Skip to main content

Grammar and memory: Evidence from agrammatic aphasia and probable Alzheimer's disease in German, Italian and Greek

Final Report Summary - MEMOGRAM (Grammar and memory: Evidence from agrammatic aphasia and probable Alzheimer's disease in German, Italian and Greek)

In Europe, there is an increasing number of people with neurological disorders, such as probable Alzheimer’s disease (pAD) or aphasia, which often result in severe language and memory impairments. Language is important for communication and daily living, but there are still significant gaps in our knowledge about how language is affected in pAD, especially at the (morpho)syntactic level. Agrammatic aphasia is an aphasia type characterized by selective deficits in (morpho)syntax. While some (morpho)syntactic phenomena have been repeatedly examined in agrammatic aphasia (e.g. subject-verb Agreement and Relative Clauses), others have received little attention thus far (e.g. Mood and Negation). Only a few studies have investigated (morpho)syntactic phenomena in pAD. It has been argued that memory deficits contribute to the language problems aphasic and pAD individuals face and, further, can predict these populations’ patterns of linguistic impairment. Interestingly, working memory (WM) deficits appear to differentially affect morpho(syntactic)/functional categories, resulting in a selective breakdown of grammar. However, there is no consensus as to the source of memory related processing difficulties for certain categories.
The project MemoGram aimed at: (1) gaining further insight into the way agrammatic aphasia and pAD manifest themselves across different languages; (2) exploring the relationship between memory systems such as WM and functional/(morpho)syntactic categories; (3) shedding light on the precise relationship between (morpho)syntactic phenomena and memory detecting the factors that contribute to the processing load of a given category/phenomenon; and (4) informing treatment programmes for German-, Italian-, and Greek-speaking agrammatic and pAD individuals.
Marie Curie (MC) fellow Valantis Fyndanis established collaborations in all three countries involved (Greece, Italy, Germany) and worked in all of them collecting data and/or interacting with colleagues that contributed to the project.
To address objective 1, we investigated the ability of German-, Italian-, and Greek-speaking individuals with agrammatic aphasia and pAD to process a number of functional/(morpho)syntactic categories, some of which have been paid little attention thus far: subject-verb Agreement, Tense/Time Reference, Aspect, sentential Negation, Mood, gender Agreement, and structural Case. We used a wide array of constrained tasks. Selective patterns of performance emerged, which differed across languages. In the verbal domain, for example, the Italian aphasic participants (N=10) performed significantly better on Negation than on Tense and Agreement, and significantly better on Tense/Agreement than on Mood. In the Greek aphasic participants (N=10), on the other hand, Mood and Agreement were found to be the best preserved categories, followed by Negation and Tense. Aspect was the most impaired category. The German aphasic participants exhibited the pattern Negation > Mood/Agreement > Tense. Several factors have been argued to determine the patterns of performance observed in agrammatic aphasia, such as involvement of integration processes in categories like Tense and Aspect (e.g. Fyndanis, Varlokosta, & Tsapkini, 2012), Tense underspecification (Wenzlaff & Clahsen, 2004, 2005), involvement of inflectional alternations (Wang, Yoshida, & Thompson, 2014), and position of a given category in the syntactic hierarchy (Friedmann & Grodzinsky, 1997). The data collected allow us to test the relevant theories/hypotheses taking a cross-linguistic approach. In September 2015 Valantis Fyndanis organized a workshop at the University of Potsdam, where he presented those cross-linguistic data that are critical for testing the Tense Underspecification Hypothesis (TUH) (Wenzlaff & Clahsen, 2004, 2005). The results of this project do not lend support to the TUH. In fact, as will be pointed out in the articles that will be published in the near future, the results do not lend support to any of the hypotheses associated with the factors mentioned above, suggesting that a unitary account is unlikely to succeed (see also Miceli, Silveri, Romani, & Caramazza, 1989). Testing these hypotheses –which is related to objective 3– taking a cross-linguistic approach is one of the significant contributions MemoGram has made. This is especially true because the majority of these hypotheses had not been tested before. As a next step, possible interactions of the various variables/factors will be explored by means of sophisticated statistical methods (e.g. generalized linear models, neural networks, clustering and learning algorithms). (This will be related to objectives 2 and 3.)
To meet objectives 2 and 3, we also designed and implemented a study to which no reference had been made in the original project description. In particular, Valantis Fyndanis interacted with Prof. David Caplan (Harvard Medical School), a world expert in memory research, and they agreed that the number of aphasic participants recruited (i.e. 10 in each language) would not allow them to reliably investigate the relationship between memory systems (such as WM) and (morpho)syntactic phenomena. Since pathology exacerbates behavioural trends/patterns of performance observed in healthy individuals (e.g. Miyake, Carpenter, & Just, 1994), the MC fellow and Prof. Caplan decided to focus on a sufficiently large number of healthy speakers presenting a large variability in WM capacity. To this end, they tested 103 neurologically-intact native speakers of Greek, aged 22-85, with cognitive/WM tasks and a constrained linguistic task that tapped into the production of Agreement, Tense, and Aspect. The linguistic task also manipulated a design variable, namely distance between cue and target, which was assumed to affect performance. The main findings of this study were that (1) cue-target distance matters (this finding relates to objective 3), and (2) verbal WM capacity differentially affects Agreement, Tense, and Aspect (affecting more Aspect than Tense and not affecting Agreement at all); this interaction, however, was only observed in individuals with lower educational levels. Interestingly, at the group level, Aspect was found more impaired than Tense, and Tense more impaired than Agreement. The same pattern has been reported for Greek-speaking agrammatic aphasia (e.g. Fyndanis et al., 2012). Therefore, the differences in the results of the healthy individuals who participated in this study and the aphasic participants reported in other studies (op. cit.) are quantitative and not qualitative, which is consistent with the claim that pathology exacerbates behavioural trends observed in healthy individuals (e.g. Miyake et al., 1994).
As far as Alzheimer’s disease is concerned, originally we targeted to recruit and test 10 pAD individuals per language. Since MemoGram has been a very ambitious project involving testing in three countries, various collaborations in the countries involved had to be established. The collaborations established have been very successful in all cases except for the one concerning the Alzheimer’s part in Germany. This, coupled with the heavy workload of the MC fellow (partly associated with the above-mentioned additional study on healthy Greek-speaking individuals), led to the decision to drop the German pAD part of the project. On the other hand, thanks to the collaborations established in Italy and Greece we recruited and tested more pAD individuals (16 in Greece and 14 in Italy) than initially planned to ensure greater statistical power and increase the chance of publishing the results in well-established international peer-reviewed journals. The patterns of performance exhibited by the Greek- and Italian-speaking pAD participants are similar to those exhibited by the aphasic participants in these two languages. This suggests that (morpho)syntactic impairments are not neurological condition-specific, but they probably emerge due to the synergistic effect of two facts: (1) in each language, some morphosyntactic phenomena are more demanding in terms of processing resources than others, and (2) neurologically impaired individuals such as aphasic and pAD speakers have reduced processing resources (which is reflected in their limited WM capacity).
The findings of the basic research that has been carried out are also expected to be useful to clinicians in Europe, as MemoGram has produced new knowledge about the way agrammatic aphasia and pAD manifest themselves in Greek, Italian, and German, as well as about the nature of the linguistic deficits observed in these neurological conditions. Therefore, MemoGram will inform treatment, which will have an obvious beneficial impact on the quality of life of aphasic and pAD individuals, of their carers/family, and on society and economy at large. Valantis Fyndanis has already started working with a well-established clinician and researcher, Prof. Maria Kambanaros (Cyprus University of Technology), and soon they will co-author an article with suggestions to clinicians in Greece, Italy, and Germany. (This relates to objective 4.)
Valantis Fyndanis presented the results of MemoGram at many international peer-reviewed conferences and workshops. Overall, he delivered 16 lectures and poster presentations, 5 of which were invited talks. As mentioned above, he also organized an Open Workshop on Cross-linguistic Research on Aphasia at the University of Potsdam, and contributed to the organization of the Collaboration of Aphasia Trialists’ (CATs) conference Future Directions for Aphasia Research, which was held at the City Univeristy in London. In both events, he presented part of the MemoGram data. He has also published five peer-reviewed conference papers in international journals and one conference proceedings paper, and is about to submit for journal publication three full-blown research papers (see deliverables attached). Moreover, he is currently preparing another paper and he has planned 10 articles that will report data collected during his MC fellowship.
Lastly, a Facebook Page has been developed which provides information about the project and MC Actions ( ). The personal webpage of the MC fellow ( ) has also been updated with information about MemoGram and Marie Curie Actions. An audio clip has also been created (podcast), in which Valantis Fyndanis briefly presents his MC project. This audio clip is available on the personal webpage of the MC fellow.

Miceli, G., Silveri, M. C., Romani, C., & Caramazza, A. (1989). Variation in the pattern of omissions and substitutions of grammatical morphemes in the spontaneous speech of so-called agrammatic patients. Brain and Language, 36, 447-492.
Miyake, A., Carpenter, P. A., & Just, M. (1994). A capacity approach to syntactic comprehension disorders: Making normal adults perform like aphasic patients. Cognitive Neuropsychology, 11, 671-717.
Friedmann, N. & Grodzinsky, Y. (1997). Tense and agreement in agrammatic production: Pruning the syntactic tree. Brain and Language, 56, 397-425.
Fyndanis, V., Varlokosta, S., & Tsapkini, K. (2012). Agrammatic production: Interpretable features and selective impairment in verb inflection. Lingua, 122, 1134-1147.
Wang, H., Yoshida, M., & Thompson, C. K. (2014). Parallel functional category deficits in clauses and nominal phrases: The case of English agrammatism. Journal of Neurolinguistics, 27, 75-102.
Wenzlaff, M. & Clahsen, H. (2004). Tense and agreement in German agrammatism. Brain and Language, 89, 57-68.
Wenzlaff, M. & Clahsen, H. (2005). Finiteness and verb-second in German agrammatism. Brain and Language, 92, 33-44.