if you agree - Università degli Studi di Verona
Transcript
if you agree - Università degli Studi di Verona
DYSLEXIA, PRONOUN INTERPRETATION AND VERBAL WORKING MEMORY Maria Vender Department of Germanic and Slavic Studies, University of Verona Lungadige Porta Vittora 1 – 37100, Verona, Italy E-mail: [email protected] Abstract In this contribution I present original data from an experimental research concerning the interpretation of referential expression performed on five groups of subjects: dyslexic children (mean age 9;15), age-matched control children (mean age 9;05), children attending the first class of a primary school (mean age 6,9), younger children (mean age 4;75) and control adults (mean age 30;25). The experiment tested the comprehension of Italian sentences containing zero and phonetically realized pronouns. Data show that dyslexics, unlike age-matched control children, are remarkably impaired in their capacity to correctly interpret referential expressions, while their performance does not differ significantly from the performance proper to younger children. Since it has been shown that retrieving the antecedent of a referential expressions is a complex task that imposes a considerable load on Working Memory, I propose that dyslexic children’s Working Memory is impaired. 1 1. Introduction 1.1. What is developmental dyslexia Developmental dyslexia can be defined as a specific impairment affecting the acquisition of reading and spelling skills, despite adequate intelligence, opportunity and social background, which occurs in absence of physical, neurological, emotional and socio-economical problems (Vellutino 1979). Recent studies show that dyslexia affects around 5-10% of the children, and that it is highly heritable: an individual who suffers from dyslexia has namely the 40%-50% of probability to have dyslexic children. Even though the existence of a specific impairment affecting reading and writing has been known since 1800, dyslexia has been considered as a specifically linguistic issue only in the last decades. Focus has been posed, in particular, on the phonological difficulties experienced by dyslexic people: studies have demonstrated that 100% of the dyslexic population exhibits difficulties in phonological processes, even though neither the exact locus, nor the nature of the phonological deficit have been identified yet. According to the phonological theory, dyslexia is caused by a cognitive deficit specific to the representation and the processing of speech sounds. Since learning to read requires gaining knowledge of the grapheme-phoneme correspondence, dyslexics have arguably difficulties in representing, storing or retrieving the sounds of their language and the respective grapheme-phoneme correspondence, and for this reason, they cannot learn to read and spell words properly. Evidence in favour of this theory comes from the results of experiments, showing that dyslexics perform poorly on tasks requiring phonological awareness, like conscious segmentation and manipulation of speech sounds (Snowling 1991, Ramus 2003, Szenkovits & Ramus 2005). Moreover, a series of experiments conducted in order to investigate the nature of the phonological deficit in developmental dyslexia, leads to the hypothesis that the phonological representations of dyslexics are intact, and that their impairment arises only in particular tasks, especially those involving short-term memory, conscious awareness and time constraints (Ramus & Szenkovits 2007). 2 However, even though the phonological domain has been the more investigated in the research conducted on developmental dyslexia since now, impairments have been found also in the vocabulary development and in the morpho-syntactic domain. (i) Vocabulary development: dyslexic children’s vocabulary has been shown to be underdeveloped in comparison to age-matched children (Scarborough 1990). Dyslexics also display a wordlength effect (the longer the word, the poorer the performance) and a frequency effect (the lower the frequency, the poorer the performance) (Wolf & Obregon 1992). (ii) Rapid naming: speeded naming of pictures, colours or letters have been found to be a predictor of reading ability. Dyslexic children exhibit more problems in comparison to age-matched control children, when required to rapidly name objects, colours and letters (De Jong & Van der Leij 1999, Wolf et. al 2002). (iii) Morpho-syntax: between 2 and 4 years of age, dyslexics manifest deficits in both comprehension and production of syntactic structures and morphemes marking morpho-syntactic information (Scarborough 1990). Moreover, Wilsenach and Wijnen (2003) showed that Dutch at-risk children from 18-23 months, differently from age-matched babies, are not able to distinguish sentences containing the correct temporal auxiliary combined with a past participle from agrammatical sentences. Furthermore, Joanisse et al. (2000) showed that 8-years old children made more errors in verb inflections in comparison to control children. (iv) Dyslexics have been found impaired in the comprehension of complex sentences; in particular, their ability to understand and repeat relative clauses (Mann et al., 1984; Stein et al., 1984, BarShalom et al. 1993) and to interpret passive sentences (Stein et al. 1984) is damaged. According to these studies, then, developmental dyslexia seems to be a specifically linguistic deficit which affects not only the phonological domain but also the morpho-syntactic domain. In this perspective, it would be important to perform specific studies in order to investigate dyslexics’ abilities in the semantic and pragmatic domain. For this reason, I developed an experimental protocol aiming to test dyslexic children’s comprehension of referential expressions, a task involving both semantic and pragmatic knowledge. 3 In order to assess if they exhibit an impaired behaviour, I also applied the experimental protocol to a group of age-matched children. I included in the sample a group of adults without linguistic problems, in order to check if there are significant differences between their performance and the control children’s performance, and two groups of younger children, aged 5 and 7 years old. 2. Comprehension and Production of Referential Expressions: the Accessibility Theory 2.1. What is Accessibility Theory? Referential expressions are considered as constants or variables that have to be assigned a referent in order to be interpreted, as shown in (1) and (2): (1) Frank explained the theory and then he left (2) Frank explained the theory and then x left, where x = Frank In this case, as described in (2), the pronoun he is interpreted as a variable which is assigned the same value as the subject of the sentence, i.e. Frank, which is its antecedent; technically, this operation is dubbed accidental coreference, since the pronoun is assigned the same reference as the antecedent. However, when there is more than one possible antecedent for a referential expression in a sentence, the situation appears to be more complicated to explain. Take for instance the Italian sentence in (3): (3) Il professorei ha spiegato la lezione allo studentej ed Øi,*j è uscito dall’aula “The professor explained the lesson to the student and Ø left the classroom” In (3), there are two possible antecedents for the pronoun “he”, i.e. il professore and lo studente. Nevertheless, Italian speakers have no troubles in identifying il professore as the correct antecedent of the pronoun. The Accessibility Theory developed by Ariel (1991) permits to explain how reference assignment works in this contexts, aiming to capture the ways in which the human mind is able to select the most appropriate antecedent for a referential expression. 4 The concept of accessibility plays an important role in the relation that is established between anaphoric expressions and antecedents. In an anaphoric relation, in fact, the anaphoric term refers back to an entity that has already been introduced in the discourse or activated by an antecedent. This entity is said to be accessible for the anaphoric expression. According to Accessibility Theory, some entities are more readily retrievable than others in the participant’s memory and the speaker can select, within a set of anaphoric expressions, the most appropriate one, in order to help the hearer access the referent. In other words, we can imagine that every anaphoric expression is provided with a label that informs the hearer about how salient the antecedent is in the discourse. In the Accessibility Theory framework, then, referential expressions are seen as accessibility markers which are ordered in a precise hierarchy from the maximum to the minimum degree. Their position in the scale provides the addressee appropriate instructions to retrieve the antecedent, by indicating how accessible it is in the discourse. To better understand how this mechanism works, I reported in (4) a very detailed version of Ariel’s Accessibility Hierarchy: (4) Accessibility Marking Scale Zero pronouns (Ø) < Reflexives < Agreement markers < Cliticized pronouns < Unstressed pronouns < Stressed pronouns < Stressed pronouns + gesture < Proximal demonstrative (+NP) < Distal demonstrative (+NP) < Proximal demonstrative (+NP) + modifier < Distal demonstrative (+NP)+ modifier < First name < Last name < Short definite description < Long definite description As indicated in (4), the zero pronoun is the highest accessibility marker: its position in the scale informs the addressee that the antecedent to be retrieved has to be the most accessible one; since the most accessible expression of a sentence, i.e. its topic, generally coincides with the subject, precisely the subject is selected 5 as the appropriate referent for the zero pronoun. This analysis permits then to explain why Italian speakers interpret the zero pronoun in (3) as referring to il professore, which is the subject of the sentence. Replacing the zero pronoun with a phonetically realized pronoun, instead, would give the sentence a quite different interpretation, as shown in (5): (5) Il professore ha spiegato la lezione allo studente e poi lui è uscito dall’aula “The professor explained the lesson to the student and the he left” According to the hierarchy in (4), the phonetically realized pronoun he occupies a lower position in the scale, in comparison to the zero pronoun, and, hence, it provides us with the information that the appropriate antecedent will not be the most salient expression, i.e. the subject, but a less salient one. For this reason, the addressee selects lo studente as the referent of the pronoun lui. Different factors contribute to determine the accessibility degree of an antecedent: salience, distance, competition and unity. Let us briefly consider these concepts: (i) Saliency: entities are salient in the discourse if they are mental representations of the participants of the conversation, or if they are discourse or sentence topics. (ii) Distance: it informs us about the distance between the anaphoric expression and its antecedent. Distance is determined by the number of NPs that occurs between the referential expression and its antecedents. Shorter is the distance, more accessible the antecedent. (iii) Competition: it informs us about the number of other possible antecedent candidates for an anaphoric expression. If there are more possible antecedents, the competition becomes higher and the antecedent is less accessible. (iv) Unity: it informs us about how related the unit in which the antecedent occurs is to the unit in which the anaphoric expression finds itself (i.e. it concerns the fact that the antecedent occurs in the same frame, world, point of view or segment of the paragraph). These factors interact with each other contributing to determine the degree of accessibility of an antecedent. 6 Ariel argues that the scale in (4) is essentially universal, since natural languages tend to encode the degree of accessibility relying on three principles: informativity, rigidity and attenuation. (i) Informativity: the degree of informativity of a marker depends on the lexical information it provides. If a form is more informative than another is, it is more likely that it will code a lower accessibility. Semantically more empty markers (as pronouns) signal higher accessibility. For instance, the long definite description the girl with the red hat is more informative than the pronoun she, and, hence, it is a lower accessibility marker. For this reason, if the speaker wishes to refer to a non salient antecedent, s/he will use a low but more informative accessibility marker. (ii) Rigidity: the degree of rigidity determines how uniquely referring an expression is. If a form is rigid (i.e. it refers unambiguously to an individual or an object, as proper names do), it marks lower accessibility. For example, the proper name Noam Chomsky is more rigid and unambiguous than the definite description the linguist, and, hence, it is a lower accessibility marker. (iii) Attenuation: the degree of attenuation depends on the phonological size of a form, and on the presence or absence of the stress. Less attenuated forms are used for lower accessibility retrievals. A stressed pronoun is more attenuated then an unstressed one, and, hence, it conveys higher accessibility. To sum up, then, the Accessibility Theory claims that a high accessibility marker, such as a zero pronoun, informs the addressee that the antecedent is highly accessible, hence salient, in the discourse, while a low accessibility marker specifies that the antecedent is not highly salient in the discourse. Hence, the hierarchy helps the addressee reduce the number of possible candidates to the minimum, excluding incorrect and improper candidates. 7 2.2. Processing costs of Accessibility Theory In the previous section we have seen that according to the Accessibility Theory referential expressions are accessibility markers ordered in a hierarchy and that their position in this scale provides the addressee with the information needed to retrieve an appropriate antecedent. In this paragraph, I will concentrate on the difference between zero pronouns and phonetically realized pronouns, focussing in particular on the processing costs that they impose on the general computation of the sentence’s meaning. 2.2.1. Zero pronouns and phonetically realized pronouns As argued above, zero pronouns (zeros henceforth) and phonetically realized pronouns (pronouns henceforth) are both high accessibility markers, even though not at the same degree. Pronouns, namely, are more rigid and fully articulated than zeros, which are, instead, uninformative, more ambiguous and attenuated. For this reason, zeros occupy the highest position in the accessibility marking scale in (4), followed by pronouns which mark a lower accessibility. Evidence in favour of this statement is provided by the generalized preference manifested by speakers to use a zero form to refer to very salient antecedents, when both zeros and pronouns are grammatical. The sentences in (6) and (7) exemplify this preference: although both utterances are grammatically correct, the former is generally preferred. Moreover, people hearing a sentence containing a pronoun referring to the topic of the sentence, as in (7), tend to attribute a slight contrastive meaning to the pronoun, deducing for instance that the speaker intended to say that John, and not Paul, is a good linguist. (6) Questo è Gianni: Anna dice che è veramente un bravo linguista “This is John: Ann says that Ø is really a good linguist” (7) Questo è Gianni: Anna dice che lui è veramente un bravo linguista “This is John: Ann says that he is really a good linguist” This tendency was noted as early as in Chomsky (1981), where the Avoid Pronoun Principle, as reported in (8), was formulated: 8 (8) Avoid Pronoun Principle Lexical pronouns are blocked by empty pronouns if possible So stated, however, this principle leaves some questions open, since it cannot explain why in some cases speakers show preferences for realized pronouns, instead of zeros, as reported by Ariel (1991) for languages like Hebrew. The Accessibility Theory, instead, is able to capture this tendency, arguing that once the accessibility of a given antecedent is perceived as relatively low, speakers tend to use a relatively lower accessibility marker, resorting to the pronoun. 2.2.2. Processing costs of zero pronouns’ and phonetically realized pronouns’ resolution: an hypothesis In this section, I will present an original hypothesis about the distinct processing costs required in the interpretation of zero pronouns and full pronouns, arguing that the resolution of overt pronouns is more complex, since it involves a finer reasoning and the activation of an implicature. Let us concentrate on Italian: as is well-known, Italian is a pro-drop language which allows the deletion of the subject-pronoun. The zero pronoun, hence, can be used uniquely to refer to the subject of a sentence, whereas the realized pronoun is generally introduced to inform the addressee that a topic shift has occurred. Thus, when the addressee finds a zero pronoun, s/he has to retrieve the sentence topic, which generally coincides with the subject of the sentence, and to identify it as the appropriate antecedent. We can formalize this procedure as indicated in (9): (9) Referent Assignment to the zero pronoun Select the topic of the sentence as the correct antecedent of a zero pronoun. Applying this rule in the resolution of the zero pronoun in (10), we obtain that its appropriate antecedent is the sentence topic, i.e. Daisy Duck, whereas it is not possible to select the non-topic element, i.e. Minnie, as coreferential with the zero form: (10) Paperina ha ballato con Minnie e poi Ø ha preparato la cena ‘Daisy Duck danced with Minnie and then Ø prepared dinner’ 9 However, this rule is of no help in presence of a realized pronoun whose resolution requires a finer and more complex reasoning. The full pronoun can be seen, in fact, as shown above, as a generic variable which could be used both to refer to the topic and to the non-topic expression of a sentence. For this reason, while interpreting sentences as the one in (11), the addressee has to cope with an ambiguous sentence. (11) Minnie ha ballato con Paperina e poi lei ha preparato la cena “Minnie danced with Daisy Duck and then she prepared dinner” To disambiguate the full pronoun she, the addressee has to perform a more subtle reasoning, considering that if the topic had been the appropriate referent, the speaker would have used a zero pronoun, and concluding therefore that the topic cannot be the correct antecedent. We can think of this operation in terms of scalar implicatures1: we assume that the use of the phonetically realized pronoun to refer to a highly accessible antecedent, such as a subject, is excluded by conversational implicature. If the speaker, in fact, wanted to convey reference to a highly accessible antecedent, s/he would have used a zero pronoun, which selects the topic (i.e. the subject) as its appropriate antecedent. This concept can be formalized as in (12): (12) Implicature for realized pronouns resolution If the topic had been the appropriate antecedent, the speaker would have used the zero pronoun to refer to it. Hence, select the non-topic expression as the correct antecedent. While interpreting the full pronoun in (11), therefore, the addressee must be aware of its ambiguity, considering that if the speaker had wished to refer to the sentence-topic, i.e. Minnie, he would have used the zero pronoun. S/he is then forced to conclude that a topic-shift has occurred and that, for this reason, Daisy Duck is the appropriate referent to be chosen. To see how the actual resolution of zeros and pronouns works, let us consider the sentences in (13) and (14): (13) Paperina ha ballato con Minnie e poi Ø ha preparato la cena 10 “Daisy Duck danced with Minnie and then Ø prepared dinner” (14) Paperina ha ballato con Minnie e poi lei ha preparato la cena “Daisy Duck danced with Minnie and then she prepared dinner” In (13) The zero pronoun resolution requires that we: (i) identify the referential expression (i.e. the zero pronoun), (ii) apply the “Referent Assignment to the Zero Pronoun” rule, (iii) find the possible antecedents (i.e. Daisy Duck and Minnie), (iv) identify the topic (i.e. Daisy Duck) and (v) select the topic as the correct antecedent. In (14), instead, the phonetically realized pronoun resolution requires us to: (i) identify the referential expression (i.e. the realized pronoun “she”), (ii) activate the implicature, (iii) find the possible antecedents (i.e. Daisy Duck and Minnie), (iv) apply the “Referent Assignment to the Zero Pronoun” rule, (v) identify the topic of the sentence (i.e. Daisy Duck), and finally (vi) select the non-topic expression as the appropriate antecedent of the realized pronoun (i.e. Minnie). To summarize, the resolution of referential expressions, and in particular of zero pronouns and phonetically realized pronoun, as stated above, seems to require a complex processing, imposing a considerable processing load on subjects’ Working Memory (see section 3). This consideration leads to the prediction that subjects who suffer from working memory impairments should have more difficulties in assigning the correct reference to pronominal expressions, i.e. there is a correlation between WM and interpretation of referential expressions. Moreover, according to the formulation above, it is possible to hypothesize that the resolution of phonetically realized pronouns requires a more complex processing than the resolution of zero pronouns, since it involves the activation of an implicature. As Reinhart (2001) points out, in fact, sentences triggering scalar implicatures are more complex, since they require subjects to keep in mind and compare two representations, and thus they are generally processed slower, even by adults. These two hypotheses will be tested respectively in EXP 1 and EXP 2 (see sections 4.1 and 4.2). 11 3. Verbal Working Memory and its relationship with Developmental Dyslexia In this section, I will briefly discuss working memory, introducing one of the most well-known models about its structure and functioning. Data showing the existence of a strict relationship between working memory and developmental dyslexia, and also between working memory and age will be further presented. I will then argue that dyslexia can be seen as a deficit arising from Working Memory (WM), and in particular from Verbal Working Memory (vWM), dysfunctions. 3.1. What is Working Memory? Working Memory can be defined as a “brain system that provides temporary storage and manipulation of the information necessary for such cognitive tasks as language comprehension, learning and reasoning” (Baddeley 1986). The conceptualization of a working memory system led to the abandonment of the idea of a single unitary short-term store which also functions as a working memory. Some of the most convincing evidence in favour of a distinction between short-term memory and working memory came in the last decades from studies of brain-damaged patients who were no longer able to have new lasting memories, but showed a normal performance in short-term memory. The first unitary model of WM proposed by Baddeley and Hitch (1974) was a three-component system, composed of a limited capacity attentional controller, the Central Executive, and two slave subsystems, one performing operations with acoustic and verbal information, dubbed Articulatory (subsequent Phonological) Loop, and the other concerned with visual and spatial information, the Visuospatial Scratchpad (subsequent Sketchpad). Due to the need to integrate information from the subsidiary systems and from long term memory allowing at the same time their manipulation and maintenance, a fourth component has been added to Baddeley’s model, the episodic buffer (Baddeley 2000) which has the task of linking information across domains, forming integrated units of visual, spatial and verbal material. 12 This multiple component model of WM has been highly influential and supported by a huge amount of experimental data, and it has thus provided a “useful conceptual tool in understanding a range of neuropsychological deficits, which in turn have thrown light on normal cognitive functioning” (Baddeley 1992). In the next sections, the phonological loop will be analyzed in more detail, shedding light on its properties and also on its limitations. 3.1.1. Phonological Loop and Verbal Working Memory: strengths and weaknesses The phonological loop is the most broadly investigated component of working memory; it is assumed to comprise two sub-systems: a phonological store, deputed to temporarily store acoustic or speech-based information, and an articulatory control process, which permits to maintain material in the store by subvocal repetition and also to register visually presented information, such as words or pictures, by subvocalization in the phonological store. The model gives a good account of a rich range of laboratory-based findings, such as: (i) The acoustic similarity effect: data show that immediate ordered recall of items is poorer when they are phonologically similar, in comparison to when they are different in sound (Conrad 1964; Baddeley 1966). This happens, arguably, because the basic code involved in the store is phonological: similar items have fewer distinguishing cues and they are so more susceptible to be forgotten. This effect does not appear in presence of similarity of meaning, suggesting that this subsystem does not support a semantic code. (Zempleni 2006). (ii) The irrelevant speech effect: data demonstrate that performance on serial recall tasks, requiring to recall a sequence of visually presented items, is disturbed by the presence of irrelevant background speech (Colle & Welsh 1976). This phenomenon implies that “disruptive spoken material gains obligatory access to the phonological memory store” (Baddeley, 1992). (iii) The word-length effect: it has been shown that memory span for words is inversely related to their spoken duration (Baddeley, 1975; Ellis & Hennelley 1980). This provides evidence for the existence of a subvocal rehearsal process that refreshes the phonological material in the store. 13 (iv) Articulatory suppression: it has been demonstrated that it is possible to disrupt the subvocal rehearsal process, asking the subject to remember a list of items while uttering an irrelevant sound, such as the word “the” (Baddeley et al. 1984). Articulatory suppression prevents the subjects from rehearsing the material they are trying to memorize, showing that the phonological store and the articulatory control process are separate. Although further experiments have provided data that further strengthened the model, there is one important consideration to be drawn. In Baddeley’s multimodal WM model a strict connection links the phonological loop to language: such relationship has been judged as inadequate and inappropriate by many researchers, since it does not seem to take into account the rich complexity of language, not considering morpho-syntax, semantics and pragmatics. Recently, Baddeley himself (2001) has shown to be aware of this problem, arguing that “it is clear that the phonological loop model is capable of more detailed computational specification and that this represents an important route for future development of the model”. For this reason, I believe that it would be more appropriate to speak, more generally, of a Verbal Working Memory (vWM), that should comprise different subcomponents, dedicated not only to phonology, but also to morpho-syntax, semantics and pragmatics. The idea to identify a vWM, deputed to the storage and processing of linguistic information, is not new: researchers working in the psychological field (Smith and Geva 2000) as well as in the neurolinguistic domain (Martin et al. 1999; Hanten and Martin 2000; Grodzinsky 2005; Zempleni 2006), have assumed that the vWM’s scope is arguably wider than just phonology and showed that it is involved in particular in semantic processing. In the following sections, a rich body of evidence showing the existence of a connection between vWM and language as a whole will be briefly presented. 3.1.1.2. Verbal Working Memory and Semantics A first claim in favour of the individuation of different linguistic subcomponents in WM comes from Almor et al. (1999) that hypothesizes the existence of a connection between vWM and semantics, showing that Alzheimer patients’ pronoun processing, in both production and comprehension, reflects the working 14 memory impairment associated with their disease. Analysing Alzheimer patients’ spontaneous speech and their performance in processing pronominal expressions, Almor et al. discovered, namely, that they had remarkable difficulties in comparison to age-matched healthy controls, i.e. they produced more empty speech2, and their comprehension resulted to be highly impaired in assigning the appropriate referent to pronouns. Both groups of participants underwent a battery of working memory tests, to assess whether there was a correlation between WM performance and interpretation/production of pronominal expressions. Results confirmed that WM performance is associated with referential deficits and led Almor et al. to argue that “vWM performance is related to pronoun processing in production and comprehension”. They also suggested that pragmatics, too, could be linked to vWM, following other authors’ proposal (Ripich and Terrel 1988; Ulatowska and Chapman 1995; Ulatowska et al. 1999) to consider pronoun overuse, typical in AD patients’ speech, as a consequence of failing to follow discourse conventions. According to their work, AD patients’ empty speech could be the consequence of a pragmatic deficiency in discourse structuring ability. Hanten and Martin (2000) and Martin et al. (1999), moreover, argue, on the basis of neuropsychological evidence, for the existence of parallel phonological and semantic short-term memory systems. 3.1.1.3. Verbal Working Memory and Morpho-Syntax Service and Tujulin (2001) conducted a research to study the effect of morphological complexity on vWM in dyslexic children, age-matched control children and control adults, to investigate in which way morphologically complex words tax vWM. Subjects had to perform two WM tasks: a word span tasks, consisting in the immediate recall of auditorily presented word lists, and a last word span task, requiring to give a truth value judgment to a sentence while remembering at the same time the last word of each sentence. Target words were divided into three groups: base words, inflected words and derived words. Data showed that base words were remembered more easily and more frequently by all subjects, in comparison to inflected and derived words, which were morphologically more complex. This result was taken as a confirmation of a correlation between vWM and morphology. Another study, from Badecker and Kuminiak (2007), established that a “content-addressable WM is needed to calculate subject-verb agreement”. 15 Caplan and Waters (1999) focus on a finer division, arguing that vWM is composed of subsystems dedicated to different types of verbal tasks; they analyse in particular syntactic processing, suggesting that “the WM system that is called on in interpretive processing at the sentence level – assigning the syntactic structure of a sentence and using that structure to determining the meaning of the sentence – constitutes a separate subsystem within vWM”. With three different experiments they investigated (i) the relationship between individual differences in WM and individual efficiency in syntactic processing, (ii) the influence of concurrent memory load on syntactic processing and (iii) how patient with poor WM cope with syntactic processing. Results led them to conclude that there is a specialization in the vWM deputed to the assignment of the syntactic structure to a sentence. 3.1.1.4. Verbal Working Memory and Language Acquisition VWM has been demonstrated to be a strong predictor of language acquisition’s proficiency (Ellis & Sinclair 1996). Different studies have emphasized the correlation between WM and vocabulary: Gathercole & Baddeley (1989) provide evidence for this association, showing that four years old children’s ability to repeat a sequence composed of unfamiliar new words predicted the size of their native language vocabulary one year later. Moreover, Gathercole & Baddeley (1990) argue that children with poor non-word repetition skills were slower at learning new vocabulary in comparison to more skilled children. Phonological memory, then, seems to be strongly involved in the long-term learning of unfamiliar phonological material: for this reason, individuals with a lower span, also display more difficulties in learning foreign words. 3.2. Working Memory and Age Different studies (Gathercole et. al 2004, Baddeley 1986, Hulme et al. 1984; 1989; Gathercole and Hitch 1993; Gathercole and Baddeley 1993; Gathercole and Adams 1993) have demonstrated that there is a strong relationship between WM and age, showing that WM capacity improves steadily until the teenage years, when it starts levelling off. 16 Gathercole et al. (2004) argue that the four main components of Baddeley’s model of WM are in place by 6 years of age, but that their capacity increases linearly from age 4 to early adolescence. This improvement can be due to the learning of different and more effective strategies to carry out tasks. Experimental investigations of WM development have shown that around 7 years of age, memory performance changes both qualitatively and quantitatively (Gathercole 1999, Gathercole and Hitch 1993). Under 7 y.o., in fact, children are unable to rehearse and they rely basically on visual short-term memory to remember information. Their performance approaches adult level around 15 years. Moreover, consistent findings have demonstrated that performance on all short-term memory tests enhances during the childhood years, owing to an increased efficiency of processing which releases more space for memory storage (Case, Kurland & Goldberg 1982). Fry and Hale (2000), Towse, Hitch and Hutton (1998) and Kail (1992), instead, claim that WM development is due to a general enhancement in the speed of cognitive processing which advantages all components of WM. 3.3. Working Memory and Dyslexia The belief that dyslexia is due to a working memory impairment has already been supported by a number of studies: inefficient working memory, in fact, has been implicated as an underlying factor in dyslexia. Mortimore (2003), for instance, argues that memory processing involves four main components, namely (i) audio memory (including phonology), (ii) visual memory (including orthography), (iii) procedural memory (including the learning of procedures, as riding a bike) and (iv) semantic memory (which deals with meaning), assuming that a person with dyslexia can experience weaknesses in any of these components, and that this will put pressure on the others. McLoughlin et al. (2002) have proposed a definition of dyslexia based on working memory, saying that “Developmental dyslexia is a genetically inherited and neurologically determined inefficiency in working memory, the information-processing system fundamental to learning and performance in conventional educational and work settings. It has a particular impact on verbal and written communication as well as on organization, planning and adaptation to change”. 17 Studies conducted until now, have shown that dyslexic people experience particular deficits in the phonological loop: Pickering (2000) states that their “phonological code in working memory” is disrupted, leading to problems in translating visual information into phonological form. She concludes assuming that this impairment affects their ability to learn new words while reading. Similarly, language disordered children, as dyslexics, were also found to be particularly poor at repeating single non-words and recalling word lists, even in comparison to younger children younger: according to Gathercole and Baddeley (1990), their repetition skills were delayed by about four years. These results were confirmed by another, more recent, study conducted by Jeffries and Everatt (2004) who compared dyslexics and age-matched typically developing children on tasks assessing phonological processing, visuo-spatial and motor coordination and both executive and inhibitory functioning. They found that dyslexics performed worse than controls on all working memory phonological loop measures, whereas they behaved as well as controls on working memory visuo-spatial sketchpad measures and visual-motor coordination tasks. Hence, they conclude arguing that dyslexic children’s working memory is damaged, while their visuo-spatial memory is unimpaired. This finding is also supported by a study conducted by Bowyer et al. (2004), who demonstrated, resorting to the MEG1 technique to determine the neuroanatomical location of WM processes, that dyslexic subjects experience difficulties in verbal WM tasks, but not in spatial WM tasks. In particular, dyslexics activated the same areas as controls during the spatial tasks (right superior temporal gyrus and right angular gyrus), whereas they activated different areas during verbal tasks: controls activated the left superior temporal gyrus and the left angular gyrus, while dyslexics activated the same areas but in the right hemisphere. Boywer concludes thus arguing that there are differences in how memory is processed in subjects with dyslexia compared to normal reading subjects, suggesting that the core deficit manifested in dyslexia lies in verbal, but not in nonverbal WM. Evidence for an association between working memory skills and both reading difficulties and poor mathematics abilities is provided by Gathercole et al. (2005) who argue that working memory skills represent an important constraint on the acquisition of skills and knowledge in reading and mathematics, and that they are related to the severity of the learning difficulties manifested by dyslexics. According to other researchers, in particular, WM provides resources that allow the learner to integrate information retrieved 18 from long-term memory with current inputs; in this framework, then, poor WM skills lead to impairments in complex cognitive activities (Swanson & Beebe-Frankenberger 2004; Swanson and Saez 2003). Gathercole et al. (2007), finally, found that WM problems and the inattentive behaviour often detected in dyslexic children, co-occur: they noticed, namely, that children with a low WM span “obtained high ratings of cognitive problems and inattentive symptoms and were judged to have short attention spans, high levels of distractibility, problems in monitoring the quality of their work and difficulties in generating new solutions to problems”. These results lead the authors to argue that reduced working memory capacities may play a causal role as well in the problematic behaviour of dyslexic children. To sum up, then, a rich body of evidence demonstrates that dyslexics display working memory impairments, affecting in particular the verbal domain and the control of attention. 4. Experiments In this section, the results of two different experimental protocols will be presented. Experiment 1 was designed to test the existence of a correlation between the interpretation of sentences involving pronouns resolution and verbal Working Memory. Experiment 2, instead, aimed to check whether linguistically impaired children as dyslexics perform differently in comparison to control children, matched on their age and non-verbal IQ, in tasks requiring comprehension of sentences containing anaphoric elements. Since a control group composed of adults was also included in the protocol, specific attention was also devoted to the presence or absence of differences amongst adult’s performance and children’s performance. 4.1. Experiment 1 4.1.1. Method 4.1.1.1. Participants The experimental task was conducted on 43 typically developing children, divided in two groups according to their age: the first group was composed of 22 children attending to the third and fourth year of 19 the primary school (mean age 9;1), whereas the second one was composed of older children attending to the fifth year of the primary school (mean age 10;8). 4.1.1.2. Design and procedure The experimental protocol was designed to assess the existence of a relationship between vWM and the interpretation of referential expressions. For this reason, subjects had to perform two kind of tasks: a linguistic task, assessing their semantic competence, and a working memory task, assessing their verbal WM abilities. For the first part of the experiment a picture selection task was used: the subject was presented some pictures portraying a short story: the main characters of the story, introduced in the first picture, were four boys or girls, differentiated for one single visual feature (i.e. the colour of the dress) or for two visual features (i.e. the colour of the dress and the colour of the hair). In the second picture, one of the four characters was portrayed as performing an action. Subsequently, the subject was told that the characters were hiding behind a curtain and that nobody could see what happened later, except for a puppet, namely Little Red Riding Hood, who told what had occurred. The subject’s task was to select amongst two pictures, the one which correctly described what happened, according to what the puppet said. The protocol involved a warm up item, two fillers and eight experimental items, two for each of the four different conditions, namely: (i) Condition A, Z1f: a zero pronoun was used, when possible antecedents differed for only one visual feature; (ii) Condition B, Z2f: a zero pronoun was used, when possible antecedents differed for two visual features; (iii) Condition C, R1f: a phonetically realized pronoun was used, when possible antecedents differed for only one feature; (iv) Condition B, Z2f: a phonetically realized pronoun was used, when possible antecedents differed for two visual features; An example of Experiment 1 follows: 20 A B In the first picture, the experimenter introduces the characters, namely four blond boys wearing clothes of different colours (i.e. characters differ for one visual feature). The subject is then shown that in the second picture the red boy greets the green boy; the characters suddenly hide behind a curtain and the participant is told that only Little Red Riding Hood knows what happened later. So, the experimenter asks her to describe what happened. Here are reported the sentences uttered by the puppet for each condition: (i) Condition A, Z 1f: Prima il bambino rosso ha salutato il bambino verde e poi ha salutato il bambino blu ‘First the red boy greeted the green boy and then Ø greeted the blue boy’ (ii) Condition B, R1f: Prima il bambino rosso ha salutato il bambino verde e poi lui ha salutato il bambino blu ‘First the red boy greeted the green boy and then he greeted the blue boy’ The subject’s task was precisely to select the picture which best corresponded to what the puppet told. With regard to the example reported above, note that in Condition A, only picture B portrays correctly what happened, while in Condition B picture A is the correct one. For the second part of the experimental protocol, a digit span test was used: subjects were shown a sequence of numbers of increasing length and their task was to remember the exact sequence, typing the 21 digits in the correct order. The number of digits correctly recalled was taken as the subject’s digit span test score. Another task assessing vWM, the reading span test, was performed on the group composed by older children: in this task, subjects had to read aloud a set of sentences, composed of two to six sentences, and they had to remember the last word of each sentence. The number of words correctly remembered was considered the subject’s reading span test score. 4.1.2. Results All subjects included in the two groups gave the correct answer to all fillers, hence nobody was excluded from the sample. The error rates shown by the groups of younger children and older children are respectively 42,45% and 26,32%. The t-test carried out on these data confirmed the existence of a statistically significant difference between the younger children’s performance and the older children’s performance (t = 2,389, df= 39, p = 0.021). This difference is consistent with the claim that vWM and age are associated: as their vWM capacities increase, children’s performance grows better. To determine if there is a correlation between the interpretation of referential expressions and vWM, the performance of the whole group of children was considered, taking into account their overall error rate (35,76%) and their mean digit span test score (5,93). For a first analysis of data, the whole group was divided in two subgroups according to their digit span test score: (i) Group 1 was composed of subjects whose digit span test score was less than 6. Their mean error rate was 44,38%. (ii) Group 2 was composed of subjects whose digit span test score was more than 6. Their mean error rate was 28,26%. A t-test comparing Group 1 and Group 2 shows that there is a statistically significant difference between them: (t= 2,5902, df= 41, p= .0132). 22 A Pearson Correlation test performed on the whole group demonstrates that there is a positive and strong correlation between the performance scored in the linguistic task and in the digit span test (r = .501, p (onetailed) = .000). The first analysis, hence, confirms that there is a strong correlation between vWM and interpretation of referential expressions. Significantly, in fact, subjects with a lower span display more difficulties than subjects with a higher span. A second analysis was then conducted on the performance shown by the older children, who were divided in two subgroups according to their reading span test score: (i) Group A was composed of children whose reading span test score was less than 2,5. Their mean error rate in the picture selection task was 45,00%. (ii) Group B was composed of children whose reading span test score was more than 2,5. Their mean error rate in the picture selection task was 18,75%. A t-test confirmed that there is a statistically significant difference between Group A and Group B (t= 3,0594, df= 20, p= .0062). A Pearson correlation task conducted on the whole group revealed that there is a strong relationship between the reading span test score and the number of errors reported in the linguistic task requiring the comprehension of pronominal expressions (r = .615, p (one-tailed) = .001). Summarizing, then, both analyses lead to the conclusion that vWM and interpretation of pronouns are strictly correlated. 4.1.3. Discussion The data collected in Experiment 1 show that there is a significant relationship between the resolution of referential expression, and especially of pronominal expressions (i.e. zero pronouns and phonetically realized pronouns), and the performance in tasks assessing verbal Working Memory. These results, which are consistent with the initial predictions, demonstrate that subjects with a low working memory span are significantly more impaired in the interpretation of sentences containing pronominal expressions than subjects with a high working memory span. 23 Data are also consistent with the hypothesis linking vWM with age (see paragraph 3.2.), since the statistical analysis shows that younger children are more impaired in the picture selection task in comparison to older children, providing evidence for the prediction that working memory’s development is not fully completed until adolescence and that it increases linearly as children grow up. 4.2. Experiment 2 4.2.1. Method 4.2.1.2. Participants The experimental task was performed on 96 subjects, divided in five distinct groups: 18 dyslexic children (mean age 9;15), 17 age-matched typically developing children (mean age 9;05), 18 control adults (30;2), and two groups of younger children, the former composed of 23 children attending to the first class of the primary school (mean age 6;8) and the latter composed of 20 younger children (mean age 4;75). Dyslexic children have been chosen from those who had independently received a diagnosis of dyslexia, specifically by the “Centro Audiofonetico” in Trento: in particular, dyslexic children were selected according to different factors: (i) absence of neurological diseases or genetic pathologies, (ii) absence of sensorial diseases, (iii) absence of psychopathological diseases, (iv) IQ > 80 (WISC – R) and (v) fluent and correct reading and writing abilities under 2 SD (Tressoldi et al. Battery, Prove MT). 4.2.1.3. Materials and Procedure Experiment 2 was designed to assess the subjects’ comprehension of referential expressions, using the Truth Value Judgement Task. The subject was shown some pictures on a computer screen portraying a short story that always involved two characters performing some actions. The experimenter introduced the subject with a puppet, the Clumsy Detective, who had the task to explain what happened in the short story. The subject was told that the inspector was named “Clumsy”, since he could not always describe correctly what happened in the story; so the participant’s task was to decide whether the Clumsy Detective said the truth about what happened in the story or whether he lied. The task involved ten experimental items intertwined with six fillers; there were four different conditions: 24 (i) Condition A, Zero True: the Clumsy Detective described the story using a zero pronoun; if the zero pronoun was interpreted correctly as referring to the subject of the sentence, the utterance was true; (ii) Condition B, Zero False: the Clumsy Detective described the story using a zero pronoun; if the zero pronoun was interpreted correctly as referring to the subject of the sentence, the utterance was false; (iii) Condition C, Realized True: the Clumsy Detective described the story using a phonetically realized pronoun; if the pronoun was interpreted correctly as not referring to the subject of the sentence, but to the object, the utterance was true. (iv) Condition D, Realized False: the Clumsy Detective described the story using a phonetically realized pronoun; if the pronoun was interpreted correctly as not referring to the subject of the sentence, but to the object, the utterance was false. An example of Task 1, with all four conditions, is reported below: (15) An example of Task 1 ! 25 In the first picture, the experimenter introduces the characters: Daisy Duck and Minnie; s/he then explains to the subject that first Daisy Duck and Minnie danced together and that after dancing, Daisy Duck went to the kitchen and prepared dinner. Suddenly, the Clumsy Detective appears and tries to describe what happened. I reported below the target sentences he uttered in the four conditions: (i) Condition A (Zero T): Paperina ha ballato con Minnie e poi Ø ha preparato la cena. ‘Daisy Duck danced with Minnie and then Ø prepared dinner’ (ii) Condition B (Zero F): Minnie ha ballato con Paperina e poi Ø ha preparato la cena. ‘Minnie danced with Daisy Duck and then Ø prepared dinner’ (iii) Condition C (Real T): Minnie ha ballato con Paperina e poi lei ha preparato la cena. ‘Minnie danced with Daisy Duck and then she prepared dinner’ (iv) Condition D (Real F): Paperina ha ballato con Minnie e poi lei ha preparato la cena. ‘Daisy Duck danced with Minnie and then she prepared dinner’ The subject’s task was precisely to decide if the Clumsy Detective told the truth or if he said a wrong thing. 4.2.1.4. Research Questions and Predictions Experiment 2 was designed to provide an answer to the following questions: (i) How do dyslexic children cope with tasks requiring a complex processing? (ii) Do dyslexic children perform differently from age-matched typically developing children in tasks requiring complex processing? (iii) Do dyslexic children exhibit a non-adult performance in assigning the appropriate antecedent to a referential expression? 26 (iv) Are there differences in the interpretation of sentences containing referential expressions between the five groups? (v) Does performance increase with age? (vi) Do dyslexics have working memory limitations? According to the literature about dyslexia and Working Memory and taking into consideration the results obtained in Experiment 1, the following predictions can be drawn: (i) Since error rate correlates with vWM (see Exp 1, section 4.1), a better performance is expected in adults and age-matched control children; (ii) Since vWM correlates with age, performance is expected to increase with age; (iii) Since at an age of 4-5 years old components of WM are not working properly yet, younger children are expected to be unable to cope with the task; (iv) If dyslexic’s vWM is impaired, a higher error rate is expected in comparison to age-matched control children. According to the existing literature about verification tasks, another important prediction can be drawn: experimental findings seem to show that true affirmative sentences are verified faster than false affirmative sentences. Clark and Chase (1972) and Just and Carpenter (1975), for instance, found that there is an interaction between the kind of sentences (i.e. affirmative and negative sentences) and the type of responses (i.e. true or false), evidencing that true affirmatives are verified faster than false affirmatives. Gough (1965) suggests that it is easier to compare the representation of a sentence to that of a picture if they are congruent. Since in Condition B and Condition D the sentences are false, i.e. not congruent with the pictures, we can hypothesize that the processing costs required by the computation impose an additional burden on vWM. Consequently, a higher error rate is expected in subjects who have a poor vWM. This consideration, leads us to the following prediction: (v) Since false sentences are more difficult to verify than true sentences, a greater error rate is expected in Conditions B and D. 27 4.2.2. Results All subjects included in the four groups gave the correct answer to the vast majority of the fillers; only one of the younger children failed to give the correct answer to all fillers and therefore he was excluded from the sample. The error rates displayed by the five groups are respectively: 49.44% for Dyslexic Children (DC), 10.00% for Age-Matched Control Children (AMCC), 2.22% for Control Adults (CA), 35.65 for First Class Children (FCC) and 50.56% for Younger Children (YC). Figure 1 displays the error rates reported in Task 1 for the four groups of participants. Figure 1: The error rates displayed by the five groups in Experiment 2 A statistical analysis was conducted on these data, to verify if there were statistically significant differences between the performances of the five groups of participant. Table 1 shows the mean scores, standard deviations standard error, lower and upper bounds, minimum and maximum for each group, concerning their overall performance in Experiment 2. EXPERIMENT 2 28 95% Confidence Interval for Mean N Mean Std. Deviation Std. Error Lower Bound Upper Bound Minimum Maximum DC 18 4,94 1,259 ,297 4,32 5,57 3 7 AMC 17 1,00 1,118 ,271 ,43 1,57 0 4 CA 18 ,22 ,548 ,129 -,05 ,49 0 2 FCC 23 3,57 1,308 ,273 3,00 4,13 1 5 YC 19 5,11 1,329 ,305 4,46 5,75 2 6 Total 95 3,04 2,278 ,234 2,58 3,51 0 7 A one-way analysis of variance was carried out on the data, and a α-level of 0.05 was adopted. The ‘group’ factor was considered to be a fixed factor; ‘group’ was significant: F (4, 90) = 67.841, p=0.000. Levene’s test for homogeneity of variances resulted significant: p = .004. Therefore, a subsequent post hoc comparison test for equal variance not assumed (Dunnett’s T3), with an α-level of 0.05 showed that dyslexic children’s performance differs significantly from control age-matched children and control adults’ one (p= .000) and even from first-class children’s one (p= .015). On the other side, dyslexic children’s performance does not differ significantly from younger children’s one (p=1.000). AMCC’s and CA’s performances do not differ significantly (p=.142), while AMCC’s and FCC’s performances are statistically different (p=.000). Moreover, FCC’s performance differs significantly not only from DC and AMCC, but also from CA (p=.000) and from YC (p=.006). To conclude, YC’s performed significantly differently from all other groups except from DC (p=1.000). The performance of the participants in the four conditions of Experiment 1 was then taken in consideration. 4.2.2.1. Condition A: Zero True With respect to Condition A (‘zero T’), the mean error rate obtained in the four groups in the two experimental items (with standard deviations in brackets) was 2.78% for dyslexic children (.236), while it was absent for age-matched control children, control adults, first-class children and younger children (.000). 29 A one-way analysis of variance was carried out on the data, with a α-level of 0.05; the ‘zeroT’ factor was not significant F (4, 90) = 1.073, p= .375. 4.2.2.1. Condition B: Zero False Concerning Condition B (‘zero F’), instead, the mean error rates scored displayed by the four groups in the three items, with standard deviations in brackets, were: 51.85% for DC (.922), 1.96% for AMCC (.243), 0% for CA (.000), 34.87% for FCC (.825) and 68.52% for YC (1.100). A one-way analysis of variance was carried out on the data (α-level= 0.05), showing that the ‘zero F’ factor was significant: F (4, 90) = 26.684, p= .000. Levene’s test for homogeneity of variances resulted significant: p = .000. Therefore, a subsequent post hoc comparison test for equal variance not assumed (Dunnett’s T3), with a α-level of 0.05 showed that DC’s performance differs significantly from AMCC’s and CA’s one (p= .000). On the contrary, DC’s performance does not differ significantly from FCC’s (p=.504) and YC’s one (p= .650); AMCC’s and CA’s scores do not differ significantly, too (p=.968). Interestingly, there is instead a statistically significant difference between FCC and YC (p= .014). In this condition, all CA and AMCC (except for only one child who failed to judge correctly the sentences) interpreted correctly all experimental items. DC, instead, show a different pattern: only two children (11,11%) gave the correct answer, while 3 failed always, giving a “yes” answer (16,67%); the remaining 72,22% of the children responded correctly to one or two out of three items. Only two of the 19 YC always answered in the appropriate way to all three items (10,53%), while 10 children, the majority of them (52,63%) failed always giving a “yes” answer. 7 of them (36,84%) answered correctly to one or two out of three items. A greater number of FCC, 6 of them (26,09%) answered always correctly, while only one always failed (4,35%); the majority of them, 15 children (65,22%) responded correctly to one or two out of three items. It is possible to interpret the indecision shown by DC, FCC and YC as a sort of guessing strategy, hypothesizing that since they are unable to cope with the task, they try to guess, accepting some utterances and rejecting some others. 30 To sum up, while CA and AMCC do not display difficulties in rejecting the wrong utterances, and therefore in assigning the correct referent to the zero pronoun, the majority of DC, YC and FCC adopt a guessing strategy. 4.2.2.2. Condition C: Realized True Regarding Condition C (‘real T’), the mean error rates in the two items were: 36.11% for DC (SD= .752), 11.76% for AMCC (SD= .437), 0% for CA (SD= .000), 23.91% for FCC (SD=.665) and 19.44% for YC (SD= .597). A one-way analysis of variance was performed (α-level= 0.05) and it showed that the ‘real T” factor was significant: F (4, 90) = 4.117, p= .004. Levene’s test for homogeneity of variances was significant, with p= .000. So, Dunnett’s T3 post hoc comparison test was carried out, displaying that the only significant differences are between DC and CA (p = .007) and between FCC and CA (.022). In this condition, none of the CA displays difficulties, while amongst AMCC (21,05%) only 4 subjects exhibit an indecision, answering correctly to only one out of two items. The majority of the YC (68,42%) responded correctly, whereas 6 of them (31,58%) accepted only one out of two items. Amongst FCC, instead, 14 children always gave the correct answer (60,87%), whereas 9 of them (39,13%) accepted only one utterance. We can interpret YC’s and FCC’s performance arguing that they are biased to give a “yes” answer, as we will see in section (4.2.2.5). DC’s performance is more interesting: only 8 children answered correctly to all items, while 3 of them always failed; the remaining 7 children rejected one out of two correct sentences. Note that in this condition the correct answer is “yes” and that for this reason a “no” response is even more significant, since it means that the subject is selecting the topic as the correct antecedent of a phonetically realized pronoun, failing to activate the implicature (see section 2.2.2). Given the strong tendency to give “yes” answers noted in Condition B, we can argue that in this case the relatively high percentage of correct responses could hide the real difficulty met by the children. Furthermore, DC’s error rate (the highest one amongst the five groups) can be explained in accordance with the remarks above, arguing that they are not able to activate the implicature for phonetically realized pronoun resolution, allowing the overt pronoun to refer to the subject of the sentence. This consideration 31 gives stronger support to the hypothesis according to which the processing cost required by this task exceeds the children’s capacities. 4.2.2.3. Condition D: Realized False Finally, the last condition (‘real F’) was considered, showing that the mean error rates are the highest for all five groups: 87.04% for DC (SD= .608), 24.54% for AMCC (SD= .849), 5.56% for CA (SD= .548), 68,12% for FCC (SD= 1.107) and 88.89 for YC (SD= .582). A one-way analysis of variance showed that the ‘real F’ factor was highly significant: F (4, 90) = 37.045, p= .000. Levene’s test for homogeneity of variance resulted in p= .002, hence significant. Dunnett’s T3 post hoc comparison test was carried out, displaying that DC’s performance is different from AMC’s (p=.251) and CA’s (p= .193) ones, while it is not different from FCC’s (p= .342) and YC’s (p=1.000) ones. There is no statistically significant difference between FCC’s and YC’s behaviour (p=.188); AMCC and CA do not differ significantly, too (p=.412). Data show that this condition is the most difficult one for all five groups: CA answered correctly to all three items, except for 4 of them who showed an indecision in one or two items. The error rate increased also for AMCC: 1 child always failed to give the correct truth-value to the utterance (5,88%), 8 children answered correctly to all items (47,06%) and the remaining 8 children accepted one or two out of three items. The percentage of errors is higher also in FCC: only two children were always correct (8,70%), 12 children always failed (52,17%) and the remaining 9 (39,13%) accepted one or two sentences. DC and YC, instead, showed a similar pattern: amongst DC, nobody answered in the appropriate way to the three items, 12 of them gave always a “yes” answer, mistaking (66,67%), and 6 accepted one or two utterances (33,33%). Likewise, none of the YC responded correctly to all three items; 14 of them gave always a “yes” answer (73,68%) and 5 (26,32%) rejected one or two out of the three wrong statements. In this condition, the remarkably high error rate displayed by FCC and in particular by DC and YC can be interpreted as incapacity to cope with the sentences uttered by the Clumsy Detective, due to the high processing cost required to interpret them. 32 It is then possible to conclude that DC, YC and FCC resort to a bypassing default strategy (Reinhart 2006), preferring to give “yes” answers, since they are not able to compute the meaning of the sentences, due to their processing limitations. 4.2.2.4. Overall Performance Figure 2 displays the distributions of the five groups’ performance in the four conditions. Figure 2 Performance in the four conditions of EXP 2 The graph shows clearly that, whereas AMCC exhibit an adult-like behaviour, DC manifest a remarkably greater difficulty in giving the correct truth-value judgment to utterances which require a costly processing. Dyslexic children’s performance rather resembles the behaviour manifested by children more than 4 years younger than them, since a statistical analysis demonstrated that DC’s and YC’s performances do not differ significantly in any of the four conditions. Interestingly, FCC, more than two years younger than DC, occupy a midway position between DC and AMCC: their overall performance is significantly better than DC’s and YC’s one, even though not yet completely adult-like. 33 To sum up, the analysis of the results collected in Experiment 2 shows that there is a remarkably significant difference between DC’s interpretations of pronominal expressions in comparison to AMCC’s and CA’s ones, and in particular in conditions B, C and D. Interestingly, instead, there is no statistically significant difference between DC’s performance and YC’s one. These data permit us to conclude that DC behaved in Experiment 2 in the same way as children more than 4 years younger than them in all conditions. AMCC, instead, showed an adult-like behaviour, whereas FCC followed a different pattern: their overall performance is statistically significantly different from that of DC and YC, while they do not perform differently in Condition B, C and D. Figure 2 shows clearly that FCC’s trend in the four conditions resembles DC’s one, even though at a significantly lower level. 4.2.2.5 Tendency to give “yes” answers Another set of data to be taken in consideration in the analysis of the results obtained in Experiment 2 consists in checking if subjects manifested the tendency to give true answers to the target sentences uttered by the Clumsy Detective. First of all, it is important to make it clear that subjects who always said “true” have been excluded from the sample. However, DC and YC exhibited a strong tendency to provide “true” answers in conditions B and D, where AMCC and CA quite always give a “false” (and correct) answer. In Figures 3 and 4, two graphs display the percentages of “true” answers given by the five groups in all four conditions and in conditions B and D, respectively. Note that in Conditions A and C “true” answers are the correct ones, while in Conditions B and D, “false” answers are the only appropriate ones. 34 Figure 3 Percentage of true answers given by the subjects in the four conditions Figure 4 Percentage of true answers given by the subject in the “false” conditions 35 Similarly as the graphs depicted in Figure 2, Figure 3 and Figure 4 show that participants can be divided in two groups: on one side AMCC and CA exhibit a related tendency, accepting true utterances and rejecting false ones, while DC and YC can be grouped together, since both manifested a consistent preference to accept, answering “yes”, not only true statements, but also false ones. As noted above, instead, FCC show a dissimilar tendency, taking place between the two groups: their preference to say “yes” to false utterances is weaker than DC’s and YC’s one, even though stronger than AMCC and CA’s one. In other words, FCC do not show an adult-like behaviour, but their capacity is increasing; this trend seems to suggest that, differently from typically developing children, who are increasingly approaching an adult-like behaviour, dyslexic children remain still at the same level of children 4 years younger than them. Thus, we can interpret the tendency to give “yes” answers in Conditions B and D as incapacity to assign the correct referent to the zero pronoun and to the phonetically realized pronoun respectively. 4.2.3. Discussion Experiment 2 provided interesting data, consistent with all predictions (see section 4.2.1.4.). As expected, DC are remarkably more impaired in comparison to AMCC, while, interestingly, their poor performance does not differ from the one shown by YC, more than four years younger than them. Significantly, DC show a poorer performance also in comparison to FCC, two years younger than them. The greatest difference between DC’s and AMCC’s performances can be found in condition B, where AMCC demonstrate to master the resolution of zero pronouns in the vast majority of the cases (their error rate is only 1,96 %, in comparison to the 51,85% scored by DC), while DC perform at chance level, demonstrating that they are not able to cope with the task. As expected, Condition D is the most difficult one for all groups (even for adults who make errors only in this condition), since it requires a finer reasoning, involving the resolution of an implicature (see section 3.1.2) , which has been proven to be a complex task, imposing an extra-burden on WM (Reinhart 2006). In line with our predictions, Conditions B and D are respectively more difficult than Conditions A and C, consistently with the hypothesis claiming that false sentences are more complex and take longer to verify with respect to true utterances (Just and Carpenter 1975, Clark and Chase 1972, see section 4.2.1.4.): the 36 verification process, in fact, requires to maintain in WM two different representations and to compare them, in order to select the correct one. Since Reinhart (2006) demonstrates that keeping in mind and comparing two representations imposes a great load on WM, we can explain the high error rate detected in Condition D assuming that it is due to the processing costs required by the computation. Results are also consistent with the hypothesis arguing that there is a correlation between vWM and age, as well as between vWM and interpretation of referential expression. AMCC show already an adult-like performance, even though their error rate is higher in comparison to CA’s one: this trend can be explained noting that their WM development is not yet completed, since they are just 9 years old. This finding permits also to explain why there is a statistically significant difference between 5 years old and 9 years old children’s performance while coping with my experiment. YC, as expected, are unable to cope with the task: they manifest the tendency to give always “yes” answers, suggesting that the task is too difficult for them. As Reinhart (2006) points out, in fact, when WM resources are insufficient to carry out the required computation, subjects tend to use some bypassing strategies, as guessing (Cond. B) or selecting a default strategy (Cond. D). Note that this consideration permits to explain the chance level performance displayed by DC in condition B (51,85%; 68,52% in YC) and their striking error rate in Condition D (87,04%; 88,89% in YC). Summarizing, then, YC are unable to cope with the task and show a very poor performance, while FCC, two years older, improved their capacity and commit less errors in all four conditions; AMCC, finally, exhibit an adult-like performance, whereas CA do not display difficulties and provide, in the vast majority of the cases, correct answers. DC, instead, show a really different behavior: they are not only more impaired than age-matched typically developing children, but they show a poorer performance also in comparison to children two years younger than them, whereas their performance does not differ significantly from the very young children’s performance. Since it has been demonstrated in Experiment 1 that vWM and interpretation of pronominal expressions are correlated, data lead to conclude that dyslexics’ working memory is impaired, even though it is not yet 37 clear if their impairment arises from a general developmental delay, involving also WM development, or from a WM which is intrinsically impaired. 5. Conclusion In this paper evidence for the occurrence of a semantic and pragmatic deficit in developmental dyslexia has been reported. The experimental protocol designed to investigate dyslexic children’s semantic and pragmatic abilities has taken into account, in particular, their interpretation of referential expressions in the framework of the Accessibility Theory. Dyslexic children’s performance in the resolution of pronominal expressions (i.e. zero pronouns and phonetically realized pronouns) has been found remarkably poorer in comparison to that of age-matched typically developing children’s. Interestingly, dyslexics perform worse even in comparison to children more than two years younger than them, while their error rate does not differ significantly from the one shown by children more than four years younger than them. These results strongly confirm that developmental dyslexia has to be considered as an impairment affecting different levels of linguistic representation, beyond the phonological domain which has been more thoroughly investigated in the last decades. Since throughout the paper it has been demonstrated that there is a strict correlation between verbal Working Memory and the interpretation of pronominal expressions, the results obtained lead to another important hypothesis, namely, that dyslexic children’s working memory is impaired, in comparison to that of control children. This suggestion is consistent with a rich body of evidence showing the existence of a relationship between verbal working memory and developmental dyslexia and the results exposed in this paper go in this direction, supporting the insight that dyslexics exhibit working memory limitations or deficiencies, which lead to learning impairments. Further research is needed to investigate thoroughly dyslexic children’s working memory, comparing their performance to the one showed by control children of different ages, in order to determine which components of working memory are affected in dyslexia and to understand which role WM impairment has in dyslexia. Moreover, it would be interesting to clarify whether dyslexic children’s working memory is impaired, or whether their WM deficiencies are determined by a general developmental delay. 38 Acknowledgements A sincere thanks goes to all the children who took part in the experiments, to their parents, and to everyone who actively participated in the organization and administration of the experimental sessions. A special thanks goes to the “Centro Audiofonetico” in Trento, and in particular to Dr. Cristina Ioriatti, Dr. Emanuela Paris, and the speech therapists that collaborated with my experimental project, individuating the dyslexic children who took part to the experiments. I also wish to thank, for their kind and precious collaboration, the “Istituto Comprensivo di Fondo” in Romeno (TN) – in particular Dr. Massimo Gaburro and all the teachers, the “Istituto Comprensivo di Revò” in Banco and Revò (TN), Dr. Cinzia Salomone and all the teachers, the “Istituto Comprensivo di Quinto Veronese” in Quinto Veronese (VR) and, finally, the “Scuola Materna Paolo Crosara” in S. Bonifacio (VR), and in particular, Dr. Maria Mastella, Dr. Roberta Bacci, and all the teachers. An heartfelt thank you goes to Gaetano Fiorin, Karin Martin, Danilo Reggiani and Luisa Piccoli who helped during the experiments. Thanks to Denis Delfitto, Frank Wijnen, Chiara Melloni and Gaetano Fiorin for comments and discussions. Parts of this work were presented during talks in Patras, Rovereto, Utrecht, and Verona; I wish to thank as well the participants and organizers of these events. 39 References Alloway, T.P., S.E. Gathercole, C. Willis and A. M. Adams (2004) “A structural analysis of working memory and related cognitive skills in young children,” Journal of Experimental Child Psychology 87, 85– 106. Ariel, M. (1990) Accessing noun phrases antecedents. Routledge, London. Ariel, M. (1991) “The function of accessibility in a theory of grammar,” Journal of Pragmatics 16, 443-463. Ariel, M. (1994) “Interpreting anaphoric expressions: a cognitive versus a pragmatic approach,” Journal of linguistics 30, 3-42. Ariel, M. (2001) “Accessibility Theory: an overview,” in T. Sanders, J. Schlipperoord and W. Spooren, eds., Text representation, John Benjamins, Amsterdam, Philadelphia,. Baddeley, A. (1966) “Short-term memory for word sequences as a function of acoustic, semantic and formal similarity,” Quarterly Journal of Experimental Psychology 18 (4), 362-365. Baddeley, A. et al. (1984) “Exploring the articulatory loop,” Quarterly Journal of Experimental Psychology 36 (2), 233-252. Baddeley, A. D. (1986) Working memory. Oxford University Press, Oxford. Baddeley, A. D. (2003) “Working memory and language: an overview”, Journal of Communication Disorders 36, 189–208. Baddeley, A. D. and Hitch, G. I. (1974) “Working memory,” in G. Bower, ed., The psychology of learning and motivation, Academic Press, New York. Barde, L. H. and S. L. Thompson-Schill (2002). “Models of functional organization of the lateral prefrontal cortex in verbal working memory: Evidence in favor of the process model,” Journal of Cognitive Neuroscience 14 (7), 1054–1063. Bedny , M., J. C. Hulbert, S. L. Thompson-Schill, S. L. (2007) “Understanding words in context: The role of Broca's area in word comprehension,” Brain Research 1146, 101-114. 40 Blomert, L. (2003). “Stand van zaken dyslexie,” in Reij, R., ed., Dyslexie, naar een vergoedingsregeling, Rapport nr. 144 College voor Zorgverzekeringen. Bryant, P. (1995). “Phonological and grammatical skills in learning to read,” in B. de Gelder and J. Morais, eds., Speech and Reading. A Comparative Approach, Erlbaum Taylor & Francis, Hove. Carpenter, P. A. and M. A. Just (1975) “Sentence comprehension: a psycholinguistic model of sentence verification”, Psychological Review 82, 45-73. Carpenter, P. A., M. A. Just and E. D. Reichle (2000) “Working memory and executive function: Evidence from neuroimaging,” Current Opinion in Neurobiology 10, 195-199. Case, R. D., M. Kurland and J. Goldberg (1982). “Operational efficiency and the growth of short-term memory span,” Journal of experimental Child Psychology 33, 386-404. Chien, Y. and K. Wexler (1990) “Children’s knowledge of locality conditions in binding as evidence of the modularity of syntax and pragmatics”, Language Acquisition 1, 225–295. Chierchia, G. (1995) Dynamics of Meaning. Chicago University Press, Chicago. Chierchia G. (2004) “Scalar Implicatures, Polarity Phenomena and the Syntax/Pragmatics Interface,” in A. Belletti, ed, Structures and Beyond, Oxford University Press, Oxford. Chomsky, N. (1981). Lectures On Government and Binding. Foris, Dordrecht. Clark, H. H. and W. G. Chase (1972) “On the process of comparing sentences against pictures”, Cognitive Psychology 3, 472-517. Colle, H. A. and A. Welsh, A. (1976) “Acoustic masking in primary memory,” Journal of Verbal Learning and Verbal Behaviour 15 (1), 17-33. Conrad, R. (1964) “Acoustic confusion in immediate memory,” British Journal of Psychology 55 (1), 7584. Crain, S. and R. Thornton, (1998) Investigations in Universal Grammar: A Guide to Experiments on the Acquisition of Syntax and Semantics. MIT Press, Cambridge, MA. De Bree, E. (2007). Dyslexia and Phonology: A Study of the Phonological Abilities of Dutch Children Atrisk of Dyslexia. Doctoral Dissertation, University of Utrecht, LOT Dissertation Series, Utrecht. 41 Desroches, A. S., M. F. Joanisse and E. K. Robertson (2006) “Specific phonological impairments in dyslexia revealed by eyetracking,” Cognition 100, 32–42. Elbro, C. (1996). “Early linguistic abilities and reading development: A review and a hypothesis about distinctness of phonological representations,” Reading and Writing: An Interdisciplinary Journal 8, 36–60. Ellis, N. C. and R. A. Hennelley (1980) “A bilingual word-lenght effect,” British Journal of Psychology 71, 43-52. Evans, G. (1980). Pronouns. Linguistic Inquiry 11(2), 337–362. Fry, A. F., and S. Hale (2000). “Relationship among processing speed, working memory and fluid intelligence in children,” Biological Psychology 54, 1-34. Gathercole, S. E. and A. M. Adams (1993). “Phonological working memory in very young children,” Developmental Psychology 29(4), 770-778. Gathercole, S. E. and A. D. Baddeley (1990). “Phonological Memory Deficits in Language Disordered Children: Is There a Causal Connection?,” Journal of Memory and Language 29, 336-360. Gathercole, S. and Baddeley, A. D. (1993). Working Memory and Language, Lawrence Erlbaum Associates, Inc., Hillsdale, New Jersey. Gathercole, S. E. and T. P. Alloway (2006) “Practitioner Review: Short-term and working memory impairments in neurodevelopmental disorders: diagnosis and remedial support,” Journal of Child Psychology and Psychiatry 47(1), 4–15. Gathercole, S. E., T. P. Alloway, C. Willis, A. M. Adams (2006) “Working memory in children with reading disabilities,” Journal of Experimental Child Psychology 93(3), 265-281. Gathercole, S. and G. Hitch (1993) “Developmental Changes in Short-Term Memory: A Revised Working Memory Perspective,” in A. F. Collins, S. E. Gathercole, M. A. Conway, and P. E. Morris, eds., Theories of Memory, Lawrence Erlbaum Associates, Inc., Hillsdale, New Jersey. Gazdar, G. (1979) Pragmatics: implicature, presupposition and logical form, Academic, New York. Grodzinsky, Y., (2005) “Syntactic dependencies as memorized sequences in the brain”, Canadian Journal of Linguistics. 42 Hanten, G. and R. Martin (2000) “Contributions of phonological and semantic short-term memory to sentence processing: Evidence from two cases of closed head injury in children,” Journal of Memory and Language 43, 335–361. Horn, L. (1996) “Presupposition and implicatures”, in S. Lappin, Ed., The Handbook of Contemporary Semantic Theory, Blackwell, Oxford. Hulme, C., N. Thomson, C. Muir and A. Lawrence (1984) “Speech rate and the development of short term memory span,” Journal of Experimental Child Psychology 43, 35-52. Hulme, C. and V. Tordoff (1989) “Working memory development: The effects of speech rate, word length, and acoustic similarity on serial recall,” Journal of Experimental Child Psychology 47, 72-87. Jeffries, S. A. and J. E. Everatt (2003). “Differences between dyspraxics and dyslexics in sequence learning and working memory,” Dyspraxia Foundation Professional Journal 2, 12–21. Jeffries, S. and J. Everatt (2004) “Working Memory: Its Role in Dyslexia and Other Specific Learning Difficulties,” Dyslexia 10, 196–214. Joanisse, M. F., F. R. Manis, P. Keating and M. S. Seidenberg (2000). “Language deficits in dyslexic children: Speech perception, phonology, and morphology,” Journal of Experimental Child Psychology 77, 30–60. Kail, R., (1992). “Processing speed, speech rate and memory,” Developmental Psychology 28, 899-904. Kipp, K. H., and G. Mohr (2008) “Remediation of developmental dyslexia: Tackling a basic memory deficit” Cognitive Neuropsychology, 1, 1-18. Mann, V. A., & Liberman, I. Y. (1984) “Phonological awareness and verbal short term memory: Can they presage early reading problems?,” Journal of Learning Disabilities 17, 592–599. Mann, V. A., D. P. Shankweiler and S. T. Smith (1984) “The association between comprehension of spoken sentences and early ability: The role of phonetic representation,” Journal of Child Language 11, 627– 643. Martin, R., M. Lesch and M. Bartha (1999) “Independence of input and output phonology in word processing and short-term memory,” Journal of Memory and Language 41, 3–29. 43 Rack, J. (1994). “Dyslexia: The phonological deficit hypothesis,” in A. Fawcett and R. Nicholson, eds, Dyslexia in Children. Multidisciplinary Perspectives, Harvester Wheatsheaf, London. Ramus, F. (2001). “Outstanding questions about the phonological processing in dyslexia,” Dyslexia 7, 197-216. Ramus, F. (2003) “Developmental dyslexia: Specific phonological impairment deficit or general sensorimotor dysfunction?,” Current Opinion in Neurobiology 13, 212–218. Reinhart, T. (2006) Interface Strategies. MIT Press, Cambridge, MA. Rispens, J. (2004) Syntactic and Phonological Processing in Developmental Dyslexia, Doctoral Dissertation, University of Groningen, Groningen Dissertations in Linguistics, Ipskamp, Enschede. Sartori, G., R. Job and P. E. Tressoldi (1995) Batteria per la valutazione della dislessia e della disortografia evolutiva. O.S. Firenze. Schankweiler, D., S. Crain, L. Katz, A. Fowler, A. Liberman, S. Brady, R. Thornton, E. Lundquist, L. Dreyer, J. Fletcher, K. Stuebing, and S. S. B. Shaywitz (1995) “Cognitive profiles of reading-disabled children: Comparison of language skills in phonology, morphology, and syntax,” Psychological Science 6, 149–156. Siegal, L. S. and B. A. Linder (1984) “Short-term memory processes in children with reading and arithmetic learning disabilities,” Developmental Psychology 20, 200–207. Smith. E. E. and A. Geva (2000) “Working memory and its connection to language processing” in Y. Grodzinsky, L. Shapiro & D. Swinney, eds., Language and the Brain: Representation and Processing, Academic Press, San Diego. Snowling, M. J. (2000) Dyslexia, Blackwell, Oxford. Snowling, M. J. (2001) “From language to reading and dyslexia,” Dyslexia 7(1), 37–46. Stanovich, K. E., A. E. Cunningham and D. J. Freeman (1984) “Intelligence, cognitive skills and early reading progress,” Reading Research Quarterly 19, 278–303. Stein, C., H. S. Cairns and E. Zurif (1984) “Sentence comprehension limitations related to syntactic deficits in reading disabled children,” Applied Psycholinguistics 5, 305–321. 44 Thompson-Schill, S. L. (2003) “Neuroimaging studies of semantic memory: Inferring “how” from “where.”,” Neuropsychologia 41(3), 280–292. Thompson-Schill, S. L., M. D’Esposito, and I. P. Kan (1999) “Effects of repetition and competition on activity in left prefrontal cortex during word generation,” Neuron 23(3), 513–522. Thompson-Schill S. L., D. Swick, M. J. Farah, M. D’Esposito, I. P. Kan, and R. T. Knight (1998) “Verb generation in patients with focal frontal lesions: A neuropsychological test of neuroimaging findings,” Proceedings of the National Academy of Sciences of the United States of America 95, 15855–15860. Towse, J. N., G. Hitch, and U. Hutton (1998) “A reevaluation of working memory capacity in children”, Journal of Memory and Language 39, 195-217. Vellutino, F. R. (1979) Dyslexia: Theory and Research. MIT Press, Cambidge, MA. Wilsenach, C. (2006) Syntactic Processing in Developmental Dyslexia and in Specific Language Impairment, Doctoral Dissertation, University of Utrecht, LOT Dissertation Series, Utrecht. Zempleni, M.-Z., R. Renken, J. C. J. Hoeks, J. M. Hoogduin and L. A. Stowe (2006) “Semantic ambiguity processing in sentence context: Evidence from event-related fMRI,” NeuroImage 34, 1270–1279. 45 Appendix Experiment 1 Experiment 1 comprised two different models, each one consisting of 1 warm up, 2 fillers and 8 experimental items; there were two experimental items for each of the four conditions, namely: (i) Z1f (i.e. a zero pronoun was used, when possible antecedents differed for only one visual feature), (ii) Z 2f (i.e. a zero pronoun was used, when possible antecedents differed for two visual features), (iii) R 1f (i.e. a phonetically realized pronoun was used, when possible antecedents differed for only one feature) and (iv) R 2f (i.e. a phonetically realized pronoun was used, when possible antecedents differed for two features). The warm up and the fillers were identical in both models; for what concerns the experimental items, instead, the short stories told by the experimenter were the same for both models, while the target sentences uttered by Little Red Riding Hood were different in each model. For brevity, then, I will report in this section each story only once, followed by both target sentences. WARM UP Questi sono un bambino rosso con la sua bicicletta e una bambina coi capelli neri con il suo asinello. Guarda, il bambino rosso è salito sulla sua bicicletta! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Per fortuna Cappuccetto Rosso ha visto cos’è successo! Proviamo a chiederle che ce lo racconti. Cappuccetto, cos’è successo? «Prima il bambino rosso è salito sulla sua bicicletta, e poi la bambina coi capelli neri è salita sul suo asinello!» ‘In this picture, there are a red boy with his bicycle and a girl with black hair with her donkey. Look, the red boy has mounted on his bicycle! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Let us ask her to tell us what happened. Little Red Riding Hooh, what happened? «First the red boy mounted her bicycle, and then the girl with black hair mounted her donkey»’ FILLER 1 Questi sono quattro bambini: un bambino biondo blu, un bambino biondo verde, una bambina rossa e una bambina mora gialla. Guarda, il bambino biondo blu sta salutando il bambino biondo verde! Adesso però i 46 bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Per fortuna Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? “Prima il bambino biondo blu ha salutato il bambino biondo verde, e poi la bambina mora rossa ha salutato la bambina bionda gialla.” ‘In this picture there are four children: a blond blue boy, a blond green boy, a brown red girl and a brown yellow girl. Look, the blond blue boy is greeting the blond green boy! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened!. Little Red Riding Hood, what happened? «First the blond blue boy greeted the blond green boy and then the brown red girl greeted the brown yellow girl!»’ ITEM A “Questi sono quattro bambini biondi, uno rosso, uno verde, uno blu e uno giallo. Guarda, il bambino rosso sta salutando il bambino verde! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Per fortuna Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? N 1F «Prima il bambino rosso ha salutato il bambino verde e poi ha salutato il bambino blu» R 1F «Prima il bambino rosso ha salutato il bambino verde e poi lui ha salutato il bambino blu» ‘In this picture, there are four blond boys who wear clothes of different colours: red, green, blue and yellow. Look, the red boy is greeting the green boy! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? N 1F «First the red boy greeted the green boy and then Ø greeted the blue boy» R 1F «First the red boy greeted the green boy and then he greeted the blue boy»’ ITEM B Queste sono quattro bambine. Una mora blu, una bionda blu, una mora rosa e una bionda rosa. Guarda, la bambina mora blu ha fatto un regalo alla bambina bionda blu! 47 Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? R 1F «Prima la bambina mora blu ha dato un regalo alla bambina bionda blu, e poi lei ha dato un regalo alla bambina bionda rosa» N 1F «Prima la bambina mora blu ha dato un regalo alla bambina bionda blu, e poi ha dato un regalo alla bambina bionda rosa» ‘In this picture there are four girls: a brown blue girl, a blond blue girl, a brown pink girl and a blond pink girl. Look, the brown blue girl gives the blond blue girl a present! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? R 2F«First the brown blue girl gave the blond blue girl a present and then she gave the blond pink girl a present» N 2F «First the brown blue girl gave the blond blue girl a present and then Ø gave the blond pink girl a present»’ ITEM C “Questi sono quattro bambini biondi, uno rosso, uno verde, uno blu e uno giallo. Guarda, il bambino rosso ha dato un panino al bambino verde! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? R 1F «Prima il bambino rosso ha dato un panino al bambino verde, e poi lui ha dato un panino al bambino giallo!» N 1F «Prima il bambino rosso ha dato un panino al bambino verde, e poi ha dato un panino al bambino giallo!» ‘In this picture there are four blond boys who wear clothes of different colors: red, green, blue and yellow. Look, the red boy gave the green boy a roll! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? 48 R 1F «First the red boy gave the green boy a roll and then he gave the yellow boy a roll!» N1F «First the red boy gave the green boy a roll and then Ø gave the yellow boy a roll!»’ ITEM D – N 2f “Queste sono quattro bambine. Una mora verde, una bionda verde, una mora rossa e una bionda rossa. Guarda, la bambina mora verde ha regalato un orsetto a quella bionda verde! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? N 2F «Prima la bambina mora verde ha regalato un orsetto alla bambina bionda verde e poi ha regalato un orsetto alla bambina mora rossa». R 2F «Prima la bambina mora verde ha regalato un orsetto alla bambina bionda verde e poi lei ha regalato un orsetto alla bambina mora rossa». ‘In this picture there are four girls: a brown green girl, a blond green girl, a brown red girl and a blond red girl. Look, the brown green girl presents a teddy bear to the blond green girl! Now the girls are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? N 2F «First the brown green girl presented a teddy bear to the blond green girl and then Ø presented a teddy bear to the brown red girl» R 2F «First the brown green girl presented a teddy bear to the blond green girl and then she presented a teddy bear to the brown red girl»’ FILLER 2 “Queste sono quattro bambine: una bambina mora e una bambina bionda con la tuta azzurra, e una bambina mora e una bambina bionda con la tuta rosa. Guarda, la bambina mora azzurra ha preso un gelato! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? «Prima la bambina mora azzurra ha preso un gelato, e poi anche la bambina mora rosa ha preso un gelato!» 49 ‘In this pictures there are four girls: a brown pink girl, a blond pink girl, a brown blue girl and a blond blue girl. Look the brown blue girl is eating an ice-cream! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? «First the brown blue girl ate an ice-cream, and then the brown pink girl ate an ice-cream, too!»’ ITEM E Queste sono quattro bambine more, una blu, una gialla, una rossa e una verde. Guarda, la bambina blu ha dato un fiore alla bambina gialla! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? R 1F «Prima la bambina blu ha dato un fiore alla bambina gialla, e poi lei ha dato un fiore alla bambina verde!» N 1F «Prima la bambina blu ha dato un fiore alla bambina gialla, e poi Ø ha dato un fiore alla bambina verde!» ‘In this picture there are four brown girls who wear dresses of different colours: blue, yellow, red and green. Look, the blue girl gave the yellow girl a flower! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hooh, what happened? R 1F «First the blue girl gave the yellow girl a flower and then she gave the green boy a flower!» N 1F «First the blue girl gave the yellow girl a flower and then Ø gave the green boy a flower!»’ ITEM F Questi sono quattro calciatori. Un bambino moro rosso, uno biondo rosso, uno moro blu e uno biondo blu. Guarda, il bambino moro rosso ha lanciato la palla al bambino biondo rosso! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? N 2F «Prima il bambino moro rosso ha lanciato la palla al bambino biondo rosso e poi l’ha lanciata al bambino moro blu!» 50 R 2F «Prima il bambino moro rosso ha lanciato la palla al bambino biondo rosso e poi lui l’ha lanciata al bambino moro blu!» ‘In this picture there are four footballer. A brown red boy, a blond red boy, a brown blue boy and a blond blue boy. Look, the brown red boy threw the ball to the blond red boy. Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Let us ask her to tell us what happened. Little Red Riding Hooh, what happened? N 2F «First the brown red boy threw the ball to the blond red boy and then Ø threw the ball to the brown blue boy!» R 2F «First the brown red boy threw the ball to the blond red boy and then she threw the ball to the brown blue boy!»’ ITEM G Questi sono quattro bambini biondi con le loro biciclette: uno giallo, uno blu, uno rosso e uno verde. Guarda il bambino giallo ha colorato la bicicletta del bambino blu! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? N 1F «Prima il bambino giallo ha colorato la bicicletta del bambino blu e poi ha colorato la bicicletta del bambino rosso!» R 1F «Prima il bambino giallo ha colorato la bicicletta del bambino blu e poi lui ha colorato la bicicletta del bambino rosso!» ‘In this picture there are four blond boys with their bicycles. The boys wear clothes of different colors: yellow, blue, red and green. Look, the yellow boy colored the blue boy’s bicycle! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? N 1F «First the yellow boy colored the blue boy’s bicycle and then Ø colored the red boy’s bicycle!» R 1F «First the yellow boy colored the blue boy’s bicycle and then he colored the red boy’s bicycle!»’ ITEM H 51 Queste sono quattro bambine. Una mora azzurra, una bionda azzurra, una mora rosa e una bionda rosa. Tutte le bambine hanno il loro fiorellino. Guarda, la bambina mora azzurra ha annaffiato il fiore della bambina mora rosa! Adesso però i bambini si sono nascosti dietro una tenda e non possiamo vedere cosa succede! Però Cappuccetto Rosso ha visto cos’è successo! Cappuccetto, cos’è successo? R 2F «Prima la bambina mora azzurra ha annaffiato il fiore della bambina mora rosa, e poi lei ha annaffiato il fiore della bambina bionda azzurra!» N 2F «Prima la bambina mora azzurra ha annaffiato il fiore della bambina mora rosa, e poi Ø ha annaffiato il fiore della bambina bionda azzurra!» ‘In this picture there are four girls: a brown blue girl, a blond blue girl, a brown pink girl and a blond pink girl. Every girl has her flower and needs it to be watered. Look, the brown blue girl watered the brown pink girl’s flower! Now the children are hiding behind a curtain and we cannot see what is happening. Fortunately, Little Red Riding Hood could see what happened! Little Red Riding Hood, what happened? R 2F «First the brown blue girl watered the brown pink girl’s flower and then she watered the blond blue girl’s flower!» N 2F «First the brown blue girl watered the brown pink girl’s flower and then Ø watered the blond blue girl’s flower!»’ Experiment 2 Experiment 2 comprised four different models, each one consisting of 6 fillers and 10 experimental items. The fillers were identical in each model, while the target sentences of the experimental items were different. The experiment comprised four different conditions, namely: (i) Z T, where a zero pronoun was used in a context in which the sentence was true, (ii) Z F, where the zero pronoun was used in a false context, (iii) RT, with a realized pronoun used in a true context and (iv) R F, with a realized pronoun used in a false context. There were two experimental items for Conditions “ZT” and “RT” and three items for Conditions “Z F” and “R F”. 52 Since the short stories told by the experimenter were the same for all models, while the target sentence uttered by the Clumsy Detective were different, I report here each story only once, followed by the corresponding target sentences. FILLER 1 Guarda questi personaggi. Questa è Biancaneve, e questo è un nano. Biancaneve e il nano si incontrano davanti alla casetta dei nani e poi decidono di ballare insieme. All’improvviso, arriva l’ispettore pasticcione e dice: «Ho capito cos’è successo! Biancaneve ha incontrato un nano e poi ha ballato con lui!» ‘Look at these characters. This is Snow White, and this is a dwarf. Snow White and the dwarf meet in front of the Seven Dwarf’s House and then they decide to dance together. Suddenly, the Clumsy Detective arrives and says: «I know what happened! Snow White met a dwarf and then she danced with him!»’ FILLER 2 Guarda questi personaggi. Questa è Alice, e questo è Gianni. Gianni e Alice giocano insieme con il computer. Poi Alice torna a casa per fare i compiti. All’improvviso arriva l’Ispettore Pasticcione e dice: «Ho capito cos’è successo! Gianni ha giocato con il computer insieme ad Alice e poi è tornato a casa!» ‘Look at these characters. This is Alice, and this is John. John and Alice are playing computer games together. Then Alice goes back home to do her homework. Suddenly the Clumsy Detective arrives and says: «I know what happened! John played computer games with Alice and then Ø went back home!»’ FILLER 3 Guarda questi personaggi. Questo è Paperino e questa è la sua fidanzata Paperina. Paperino e Paperina vanno insieme a fare compere. Al ritorno Paperino porta tutti i pacchetti da solo, mentre Paperina non porta niente. All’improvviso, arriva l’Ispettore Pasticcione e dice: «Ho capito cos’è successo! Paperino è andato a fare spese con Paperina e poi lei ha portato tutti i pacchetti da sola!» 53 ‘Look at these characters. This is Donald Duck, and this is his girlfriend, Daisy Duck. Donald Duck and Daisy Duck go shopping together. On the return, Donald Duck carries all bags alone, while Daisy Duck does not carry anything. Suddenly the Clumsy Detective arrives and says: «I know what happened! Donald Duck went shopping with Daisy Duck and then she carried all bags alone!»’ ITEM A Guarda questi personaggi. Questo è Paperino e questo è Topolino. Topolino e Paperino giocano a calcio insieme. Poi Paperino va a suonare il tamburo. All’improvviso arriva l’Ispettore Pasticcione e dice: Z F «Ho capito cos’è successo! Topolino ha giocato a calcio con Paperino e poi ha suonato il tamburo!» Z T «Ho capito cos’è successo! Paperino ha giocato a calcio con Topolinoi e poi ha suonato il tamburo!» R F «Ho capito cos’è successo! Paperino ha giocato a calcio con Topolino e poi lui ha suonato il tamburo!» R T «Ho capito cos’è successo! Topolino ha giocato a calcio con Paperino e poi lui ha suonato il tamburo!» ‘Look at these characters. This is Donald Duck, and this is Mickey Mouse. Donald Duck and Mickey Mouse are playing football together. Then Donald Duck beats the drum. Suddenly the Clumsy Detective arrives and says: Z F «I know what happened! Mickey Mouse played football with Donald Duck and then Ø beat the drum» Z T «I know what happened! Donald Duck played football with Mickey Mouse and then Ø beat the drum» R F «I know what happened! Donald Duck played football with Mickey Mouse and then he beat the drum» R T «I know what happened! Mickey Mouse played football with Donald Duck and then he beat the drum»’ ITEM B 54 Guarda questi personaggi. Questo è Winnie the Pooh, e questo è il suo amico Tigro. Oggi è il compleanno di Tigro. Per festeggiare, Tigro organizza una festa e invita il suo amico Winnie. Winnie va alla festa e porta una torta a Tigro. Tigro lo ringrazia. All’improvviso arriva l’Ispettore Pasticcione e dice: Z F «Ho capito cos’è successo! Tigro ha festeggiato con Winnie e gli ha regalato una torta!» Z T «Ho capito cos’è successo! Winnie ha festeggiato con Tigro e gli ha regalato una torta» R F « Ho capito cos’è successo! Winnie ha festeggiato con Tigro e lui gli ha regalato una torta» R T « Ho capito cos’è successo! Winnie ha festeggiato con Tigro e lui gli ha regalato una torta» ‘Look at this characters. This is Winnie the Pooh and this is his friend Tigger. Today is Tigger’s birthday. To celebrate his birthday, Tigger organizes a party and invites his friend Winnie. Winnie goes to the party and gives Tigger a cake. Tigger thanks him. Suddenly, the Clumsy Detective arrives and says: Z F «I know what happened! Tigger has a party with Winnie and Ø gave him a cake» Z T «I know what happened! Winnie has a party with Tigger and Ø gave him a cake» R F «I know what happened! Winnie has a party with Tigger and he gave him a cake» R T «I know what happened! Tigger has a party with Winnie and he gave him a cake»’ ITEM C Guarda questi personaggi. Questo è Winnie the Pooh e questo è il suo amico Pimpi. Winnie e Pimpi vanno a fare una passeggiata nel bosco ma c’è un vento fortissimo. Visto che il vento è molto forte, il povero Pimpi vola via. All’improvviso, arriva l’Ispettore Pasticcione e dice: Z T «Ho capito cos’è successo! Pimpi è andato a fare una passeggiata con Winnie è poi a causa del vento è volato via!» Z F «Ho capito cos’è successo! Winnie è andato a fare una passeggiata con Pimpi è poi a causa del vento è volato via!» R F «Ho capito cos’è successo! Winnie è andato a fare una passeggiata con Pimpi è poi lui a causa del vento è volato via!» R T «Ho capito cos’è successo! Pimpi è andato a fare una passeggiata con Winnie è poi lui a causa del vento è volato via!» 55 ‘Look at these characters. This is Winnie the Pooh and this is his friend Piglet. Winnie and Piglet go for a walk together, but a very strong wind is blowing. The wind is so strong, that the poor Piglet flies away! Suddenly, the Clumsy Detective arrives and says: Z T «I know what happened! Piglet went for a walk with Winnie and then, because of the wind, Ø flew away!» Z F «I know what happened! Winnie went for a walk with Piglet and then, because of the wind, Ø flew away!» R T «I know what happened! Winnie went for a walk with Piglet and then, because of the wind, he flew away!» R F «I know what happened! Piglet went for a walk with Winnie and then, because of the wind, he flew away!»’ ITEM D Guarda questi personaggi. Questa è Lisa e questa è la sua mamma. Lisa e la mamma giocano insieme: Lisa dà la bambola alla mamma. Poi Lisa prende in braccio l’orsetto. All’improvviso, arriva l’Ispettore Pasticcione e dice: Z T «Ho capito cos’è successo! Lisa ha dato la bambola alla mamma e poi ha preso l’orsetto!» Z F «Ho capito cos’è successo! La mamma ha preso la bambola che le ha dato Lisa e poi ha preso l’orsetto!» R T «Ho capito cos’è successo! La mamma ha preso la bambola che le ha dato Lisa e poi lei ha preso l’orsetto!» R F «Ho capito cos’è successo! Lisa ha dato la bambola alla mamma e poi lei ha preso l’orsetto!» ‘Look at these characters. This is Lisa, and this is her Mum. Lisa and her Mum are playing together: Lisa gives a puppet to her mother. Then Lisa takes the Teddy Bear in her arms. Suddenly, the Clumsy Detective arrives and says: Z T «I know what happened! Lisa gave Mum a puppet and then Ø took the Teddy Bear in her arms» Z F «I know what happened! Mum took the puppet Lisa gave her and then Ø took the Teddy Bear in her arms» 56 R T «I know what happened! Mum took the puppet Lisa gave her and then she took the Teddy Bear in her arms» R F «I know what happened! Lisa gave Mum a puppet and then she took the Teddy Bear in her arms»’ ITEM E Guarda questi personaggi. Questa è Minnie e questa è Paperina. Minnie e Paperina passano il pomeriggio insieme e ballano le loro canzoni preferite. Poi Paperina va in cucina e prepara la cena. All’improvviso, arriva l’Ispettore Pasticcione e dice: Z T «Ho capito cos’è successo! Paperina ha ballato insieme a Minnie e poi Ø ha preparato la cena!» Z F «Ho capito cos’è successo! Minnie ha ballato insieme a Paperina e poi Ø ha preparato la cena!» R T «Ho capito cos’è successo! Minnie ha ballato insieme a Paperina e poi Ø ha preparato la cena!» R F «Ho capito cos’è successo! Paperina ha ballato insieme a Minnie e poi lei ha preparato la cena!»’ ‘Look at these characters. This is Minnie and this is Daisy Duck. Minnie and Daisy Duck spend the afternoon together and they dance her favourite songs. Then Daisy Duck goes to the kitchen and prepares dinner. Suddenly, the Clumsy Detective arrives and says: Z T «I know what happened! Daisy Duck danced with Minnie and then Ø prepared dinner!» Z F «I know what happened! Minnie danced with Daisy Duck and then Ø prepared dinner!» R T «I know what happened! Daisy Duck danced with Minnie and then she prepared dinner!» R F «I know what happened! Minnie danced with Daisy Duck and then she prepared dinner!»’ ITEM F Guarda questi personaggi. Questo è Paperino e questo è Pippo. Paperino e Pippo tornano a casa da scuola insieme. Appena tornato a casa, Paperino inizia subito a fare i compiti prima di uscire a giocare. Arriva l’ispettore pasticcione e dice: Z T «Ho capito cos’è successo! Paperino è tornato a casa insieme a Pippo e poi Ø si è messo a fare i compiti» Z F «Ho capito cos’è successo! Pippo è tornato a casa insieme a Paperino e poi Ø si è messo a fare i compiti» 57 R T «Ho capito cos’è successo! Pippo è tornato a casa insieme a Paperino e poi lui si è messo a fare i compiti» R F «Ho capito cos’è successo! Paperino è tornato a casa insieme a Pippo e poi lui si è messo a fare i compiti» ‘Look at these characters. This is Donald Duck and this is Goofy. Donald Duck and Goofy are going home together. At home, Donald Duck starts to do his homework. Suddenly, the Clumsy Detective arrives and says: Z T «I know what happened! Donald Duck went home with Goofy and then Ø started to do his homework» Z F «I know what happened! Goofy went home with Donald Duck and then Ø started to do his homework» R T «I know what happened! Goofy went home with Donald Duck and then he started to do his homework» R F «I know what happened! Donald Duck went home with Goofy and then he started to do his homework»’ ITEM G Guarda questi personaggi. Questa è Cappuccetto Rosso, e questa è la sua mamma. La mamma dà a Cappuccetto un cestino colmo di dolci da portare alla nonna. Cappuccetto prende il cestino e si incammina nel bosco per andare dalla nonna. All’improvviso, arriva l’Ispettore Pasticcione e dice: Z T «Ho capito cos’è successo! Cappuccetto ha preso il cestino che le ha dato la mamma e poi è andata dalla nonna!» Z F «Ho capito cos’è successo! La mamma ha dato il cestino a Cappuccetto e poi è andata dalla nonna!» R T «Ho capito cos’è successo! La mamma ha dato il cestino a Cappuccetto e poi lei è andata dalla nonna!» R F «Ho capito cos’è successo! Cappuccetto ha preso il cestino che le ha dato la mamma e poi lei è andata dalla nonna!» 58 ‘Look at these characters. This is Little Red Riding Hood, and this is her Mum. Mum gives Little Red Riding Hood a small basket full of sweets, and she asks Little Red Riding Hood to take it to her Grandmother. So Little Red Riding Hood takes the basket and goes to the Grandmother. Suddenly, the Clumsy Detective arrives and says: Z T «I know what happened! Little Red Riding Hood took the small basket that Mum gave her and then Ø went to the Grandmother!» Z F «I know what happened! Mum gave Little Red Riding Hood the small basket and then Ø went to the Grandmother!» R T «I know what happened! Mum gave Little Red Riding Hood the small basket and then she went to the Grandmother!» R F «I know what happened! Little Red Riding Hood took the small basket that Mum gave her and then she went to the Grandmother!»’ ITEM H Guarda questi personaggi. Questa è Cenerentola. E questa è la fata madrina. Cenerentola è triste perché non ha il vestito adatto per andare al ballo del principe e chiama la fata turchina per chiederle aiuto. La fata con la sua bacchetta fa una magia e fa apparire per Cenerentola un abito bellissimo. All’improvviso arriva l’Ispettore Pasticcione e dice: Z T «La fata madrina è andata da Cenerentola e ha fatto apparire un bellissimo vestito!» Z F «Cenerentola ha chiamato la fata madrina e ha fatto apparire un bellissimo vestito!» R T «Cenerentola ha chiamato la fata madrina e lei ha fatto apparire un bellissimo vestito!» R F «La fata madrina è andata da Cenerentola e lei ha fatto apparire un bellissimo vestito!» ‘Look at these characters. This is Cinderella, and this is the Fairy Godmother. Cinderella is unhappy, because she doesn’t have a beautiful dress and she cannot go to the ball. So she calls the Fairy for help. The Fairy with her magic wand creates a very beautiful dress for Cinderella. Suddenly the Clumsy Detective arrives and says: Z T «I know what happened! The Fairy Godmother went to Cinderella and Ø made a very beautiful dress appear!» 59 Z F«I know what happened! Cinderella called the Fairy Godmother and Ø made a very beautiful dress appear!» R T «I know what happened! Cinderella called the Fairy Godmother and she made a very beautiful dress appear!» R F «I know what happened! The Fairy Godmother went to Cinderella and she made a very beautiful dress appear!»’ ITEM I Guarda questi personaggi. Questo è Pluto e questo è Paperino. Pluto e Paperino vanno in cerca di un tesoro nascosto nella montagna. Paperino, che è più fortunato, trova un diamante. All’improvviso, arriva l’Ispettore Pasticcione e dice: Z F «Ho capito cos’è successo! Pluto è andato a cercare un tesoro insieme a Paperino e poi ha trovato un diamante!» R F «Ho capito cos’è successo! Paperino è andato a cercare un tesoro insieme a Pluto e poi lui ha trovato un diamante!» ‘Look at these characters. This is Pluto, and this is Donald Duck. Pluto and Donald Duck go in search of a treasure hidden in the mountains. Donald Duck, who is more lucky, finds a diamond. Suddenly, the Clumsy Detective arrives and says: Z F «I know what happened! Pluto went in search of a treasure with Donald Duck and then Ø found a diamond!» R F «I know what happened! Donald Duck went in search of a treasure with Pluto and then he found a diamond!»’ ITEM L Guarda questi personaggi. Questo è Gianni e questo è il suo gatto. Gianni dà al gatto una palla per farlo giocare. Il gattino gioca con la palla e la fa rotolare. All’improvviso arriva l’Ispettore Pasticcione e dice: Z F «Ho capito cos’è successo! Gianni ha dato la palla al gatto e poi l’ha fatta rotolare!» R F «Ho capito cos’è successo! Il gatto ha preso la palla che gli ha dato Gianni e poi lui l’ha fatta rotolare!» 60 ‘Look at these characters. This is John and this is his cat. John gives the cat a ball to make him play. The cat plays with the ball and makes it roll. Suddenly, the Clumsy Detective arrives and says: Z F «I know what happened! The cat took the ball that John gave him and then he made it roll!» R F «I know what happened! The cat took the ball that John gave him and then he made it roll!»’ FILLER 6 Guarda questi personaggi. Questo è Paperino e questa è Paperina. Paperino canta una canzone a Paperina. Poi Paperina dà un bacio a Paperino. All’improvviso, arriva l’Ispettore Pasticcione e dice: «Ho capito cos’è successo! Paperino ha cantato una canzone a Paperina e poi lei lo ha baciato!» ‘Look at these characters. This is Donald Duck and this is Daisy Duck. Donald Duck sings a song for Daisy Duck. Then Daisy Duck gives him a kiss. Suddenly, the Clumsy Detective arrives and says: «I know what happened! Donald Duck sang a song for Daisy Duck and then she kissed him»’ 61 Footnotes 1 Scalar implicatures are quantity implicatures based on the use of an informationally weak term from an implicational scale. For further discussions, see Horn (1996), Gazdar (1979), Chierchia (2004). 2 Empty speech can be defined as the overuse of empty words, such as “thing”, “do” or “it”, which characterizes Alzheimer Patients’ speech. 3 MEG is the Magnetoencelography, a translational technology which brings classical cellular Neurophysiology information into clinical applications. MEG provides a precise picture of neuronal activation at different instants of both memory and language processing (Bowyer et al. 2004). 62