alzheimer’s disease

This special issue presents papers on the characteristics of speech and language production in Alzheimer’s disease (AD), ranging from the level of the single word to discourse. Although language deficits were not historically considered a hallmark of AD, recent updates to diagnostic criteria for AD recognize word-finding difficulties as an early sign, reflecting research findings of the past two to three decades. The papers presented have implications for both theoretical models of production and for clinical assessment. Three main themes emerge from the five papers of this special issue of Aphasiology.

First, the studies presented here demonstrate that the investigation of language difficulty in AD has moved beyond the single word level. Whereas anomia has been long identified as an indicator of language difficulties in AD, researchers today recognize the need to examine word-finding difficulties in the context of communication in connected speech. This approach is evident in Abdalla, Rudzicz, and Hirst (this issue), and in Kavé and Goral (this issue). In addition, characterizing lexical retrieval difficulties beyond the single word level extends also to understanding phonetic, phonological, semantic, and morphological aspects of word retrieval, as addressed in the papers of Cera, Ortiz, Bertolucci, and Minett (this issue), Fyndanis et al. (this issue), and Kavé and Goral (this issue). A second, related, theme is the call for automated analysis of language production. If the language abilities and disabilities of individuals with AD are to be analysed in the context of connected speech production, automated analyses could alleviate some of the challenges associated with analysing connected speech, such as the time-consuming and error-prone nature of the work. Both Kavé and Dassa (this issue), and Abdalla et al. address this exciting development. A third theme that emerges from the works presented here concerns stages of severity of the disease. The progressive changes associated with AD are inherent to the nature of the dementia. A challenge that researchers of AD face is determining the extent to which language impairment associated with AD can be examined across a range of individuals and severity levels. Two studies address this issue directly, Fyndanis et al. and Kavé and Dassa (this issue), and other studies have wrestled with it as the authors define their study population. Beyond these themes, each paper contributes to our understanding of specific aspects of language changes in AD.

Indeed, in Kavé and Goral’s paper, the manifestation of word finding difficulties in connected speech production are investigated in a meta-analysis. This analysis was conducted on the performance of over one thousand individuals with AD and over one thousand healthy controls. The findings indicate that lexical retrieval difficulties are present in the connected speech deficits found in AD patients. The nature of many of these lexical retrieval difficulties (e.g., use of pronouns, paraphasias), is further characterized in this investigation.

The paper by Cera and colleagues addresses whether the underlying nature of word production errors is more properly characterized by motor level (i.e., phonetic) or language level (i.e., phonological) errors reflecting apraxia of speech or aphasia, respectively. Although usually thought of as being primarily phonological, Cera et al.’s findings point to repetition errors in AD that are reflective of a combination of phonetic and phonological changes, supporting theoretical models that posit interactive activation of phonological and articulatory stages of speech production.

Moving beyond the single word level, Fyndanis and colleagues investigate potential morphosyntactic impairments related to agreement, time reference, and mood in Greek- and Italian-speaking individuals with mild-to-moderate AD. Their work demonstrates that morphosyntactic impairments do exist, but that they are variable within and across the two languages, and seem only partially related to severity and to working memory deficits. Interestingly, the results are discussed relative to two hypotheses originally posited to explain agrammatic aphasia.

The final two papers in this special issue have in common an interest in characterizing the structure of the connected speech of individuals with AD by the use of automated analyses. Both also point to the potential clinical use of automated analyses for early detection of AD and to track language decline. In Abdalla et al., discourse relations (e.g., elaboration, logical contingency) are automatically extracted in spontaneous speech (using data from existing datasets) and written novels of authors with and without AD. The results indicate alterations in discourse relations as a function of AD in both speech and writing. Kavé and Dassa look more closely at the relationship between disease severity and lexical characteristics (both semantic and morphological) of the picture descriptions produced by individuals with AD. Their results suggest a relationship between disease severity and the semantic characteristics of impaired connected speech (e.g., lexical diversity, word frequency, information content), but not with the grammatical features of language.

Together, the papers contribute important findings to our understanding of language deficits in AD and highlight a number of issues that we think will propel the field forward into the next few decades. Future areas might include investigation of the similarities and differences in language difficulties that are revealed by studying different aspects of the language code from single words through to discourse; whether the same patterns are observed across different languages and/or in bilingual individuals; what the progression of impairment might be with disease severity and whether language difficulties can be detected in an even earlier prodromal stage, such as in individuals with mild cognitive impairment. In addition, new methods of analysis, such as the automated ones used in two of the articles in this issue, may well provide the field with reliable and expedient ways to investigate these important questions. Our continuing study of patterns of language impairment in AD will further both theoretical aspects and clinical considerations of dementia and its implications.

To conclude, we thank all the authors and reviewers for their contributions, with a special thanks to Chris Code for the opportunity to serve as editors of this special issue.

Disclosure statement

No potential conflict of interest was reported by the authors.