On the processing of sentential negation in dyslexic adults. An ERP study.

ll presente lavoro nasce allo scopo di rispondere a due principali obiettivi: (a) contribuire al dibattito generale sulle dinamiche temporali dell’interpretazione della negazione linguistica; (b) approfondire l’analisi delle difficoltà di interpretazione delle frasi negative presentate oralmente da parte di soggetti dislessici . A questo scopo, il lavoro presenta i risultati di due studi comportamentali preliminari e due studi EGG (Studio 1 e Studio 2). Lo Studio 1 è stato elaborato con l’obiettivo di fornire ulteriori conferme all’approccio ‘non-incrementale’ alla negazione. Recentemente, due studi ERP sono stati svolti allo scopo di approfondire le dinamiche temporali che sottostanno all’interpretazione della negazione linguistica. Lüdtke, et al. (2008) hanno analizzato i potenziali evocati associati all’interpretazione di frasi negative e affermative vere e false durante l’esecuzione di un sentence-picture verification task. I risultati hanno mostrato che durante le prime fasi di interpretazione i potenziali evento-correlati delle frasi negative non sono influenzati né dal significato effettivo né dalla polarità della frase presentata ma soltanto dalla corrispondenza/non corrispondenza tra le entità raffigurate nell’immagine e quelle menzionate nelle frase precedentemente ascoltata. Lo studio di Nieuwland & Kuperberg (2008) si è servito di frasi licenziate pragmaticamente, ovvero, frasi affermative e negative introdotte da una semplice proposizione che, restringendo le condizioni di verità dell’enunciato successivo, ne rendeva l’uso più adeguato dal punto di vista pragmatico. I risultati hanno mostrato che, in queste condizioni, l’effetto N400 rifletteva il valore di verità della frase, come dimostrato dalla maggiore ampiezza dell’effetto nelle condizioni false in confronto con le condizioni vere, indipendentemente dalla polarità della frase. Lo Studio 1 riportato nel presente lavoro si è riproposto di replicare i risultati ottenuti da Lüdtke et al. (2008) tramite l’utilizzo di un sentence-picture verification task pragmaticamente più adeguato. Lo studio ha coinvolto un gruppo di 27 adulti italiani (età media 27;6). I risultati comportamentali e elettrofisiologici ottenuti sono in linea con i modelli non-incrementali d’interpretazione della negazione. Lo Studio 2 ha esteso l’analisi a un gruppo di adulti dislessici italiani (15 soggetti, età media 22;4). I partecipanti dello Studio 1 sono stati inclusi come gruppo di controllo. La motivazione dell’inclusione dei soggetti dislessici è costituita da studi comportamentali precedenti che hanno osservato una maggiore difficoltà nell’interpretazione delle frasi negative da parte di soggetti dislessici italiani rispetto a soggetti non dislessici della stessa età (Vender & Delfitto, 2010; Rizzato, Scappini, Cardinaletti, 2014). Una serie di test volti alla valutazione della memoria di lavoro dei partecipanti sono stati inclusi in questa seconda parte dello studio. I risultati di questi test hanno messo in luce l’esistenza di significative differenze nella memoria di lavoro dei soggetti dislessici e non dislessici coinvolti. L’analisi del sentence-picture verification task ha evidenziato un generale rallentamento nell’esecuzione del test da parte dei soggetti dislessici in confronto ai soggetti non-dislessici in tutte le condizioni sperimentali, probabilmente derivante dalla difficoltà di esecuzione del compito sperimentale stesso. L’analisi dei potenziali evocati, infine, ha suggerito che i dislessici adulti utilizzino strategie cognitive diverse da quelle utilizzate da adulti non-dislessici nell’interpretazione delle frasi negative.

The present work aimed at dealing with two main issues: (a) to contribute to the broad debate on the processing of negation trough an ERP investigation of negative sentences processing in Italian adults, still lacking in the literature on the topic; and (b) to provide further evidence and deepen analysis of dyslexics’ difficulty in the comprehension of orally presented negative sentences. To these purposes, the results of two behavioral preliminary studies and two main EEG-ERP studies (Study 1 and Study 2) are presented. The two preliminary studies involved a group of 10 adults (mean age 27;9) and a group of 33 Italian children (mean age 9;4). These two studies were principally intended to assess the experimental settings of the protocol and to evaluate its suitability for the performance of an EEG study with the two different populations. Besides having led to important adjustments in the experimental design, the results of the preliminary studies showed the difficulty of involving children in an EEG study of the type planned. As a consequence, the two EEG studies subsequently performed exclusively involved adult participants. Study 1 was intended to provide further confirmation to the non-incremental account of negation processing, according to which the interpretation of negative sentences inherently differs from the interpretation of other linguistic structures, in that it requires the comprehender to construct two different, subsequent mental simulations: a first representation corresponding to the negated state affairs, and a second referring to the actual, negative meaning of the sentence. This latter proposal is in contrast with alternative incremental approaches, according to which the processing of negative sentences does not intrinsically differ from that of affirmative sentences, except than for higher pragmatic sensitivity, which would determine longer processing times in all cases in which the negative sentence to be interpreted is not presented in adequately supportive linguistic contexts – as it mostly happens in experimental paradigms. Recently, two main EEG studies have been performed with the purpose of providing a more precise insight into the temporal dynamics of negation processing. Lüdtke, Friedrich, Filippis, & Kaup (2008) analyzed ERP effects associated with the execution of a sentence-picture verification task including both true and false affirmative and negative sentences. The results showed that the early stages of negative sentences verification are not influenced by the actual meaning of the sentence, but rather by the matching/mismatching between the picture content and the entity mentioned in the preceding sentence, resulting in larger N400 amplitudes for non-priming conditions (i.e., false affirmative and true negative sentences) than for priming conditions (i.e., true affirmative and false negative sentences). The fact that the entity mentioned in the sentence was not always depicted in the picture, however, determined a sense of pragmatic infelicity, possibly affecting event-related potentials in all cases in which the sentences involved predication about an entity not relevant to the following picture. To avoid this kind of influence, Nieuwland & Kuperberg (2008)’s ERP study focused on the interpretation of pragmatically licensed sentences, that is, affirmative and negative sentences introduced by a simple proposition restricting the conditions of validity of the target statement. In this latter study, N400 amplitudes reflected the actual sentences’ truth-value, reporting larger N400 amplitudes for false conditions than for true conditions, regardless of sentence polarity. Far from constituting a neutral discourse context, however, the short propositions inserted with the aim of ensuring pragmatic licensing determined a strong expectation for true sentences and, consequently, a sense of unnaturalness for false sentences that might represent the main cause of N400 amplitudes. In addition, the use of sentences related to general world-knowledge did not allow the experimenters to temporally and functionally distinguish the cognitive activity connected with the processing of the strictly linguistic material from that determined by truth-value evaluation, both crucially associated to N400 component. On the basis of these considerations, Study 1 aimed at replicating Lüdtke et al. (2008)’s findings through a sentence-picture verification task arguably free from the pragmatic infelicity characterizing that previous study. A group of 27 Italian adults (mean age 27;6) took part in Study 1. The behavioral and EEG results are mainly consistent with the non-incremental models of negation processing. In particular, the first stages of sentence verification were characterized by a N400 modulations very similar to that observed by Lüdtke and colleagues, revealing no initial sensitivity to negation. Importantly, the adoption of a pragmatically controlled design prevented the effect from being influenced by the simple priming between the picture and the sentence. In the following time windows, in addition, the emergence of electrophysiological components reflecting the truth-value and polarity effects provided important evidence of the successful integration of the negative meaning into the interpretative process. The Study 2 extended the investigation to a group of Italian dyslexic adults (15 subjects, mean age 22;4). The non-dyslexic participants of the Study 1 were included as control group. The rationale for the involvement of dyslexic subjects was provided by previous behavioral studies reporting increased difficulty in the interpretation of negative sentences by Italian dyslexic children (Vender & Delfitto, 2010) and adults (Rizzato, Scappini, Cardinaletti, 2014) in comparison with non-dyslexic peers. In order to verify Vender (2011) hypothesis, according to which the dyslexics’ difficulties in the spoken language comprehension may depend on their poor working memory capacity, a series of working memory tests were also included in the experimental protocol for this second study. The results of these latter tests highlighted significantly weaker phonological working memory in the dyslexic subjects in comparison with non-dyslexic controls. The analysis of the sentence-picture task performance showed an overall slowdown and poorer accuracy for dyslexic subjects in comparison with controls, suggesting that the dyslexics experienced more troubles in all conditions, most likely deriving from the working memory load associated with the execution of the experimental task. Although the negative sentences were characterized by slower response times and higher error rates than affirmative sentences, the performance resembled that observed in the control group. Therefore, no confirmation of specific difficulty in the interpretation of negative sentences by dyslexic subjects was reported. Contrary to Vender (2011), in addition, no clear evidence of working memory involvement in negative sentences interpretation was found. The analysis of the ERPs and, in particular, the observation in the dyslexic group of (a) a crucial lack of the N400 effect characterizing the first stage of sentence verification in the control group, and (b) the emergence of an early polarity effect, unreported in the control group, suggests that dyslexic adults may rely on different cognitive strategies for the interpretation of affirmative and negative sentences in comparison with non-dyslexic subjects. Worth noting, however, this latter hypothesis is at variance with the predictions of the two-step simulation hypothesis by Kaup and colleagues, according to which the interpretation of the negative sentences necessarily requires the initial simulation of the negated state of affairs.