Clinical Linguistics & Phonetics

ISSN: 0269-9206 (Print) 1464-5076 (Online) Journal homepage: http://www.tandfonline.com/loi/iclp20

Processing complex pseudo-words in mild cognitive impairment: The interaction of preserved morphological rule knowledge with compromised cognitive ability Christina Manouilidou, Barbara Dolenc, Tatjana Marvin & Zvezdan Pirtošek To cite this article: Christina Manouilidou, Barbara Dolenc, Tatjana Marvin & Zvezdan Pirtošek (2015): Processing complex pseudo-words in mild cognitive impairment: The interaction of preserved morphological rule knowledge with compromised cognitive ability, Clinical Linguistics & Phonetics, DOI: 10.3109/02699206.2015.1102970 To link to this article: http://dx.doi.org/10.3109/02699206.2015.1102970

Published online: 20 Nov 2015.

Submit your article to this journal

Article views: 2

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=iclp20 Download by: [Christina Manouilidou]

Date: 24 November 2015, At: 09:42

CLINICAL LINGUISTICS & PHONETICS http://dx.doi.org/10.3109/02699206.2015.1102970

Processing complex pseudo-words in mild cognitive impairment: The interaction of preserved morphological rule knowledge with compromised cognitive ability Christina Manouilidoua, Barbara Dolencb, Tatjana Marvinc, and Zvezdan Pirtošekd,e Department of Philology – Linguistics Section, University of Patras, Rio-Patras, Greece; bLaboratory for Cognitive Neuroscience, Department of Neurology, Ljubljana University Medical Centre, Ljubljana, Slovenia; c Department of Comparative and General Linguistics, University of Ljubljana, Ljubljana, Slovenia; dNeurology Clinic, Ljubljana University Medical Centre, Ljubljana, Slovenia; eDepartment of Neurology, University of Ljubljana, Ljubljana, Slovenia

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

a

ABSTRACT

ARTICLE HISTORY

Mild cognitive impairment (MCI) affects the cognitive performance of elderly adults. However, the level of severity is not high enough to be diagnosed with dementia. Previous research reports subtle language impairments in individuals with MCI specifically in domains related to lexical meaning. The present study used both off-line (grammaticality judgment) and on-line (lexical decision) tasks to examine aspects of lexical processing and how they are affected by MCI. 21 healthy older adults and 23 individuals with MCI saw complex pseudo-words that violated various principles of word formation in Slovenian and decided if each letter string was an actual word of their language. The pseudo-words ranged in their degree of violability. A task effect was found, with MCI performance to be similar to that of healthy controls in the off-line task but different in the on-line task. Overall, the MCI group responded slower than the elderly controls. No significant differences were observed in the off-line task, while the online task revealed a main effect of Violation type, a main effect of Group and a significant Violation × Group interaction reflecting a difficulty for the MCI group to process pseudo-words in real time. That is, while individuals with MCI seem to preserve morphological rule knowledge, they experience additional difficulties while processing complex pseudo-words. This was attributed to an executive dysfunction associated with MCI that delays the recognition of ungrammatical formations.

Received 5 February 2015 Accepted 29 September 2015 KEYWORDS

Executive functions; language disorders; MCI; morphology; psycholinguistics

Introduction The term mild cognitive impairment (MCI) refers to a condition between normal aging and dementia which is often seen as the first step towards dementia (Chertkow, 2002; Petersen et al., 2001). When MCI individuals demonstrate impairments in domains other than memory, including language, they are more likely to develop dementia than are those with a pure memory impairment (Petersen, 2003). Thus, understanding the nature of language impairment and possibly identifying sensitive measures of linguistic impairment constitutes a vital tool in early detection of dementia. While there exists plentiful evidence CONTACT Christina Manouilidou Rio-Patras 26504, Greece. © 2015 Taylor & Francis

[email protected]

Department of Philology, University of Patras,

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

2

C. MANOUILIDOU ET AL.

of language deficits in MCI mainly from standardized assessment tools (for a review see Taler & Phillips, 2008), psycholinguistic studies of language processing are scarce. The few studies that have employed psycholinguistic methodology have revealed disturbances in performance mainly at the lexical-semantic level reflecting an impaired semantic network in this population. Specifically, MCI individuals exhibit a deficit in the sentence–picture matching task (Taler & Jarema, 2004), no advantage in processing ambiguous words in lexical decision task compared to controls (Taler & Jarema, 2006), impaired semantic categorization (Olichney et al., 2002), impaired semantic encoding (Puregger et al., 2003), impaired naming and semantic knowledge of objects (Joubert et al., 2010), as well as impaired semantic priming (Davie et al., 2004; Manouilidou et al., 2014). Verbal fluency has proven to be a controversial domain for MCI. A number of studies report significant deficits (Murphy, Rich, & Troyer, 2006; Petersen et al., 1999; Wang, Lirng, Lin, Chang, & Liu, 2006), while at the same time, a number of studies fail to find a deficit in verbal fluency tasks (Albert et al., 2007; Collie, Maruff & Currie, 2002; Lambon Ralph et al., 2003). Similarly, structural aspects of language, namely phonological and morphological structure as well as syntax, are thought to show no alterations. This claim is particularly interesting given that in syntactic tasks for instance, MCI individuals (like people with early Alzheimer’s) either demonstrate no significant differences compared to age-matched controls (De Jager et al., 2003; Hodges et al., 2006) or when they have been found to have difficulties, these findings have been attributed to deficits in domains other than syntax (e.g. Lambon Ralph et al., 2003). Given the unclear language performance of MCI individuals, some researchers have claimed that their performance depends not so much on their language knowledge but on the kind of task they are asked to perform (Duong et al., 2006). For instance, patient performance deteriorates with increased task complexity, an indication that other-than-language cognitive systems, such as executive functions, might be interfering in language processing. The term executive function (also referred to as cognitive control) is not a unitary concept but rather a cover term which defines complex cognitive processing requiring the coordination of several subprocesses in order to achieve a particular goal (Elliot, 2003). Executive functions, for instance, allow the language user to adjust to cognitive and linguistic demands while interpreting a sentence (Miller & Cohen, 2001). Novick et al. (2005) have shown that executive functions affect language comprehension by allowing the speaker to bypass initial erroneous interpretations of linguistic input, thus preventing a failure in comprehension. The same researchers have further proposed that reinforcement and treatment of certain executive functions (e.g. working memory, conflict resolution) could be beneficiary for language processing given that one finds similar underlying mechanisms when dealing with linguistic phenomena. Specifically, in a recent study, Novick & colleagues tested if training in executive functions improves readers’ ability to recover from misanalysis of the type found in garden-path sentences during language processing (Novick et al., 2014). In order to correctly interpret a garden-path, participants have to re-characterize the information already received based on new evidence. In other words, the processor has to choose between two conflicting representations and opt for the appropriate one. The reasoning behind the study was that readers use a great deal of executive function mechanisms in order to revise their initial interpretation when they encounter new information which contradicts the initial analysis and representation. This

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

3

was based on the hypothesis that executive functions help readers deal with contradictory representations by assisting them in the process of organizing and regulating conflicting information (Miller & Cohen, 2001). Results of this and subsequent studies showed that an improvement in executive functions can cause an improvement in linguistic performance (in error rate and/or in response times). Executive function reinforcement (either by working memory enhancement or by conflict resolution treatment) of healthy adults as well as of stroke patients with damage at the left ventro-lateral prefrontal cortex (an area related to executive functions) has improved performance at comprehension tasks (Hoffman et al., 2010), verbal fluency (Kan & Thompson-Schill, 2004; Novick et al., 2009; Robinson et al., 1998; Schnur et al., 2008) as well as lexical and syntactic ambiguity resolution (Bedny et al., 2007; Hussey et al., 2010; Hussey & Novick, 2012; Novick et al., 2005). A number of studies have brought to light an executive dysfunction in MCI in a variety of tasks (e.g. Belleville, Bherer, Lepage, Chertkow, & Gauthier, 2008; Brandt et al., 2009; Traykov et al., 2007). For instance, Belleville et al. (2008) reported that individuals with MCI showed a deficit when required to concurrently maintain two relevant task sets when they had to perform spatial switching. Their performance was significantly better compared to Alzheimer’s patients but worse compared to healthy controls. There are also attempts to connect executive dysfunction in MCI with later progression to dementia (e.g. Albert, Blacker, Moss, Tanzi, & McArdle, 2007; Belanger & Belleville, 2009). Duong et al. (2006) as well as a recent study by Azuma et al. (2013) highlight the role of executive functions in language processing in MCI. In Duong et al. (2006), MCI participants performed significantly worse than their age-matched controls in a Stroop–picture naming task in which the secondary task of picture naming involved a semantic search component. The authors suggest that this is the result of a deficit which prevents inhibiting information while performing a semantic task, thus attributing it to an executive control dysfunction. By employing an ambiguity decision task (participants had to decide if the words presented to them had one or more than one meaning), Azuma et al. (2013) found that MCI individuals were slower to make their decision and they were prone to more errors. However, the fact that there was an effect of the number of meanings (few versus many) argues, according to the authors, for the existence of relatively intact semantic representations, thus isolating executive dysfunction as the sole responsible factor for the impaired performance. Taking the above into consideration, in the present study we aimed at investigating morphological knowledge in patients with MCI by looking at their capacity to detect violations of word-formation rules in Slovenian. Two types of tasks were used, an off-line grammaticality judgment task and an on-line lexical decision task, in order to obtain evidence from both non-chronometrized (which is closer to controlled processing) and chronometrized (closer to automatic processing) tools. In an attempt to investigate possible links between impaired linguistic performance and executive dysfunction, we manipulated the linguistic stimuli in such a way that their lexical decision could also require conflict resolution. In other words, we explore the possibility that even though patients would be performing a purely linguistic task, the idea of using pseudo-words which violate different rules of word-formation could provide a fertile ground to examine

4

C. MANOUILIDOU ET AL.

patients’ executive control in purely linguistic terms. This innovative, exploratory approach is further explained in the following section.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Morphological processing, the nature of complex words and executive functions Morphological processing on its own has not been used in the literature as requiring executive control. In this section, we present the reasons that lead us to form the hypothesis that morphological processing does require executive control, based on the nature of morphologically complex words and the way we process them. A morphologically complex word constitutes a multidimensional stimulus which requires evaluation of two types of information: information concerning the suffix on the one hand and the root on the other hand. The two together give rise to certain representations that are open to interpretation. For instance, re+lock yields the representation of relock while un+lock gives rise to the representation of unlock. Even though sharing the same stem, their correct interpretation depends on the evaluation of both stem and affix. Psycholinguistic research on the processing of morphologically complex words supported by neuroimaging data has revealed that the human processor accesses morphological information immediately when confronted with a complex lexical item (see Fruchter et al., 2013 for neuroimaging data; Morris & Stockall, 2013; Rastle & Davis, 2008 for a review of behavioural studies; Royle et al. 2012). This suggests that in healthy populations, morphological constituents are rapidly activated immediately after word viewing. Given that lexical access is a process consisting of various subprocesses, several operations take place after initial decomposition in order to validate the combination of stem and suffix. For instance, behavioural masked-priming experiments (e.g. Rastle et al., 2004) showed that the processor initially decomposes even words like corner into corn + er. In order to avoid the erroneous interpretation of initial decomposition, validation processes take place during the last stage of lexical access where speakers engage their semantic knowledge in order to resolve this morphological garden-path. This has been described either as licensing (Burani et al., 1999) or as recombination (e.g. Fruchter & Marantz, 2015), and it refers to the stage where speakers try to make sense of the decomposed parts of the word. In other words, speakers revisit this incompatible information (corn+er) about how to best characterize the stimulus corner and how to best respond to it, in a similar way as participants solve sentence garden-paths with the involvement of executive functions in Novick et al.’s (2014) terms. That is, in an incongruent condition, where stem and suffix do not match (e.g. corn+er, dis+lock etc.), greater search mechanisms are required since a conflict arises from the processing of these two types of information that are not compatible and give rise to two conflicting representations. This type of conflict is compatible with the conflict described in the literature of executive function and sentence processing, which refers to cases “in which an individual receives incompatible information either about how best to characterize a stimulus or how best to respond to that stimulus” (Novick et al., 2014, p. 187). All this becomes even more complicated when the information stemming from root and suffix has to be evaluated on various grounds, that is, semantic, syntactic, aspectual etc., as explained earlier. Therefore, even though the kind of information that needs to be processed does not originate from two different levels of processing as in mainstream

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

5

executive function tasks, such as the colour words of the Stroop task, and is entirely linguistic in nature (as it is the case in sentence comprehension and garden-path processing), the fact that the processor has to evaluate information from two resources and decide on their compatibility, in terms of multiple linguistic levels, presents a situation of conflict that requires executive control mechanisms. In other words, conflict resolution is demanded not because the processor has to suppress attention from one source and direct it to the other (as in the narrow neuropsychological notion of the term) but because the processor has to revisit and to reevaluate two types of incompatible information which originate from two sources in a single multidimensional stimulus: the morphologically complex word. In this sense, the similarity between the Stroop task and processing incongruent morphologically forms lies on the need to engage control mechanisms. Participants have to organize the linguistic information they know as speakers of a certain language and they have to evaluate this information in order to respond to a stimulus. Thus, we hypothesize that an incongruent type of stimulus, like the ones used in the present study, requires the involvement of linguistic knowledge as well as executive functions given that the processor has to resolve among competing characterizations of linguistic input, where the characterizations in this case pertain to conflicting representations. This idea was explored through the performance of MCI participants, a population which is thought to present with executive functions difficulties.

Psycholinguistic and linguistic background Apart from the fact the initial decomposition takes places immediately upon word encounter, psycholinguistic research has also shed light on the ensuing questions of what kind of information is available after the initial decomposition, how much information is accessed by our processor once the initial decomposition and morpheme recognition has taken place and how much is needed when building word structure. In a recent study by Manouilidou & Stockall (2014), Greek-speaking and English-speaking participants had to make lexical decisions to pseudo-words which violated word-formation rules in both languages, in an attempt to tease apart the contribution of syntactic category information and argument structure information in building deverbal word formations. To this end, the suffix -tis (the semantic equivalent of -er) was used for Greek in order to create agentive nominalizations (e.g. xtis-tis ‘builder’) and pseudo-words violating the c-selection properties of -tis (it requires a +verbal base) as well as its argument structure properties (it requires a +Agent verb). For English, the prefix re- was used which selects unaccusatives (with argument structures, e.g. reappear) and transitive accomplishment verbs (e.g. rebuild), but not unergatives or transitive achievements, activities or stative verbs (see Manouilidou & Stockall, 2014, for a discussion of the theoretical analysis). When these specifications are not respected, then we end up with pseudowords which violate word-formation rules, such as the ones in (1) which show a categorial violation and the ones in (2) which show a thematic or argument structure violation. The stimulus set of this study also included possible words (unattested, well-formed combinations, e.g. rehold) as well as non-words (containing a real affix and a non-attested stem, e.g. *reclow).

6

C. MANOUILIDOU ET AL.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

(1) a. GR, tis + noun: *karekla-tis ‘chairer’, *potiri-tis ‘glasser’ b. ENG, re + adj: *rehappy, *reflat, *reangry (2) a. GR, tis + non-agentive verb: *ksexas-tis ‘forgetter’, *thymi-tis ‘rememberer’ b. ENG, re + unergatives: *resmile, *reboast Results revealed distinct reaction times (RTs) and accuracy rates for each type of pseudo-word, with the ones with categorial violations (rehappy) being faster and easier to reject, followed by the ones with thematic violations (resmile). Both types of violations yielded distinct RTs compared to: (a) pure non-words (*reclow) which were responded to faster than categorial violations and (b) novel words (rehold) which yielded longer RTs compared to thematic violations. This pattern suggests that participants from both languages processed each type of information (syntactic category versus argument structure) at distinct stages, isolating their contribution to word formation. In other words, participants were able to detect the inappropriateness/ conflict in the erroneous combinations of stem and suffix, and they did this in different time courses for different types of information, as suggested by the significantly different RTs each type of pseudo-word elicited. Thus, some pseudo-words were easier to detect and reject (such as those with categorial violations, e.g.*rehappy, *kareklatis ‘chairer’), while others (such as the ones with thematic violations, e.g. *resmile, *ksexastis ‘forgetter’) were harder, suggesting a gradient violability. This finding replicated previous findings from Greek (see Manouilidou, 2006, 2007). A closer look into the processing of these pseudo-words can help us understand the differences between the stimuli and the way they were processed. At first sight, resmile appears to be a well-formed word of English. It does not violate phonotactic rules of English neither pragmatic/world knowledge constraints nor morphological (re- attaches to verbs). Thus, resmile passes successfully through all these levels already creating a representation of “resmile” and it plays in favour for pressing the YES button. However, the processor has to evaluate one more piece of information and try to add it to the already existent representation. This is the fact that re- cannot be attached to unergative verbs, information which stems from the innate linguistic knowledge of every speaker. Thus, the processor faces a conflict which needs to be resolved in order to make the final decision and press the YES or NO button. Since the activation of morphemes is an automatic process, MCI individuals should not have any difficulty with the processing of complex words, consisting of more than one morpheme. However, the processing of pseudo-words containing different kinds of violations requires the evaluation of the combinatorial properties of both base and suffix which involves more controlled, conscious processing. Participants have to access and maintain in working memory the meaning and properties of both components (word stem + affix) and verify their combination as appropriate or not on multiple levels, such as the syntactic category of the base stem and the argument structure properties of the prefix and stem. This process, apart from morphological rule knowledge possibly stemming from an

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

7

intact linguistic system, requires a well-functioning mechanism of conflict resolution. Participants are confronted with formations which violate combinatorial rules of stems and suffixes such as resmile, and they are basically asked to detect this conflict and reject the word as quickly as possible. Thus, by manipulating the number and kind of violations that have to be detected, one manipulates the amount of executive functioning required in the processing of pseudo-words. Slovenian, a southern Slavic language, is particularly suited to allow study of the various steps of derivation and possible violations in this process given that agentive deverbal word formation, apart from syntactic category of the base and argument structure specifications, also needs to take into account the aspectual properties of the base verb. That is, for agentive formation with the suffix -ec (the equivalent of -er) only imperfective verbal stems are required. Any combination between -ec and perfective stem yields ungrammatical word formations (Marvin, 2002; Stramljič Breznik, 1999), as illustrated in example (3) below. (3) a. plavalec ‘swimmer’, from plavati ‘to swim-imp’; *preplavalec (from preplavati ‘to swim-pf’) b. morilec, ‘ murderer’, from moriti ‘to murder-imp’ ; *umorilec (from umoriti ‘to murder-pf’) Thus, the information that needs to be evaluated for Slovenian deverbal word formation requires knowledge and use of the following information in sequential order: grammatical category of the base (needs to be +verb), argument structure information of the base (+agentive) and aspectual information of the base (needs to be +imperfective). This knowledge and the ability to use it when conflicts arise was evaluated in the current study for Slovenian-speaking MCI participants and their healthy age-matched controls.

Methods Participants 21 healthy volunteers (6 males, 15 females, aged 60–79, MEAN: 67.8) and 23 individuals with MCI (5 males, 18 females, aged 55–82, MEAN: 68.6) participated in the study. Patients were recruited from the Neurology Clinic of the University Medical Centre in Ljubljana. All were diagnosed by a neurologist and a neuropsychologist at the Neurology Clinic. Patients were first assessed by the neurologist, who made the first preliminary diagnosis, more in the form of a hypothesis for further examination. Then, all patients went through neuropsychological examination and after that the neurologist examined all of the data and made the final diagnosis. In some cases, patients also made lumbar puncture as well as MR and PET scans. Their neuropsychological performance was examined by the means of standardized tests translated and adapted versions for Slovenian as well as a series of neuropsychological tests. All patients performed below the critical cut-off scores, indicating the presence of cognitive disturbances. The cognitive test battery included the following tasks. For general screening, the Mini-Mental State Examination (MMSE) (Folstein et al., 1975) and MoCA test (Nasreddine et al., 2005) were used. Attention and cognitive speed was tested by using the Stroop task (Stroop, 1935) and

8

C. MANOUILIDOU ET AL.

Table 1. Participant demographic and neuropsychological information. Mean age Education level MMSE MoCA Forward digit span Backward digit span Executive Functions Tower test Letter fluency Category fluency Category switching Category switching (switch accuracy)

MCI 68.6 12.4 (3.1) 28.1 (1.4) 24.1 (3.4) 8.5 (2.1) 7.0 (1.1) 7.5 (3.5)

Older adults 67.8 14.1 (2.24) 28.8 (1.3) 27.1 (2.1) 10.3 (1.4) 8.7 (1.4) 11.2 (1.9)

t(43) t(43) t(43) t(43) t(43) t(43) t(43)

= = = = = = =

0.73, 1.78, 0.23, 5.69, 2.35, 2.76, 3.14,

p p p p p p p

= = = < = = =

ns ns ns .01 0.02 0.01 0.005

7.9 9.7 9.1 9.4

11.1 11.7 11.4 11.5

t(43) t(43) t(43) t(43)

= = = =

2.38, 2,24, 1.83, 1.72,

p p p p

= = = =

0.02 0.03 0.07 0.09

(3.4) (2.1) (2.6) (3.5)

(3.4) (2.2) (2.6) (2.3)

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Standard deviations are given in parentheses. T-scores and p-values from independent samples t-tests comparing the two groups are reported in the last column.

the trail making tests (Reitan & Wolfson, 1993). Attention span and working memory were measured with forward and backward digit span. Language was tested by the WAISIV test (Wechsler adult intelligence scale – IV, Wechsler, 2008), while for verbal memory and visual memory the CVLT-II (California Verbal Learning Test – II, Delis et al., 1987), the RCF test (Rey Complex Figure Test and Recognition Trial, Meyers & Meyers, 1995) and the Vienna-continuous recognition task were used. Finally, executive functions were tested by the Tower test (D-Kefs) and the verbal fluency (D-Kefs) tests. Demographics for all participants, as well as neuropsychological data, are shown in Table 1. The two groups did not differ significantly in terms of mean age, education and MMSE scores, but they differed in all other cognitive measurements with the exception of Category Switching and Switch Accuracy which showed a trend towards significance. Materials Materials comprised three groups of words violating certain constraints of word formation in Slovenian (2–4), one group of unattested, well-formed, possible words without violations (5), one group of real words (6) and one group of non-words (1). All were formed with a masculine-gender nominal -ec suffix. Materials were selected based on a normative study for Slovenian (Marjanovič et al., 2013). The stimulus set comprised the following word categories: (1) Non-words (NWs) based on nonexisting stems and existing suffixes (*lastje, *dovina, ‘katakt-er’) (n = 30) (2) PseudoWs violating grammatical category (CatViol) constraints of the base (*črkilec ‘letter-er’) (n = 30) (3) PseudoWs violating thematic constraints (ThemViol) of the base (*trpelec ‘sufferer’) (n = 22) (4) PseudoWs violating aspectual constraints (AspViol) of base (*preplavalec (from preplavati ‘to swim-perfective’)) (n = 30) (5) Possible unattested Ws without violations (*kuhalec (possible but blocked by kuhar ‘cook’)) (n = 30) (6) Real words (plavalec (‘swimmer’) (n = 30)

CLINICAL LINGUISTICS & PHONETICS

9

The stimulus materials also comprised 60 real words as fillers, which were not included in the analysis. Thus, the ratio between well-formed and ill-formed words was 1:11. All stems of pseudo-words were matched on average for frequency by using the FidaPlus corpus and the actual pseudo-words were also matched for length and number of syllables.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Procedure Two studies were conducted, a grammaticality judgment task and an on-line lexical decision task. Participants were tested on the same day on both tasks. First, they performed the on-line task, followed by parts of the neuropsychological evaluation and then the off-line task. The on-line lexical decision task was run on an IBM computer using E-prime professional. Stimuli (a list of words and pseudo-words – see Materials) were presented at the centre of a computer screen in black font on a white background and were randomized for each participant. Each item was preceded by a row of hashmarks that remained on the screen for 200 ms, and a pause of 150 ms. Participants had to decide as quickly as possible and as accurately as possible whether or not the word that appeared on the screen was a word of their language. Participants had unlimited time to press either the YES or the NO key. The main experiment was preceded by a practice period (which included 10 items) during which participants familiarized themselves with the task and had the chance to ask questions. The same materials were used in the acceptability judgment task. Participants were presented with the list of words and pseudo-words in written form, and for each item, they had to say “yes” or “no”, answering the question “In your opinion, is this word part of Slovenian vocabulary?”. If their answer was “yes,” they were also asked to provide the meaning of this word to ensure that they were not responding randomly.

Analysis and results Mean percentages of correct responses and mean correct RTs in ms were analysed using a 6 (word type: word, non-word, CatViol, ThemViol, AspViol, Possible) × 2 (group: MCI versus age-matched controls) mixed model analysis of variance (ANOVA). Post-hoc comparisons were evaluated using independent samples t-tests with Bonferroni corrections. Subsequent analyses were performed where necessary.

Table 2. Off-line acceptability task: Mean accuracy (with standard deviations) and statistical significance for the between groups comparison. MCI Control

Non-words 94 (7.2) 98 (2.9) t(42) = −1.6 p = 0.11

Categorical 94 (17.1) 98 (4.0) t(42)=−1.78 p = 0.10

Thematic 79 (23.8) 92 (12.2) t(42)=−2.08 p = 0.08

Aspectual 63 (29.6) 80 (22.2) t(42)=−1.95 p = 0.06

Possible 52 (34.4) 76 (27.0) t(42)=−1.62 p = 0.01

Real words 98 (2.9) 98 (2.7) t(42)=−2.2 p = 0.82

10

C. MANOUILIDOU ET AL.

120

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Percentages of YES responses

100

80 MCI

60

Controls 40

20

0 Non-words Categorial Thematic Aspectual Possible

Words

Figure 1. Percentages of YES responses in the off-line task with standard error bars.

Off-line task Percentages of correct responses (“Yes” for words, “No” for all pseudo-words) were calculated for each group of words for MCI individuals and age-matched controls and are displayed in Table 2. An ANOVA revealed a main effect of word type (F(1,42) = 44.82, p < 0.001, ηp2 = 0.52), a main effect of group (F(1,42) = 6.00, p = 0.02, ηp2 = 0.13) and no significant interaction (F(1,42) = 2.95, p = 0.09, ηp2 = 0.06). Subsequent post-hoc comparisons showed that patients’ data were similar to controls’ with the exception of possible words (t(42) = −1.62, p = 0.01), and pseudo-words with aspectual violations where there is a tendency for MCI patients to accept them as words of Slovenian more often than the controls (t(43) = −1.95, p = 0.057) All other word types did not differ significantly across groups. Figure 1 demonstrates how many times participants responded YES to the question ‘is this a word of your language?’. Both groups followed the same pattern of accuracy and that there is a continuum from non-words to real words, with thematic and aspectual violations lying in the middle together with possible words and with distinct patterns of acceptance from each other violation.

On-line task Accuracy Mean percentages of correct responses are shown in Table 3. ANOVA revealed a main effect of word type (F(1,42) = 22.04, p < 0.001, ηp2 = 0.74) but no effect of group (F(1,42) =

Table 3. On-line task: Mean accuracy (with standard deviations). MCI Control

Non-words

Categorial

Thematic

Aspectual

Possible

Real words

90 (8.8) 94 (7.8)

87 (13.1) 89 (8.8)

61 (27.5) 68 (20.9)

45 (30.7) 50 (30.9)

43 (31.2) 55 (27.2)

97 (7.5) 99 (1.2)

CLINICAL LINGUISTICS & PHONETICS

11

Table 4. On-line task: Mean RTs results for both groups of participants on each type of word (with standard deviations). MCI Control

Non-words 1506 (390) 1057 (130) t(42)=−2.5 p = 0.003

Categorial 1684 (231) 1039 (199) t(42)=−3.9 p = 0.001

Thematic 1643 (322) 1194 (198) t(42)=−2.63 p = 0.005

Aspectual 1843 (360) 1450 (190) t(42)=−2.34 p = 0.007

Possible 1713 (330) 1358 (337) t(1,42)=−2.26 p = 0.02

Real words 1282 (211) 960 (90) t(42)=−1.3 p = 0.07

Table 5. Correlations between MCI individuals’ performance on executive function tests and their mean RTs per word category.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Mean RTs per word type Nonword Categorial Thematic Aspectual Possible Real words

Tower test

Letter fluency

Category fluency

r(21) = 0.05 r(21) = −0.50* r(21) = −0.57** r(21) = 0.08 r(21) = −0.47* r(21) = −0.50* r(21) = −0.13 r(21) = −0.49* r(21) = −0.40* r(21) = −0.19 r(21) = −0.37* r(21) = −0.32* r(21) = −0.18 r(21) = −0.31, p = 0.06 r(21) = −0.35* r(21) = −0.06 r(21) = −0.48* r(21) = −0.43*

Category switching

Switch category

r(21) = −0.23 r(21) = −0.25 r(21) = −0.43* r(21) = −0.59** r(21) = −0.20 r(21) = −0.28

r(21) = −0.20 r(21) = −0.24 r(21) = −0.41* r(21) = −0.52* r(21) = −0.18 r(21) = −0.19

*Correlation is significant at the 0.05 level. **Correlation is significant at the 0.01 level.

0.91, p = 0.34), and no significant word type × group interaction (F(1,42) = 0.69, p = 0.41, ηp2 = 0.08). Reaction times As for response times, mean correct RTs in ms are shown in Table 4. Analysis revealed a main effect of word type (F(5,38) = 119.023, p < 0.000, ηp2 = 0.08), a main effect of group (F(1,38) = 811.86, p < 0.000, ηp2 = 0.11) and a significant (word type × group) interaction (F(5,38) = 12.94, p < 0.000, ηp2 = 0.01). The MCI group was significantly slower than their age-matched controls. The high RTs of the MCI group on most word types, especially on thematic and aspectual violations, negatively correlate with their performance on executive function tests, with the exception of the Tower test. Pearson’s r correlations are reported in Table 5. Looking into each group separately, ANOVA revealed significant differences between each group of words for the age-matched controls, while the same differences did not reach significance for most of the word categories for the MCI group. Specifically for the age-matched group, ANOVA showed a main effect of violation type (F(5,25) = 89.28, p <0.001, ηp2 = .12). Post-hoc analyses using LSD (which adjusts for multiple comparisons) indicated a lack of significant differences between non-words and categorial violations (p = 0.517), but a significant difference between categorial and thematic violations (p <0.001), thematic and aspectual violations (p < .001) and aspectual violations and possible words (p = .002). As for the MCI group, ANOVA showed a main effect of violation type (F(5,25) = 51.94, p > 0.001, ηp2 = 0.07). Post-hoc analysis using LSD indicated that AspViol seemed to make a difference, since this pseudo-word category appears to differ significantly from both possible words (p = 0.001) as well as from ThemViol (p < 0.000) and real words differed significantly from all types of pseudo-words but not from non-words (p = 0.15). There are no significant differences between any other word type. However, if we look at Figure 2, we see that both groups of participants follow the same pattern of behaviour

12

C. MANOUILIDOU ET AL.

Mean correct Response Times (ms)

2500 2000 1500 MCI 1000

Control

500 0 Non-words

Categorial

Thematic

Aspectual

Possible

Real Words

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Figure 2. RTs for MCI and controls in the six types of violations and real words.

(with the exception of CatViol), suggesting that the process of evaluation of each pseudoword group was similar, albeit significantly slower for MCI patients. Finally, an interesting observation derives from the comparison between the accuracy data obtained from the two tasks both for MCI patients and for controls. Namely, a 6 (word type) × 2 (group) × 2 (task) mixed model ANOVA showed a main effect of task (F (1,42) = 5.73, p = 0.02, ηp2 = 0.12) and a word type × task interaction (F(1,42) = 4.1, p = 0.01, ηp2 = 0.08). Paired samples t-tests showed that MCI patients accepted significantly more pseudo-words with aspectual violations (t(22) = −4.353, p < 0.000), thematic violations, (t(22)= −5.247, p < 0.000), categorial violations (t(22) = −2.512, p = 0.05) and possible words (t(22) = −2.773, p = 0.05) in the on-line task than in the off-line, revealing an important task effect in the processing of pseudo-words but not in the processing of real words and non-words. Similarly, age-matched controls accepted significantly more pseudo-words with categorical violations (t(20)= 3.677, p = 0.003), thematic violations (t(20) = 5.277, p < 0.001), aspectual violations (t(20) = 5.637, p < 0.001) and possible words (t(20) = 4.947, p > 0.001), in the on-line task.

Discussion The aim of the study was to investigate the knowledge of morphological rules in Slovenian by MCI individuals and the ability to use them in detecting violations in morphologically complex word formations. Given that MCI individuals have been reported to have deficits in executive function, we further explored the option that executive function impairment would interfere with the delineation of these morphological combinations. We conducted two studies aiming at examining two levels of knowledge, namely metalinguistic knowledge (acceptability judgment task) and automatic lexical access (lexical decision task). While acceptability judgment task provided evidence about the knowledge of rules, the nature of the lexical decision task and its post-access strategic component in combination with the nature of the stimulus set (various types of word-formation violations) was more informative about the involvement of other cognitive resources, i.e. executive functions during word recognition. There were three main findings of the study. First of all, acceptability judgments showed that patients’ performance was strikingly similar to that of age-matched controls, an indication that MCI did not interfere with their ability to detect violations, which

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

13

further suggests the preservation of word-formation rules. However, the on-line task brought to light two main differences between the two populations revealing that under time pressure, MCI individuals fail to behave as healthy participants do. First, they were overall much slower than their age-matched controls. In other words, MCI individuals perform within normal range when there is no time pressure, but do worse under time pressure, suggesting a reduced speed of processing. Second, while age-matched controls differentiate between each type of pseudo-word, revealing the same steps of derivation as Greek-speaking and English-speaking participants did (Manouilidou & Stockall, 2014), RT differences among conditions in MCI individuals did not reach significance, suggesting that MCI individuals do not differentiate between these word types under time pressure. However, a noticeable pattern nonetheless exists in MCI data, similar to that found for age-matched controls (see Figure 2). Namely, RTs for each violation went in the same direction as age-matched controls, albeit much slower. This, together with the grammaticality judgment results (Figure 1), argues that MCI individuals have not lost the knowledge of word-formation rules or the ability to use them. In contrast, what appears to be highly problematic is the slow processing speed which obscures the differences and a delay in detecting violations induced by the inappropriate combination between stem and suffix. Interestingly enough, this delay correlates with their performance in the executive function tasks (Table 5), pointing towards a general inability to detect and efficiently resolve conflict. This is in accordance with previous studies on MCI individuals showing an impairment in cognitive and language tasks because of an executive dysfunction (Azuma et al., 2013; Belleville et al., 2008; Brandt et al., 2009). As stated in the Introduction, executive functions mainly affect two ways of human behaviour, organization and regulation (Miller & Cohen, 2001). First, at the level of organization, the individual gathers all available information and structures it for evaluation. Then, he/she estimates the current situation and responds accordingly. Executive functions regulate mental activity in order to enable individuals to evaluate and select between appropriate and inappropriate information during tasks. It is in this way that executive function has been found to mediate sentence comprehension, in the sense that it enables the individual to override initial misinterpretations and to choose among conflicting syntactic representations (Novick et al., 2005). As shown in a previous section, morphological processing can lead to initial misinterpretations (e.g. corner = corn+-er). The role of executive control is to enable the individual to revise an early parsing commitment after encountering new evidence. This is precisely what is needed in order to realize that word formations such as *črkilec ‘letter-er’, *trpelec ‘sufferer’, *preplavalec ‘swimmer-perfective’ each present with a different type of conflict that needs to be resolved in order to make the lexical decision and press the YES or NO button. In case of categorial violations (*črkilec ‘letter-er’), the mismatch is very robust and easily detectable as it emerges because of the inappropriate lexical category of the stem and, thus, it is easily resolved before it creates a conflicting representation. The cases of thematic (*trpelec ‘sufferer’) and aspectual violations (*preplavalec ‘swimmer-perfective’) are far more complicated, and they are the only type of violations that correlate with category switching and switch category, the most demanding executive function tasks in terms of cognitive flexibility. In category switching and switch category, participants have to shift their attention between one task and another. The tasks bring into light participants’ ability to rapidly and efficiently adapt to different situations. When it comes to

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

14

C. MANOUILIDOU ET AL.

aspectual and thematic violations, the conflict between competing representations is much greater compared, for instance, to categorial violations. That is, participants have to shift their attention and make a choice between a word that can potentially be interpreted (e.g. *preplavalec ‘swimmer’-perfective) and the already established representation of the word (plavalec –’swimmer’). There is no pragmatic/word knowledge reason to disallow preplavalec; thus the processor, upon encounter, builds up a representation of “*preplavalec” which immediately after has to compete with innate linguistic knowledge that –ec selects imperfective verbal bases and also with the existing representation of plavalec. Thus, participants have to shift attention and decide between world knowledge appropriateness and linguistic knowledge non-appropriateness, in a comparable way to their attention shift in switch category tasks. Similarly, a word with a thematic violation such as *trpelec ‘suffer-er’ is also interpretable and the processor builds up the representation of “someone who suffers” which is in conflict with innate linguistic knowledge and prototypical representations of words formed with –ec (requirement of agentive verbal base) making participants wiggle between the two and ponder upon encountering the word. A similar conflict does not arise with other types of violations, such as categorial violations (črkilec ‘letter-er’), simply because it is not possible to build any kind of representation and the word is rapidly and easily rejected. Taking the above into consideration, the results of the current study point towards the involvement of executive functions in processing these morphologically complex words. That is, we assume that MCI individuals are able to isolate the two morphemes that participated in each derived pseudo-word (as age-matched controls did), but they were much slower in detecting the conflict in the inappropriate word formations, especially in the cases of *trpelec ‘sufferer’ and *preplavalec ‘swimmer-perfective’. An anonymous reviewer suggests a different view of the data according to which MCI individuals sacrifice speed for accuracy. This is a valid concern especially given the fact that the lexical decision is dependent on decision criteria which could create speedaccuracy trade-offs. In other words, one could postulate that MCI individuals follow a strategy according to which they opt to be slower in order to be more accurate. Indeed, while there are significant differences in accuracy between off-line and on-line results, there are no differences between MCI and controls in accuracy in the on-line task. However, as one can see, age-matched controls show exactly the same difference as MCI participants do between accuracy patterns in off-line and on-line task (see Results). Thus, if MCI individuals adopted this strategy, then we would have to admit that agematched controls adopted it too. Yet, the difference between the two groups in terms of RTs remains highly significant in every word category. We take this difference to be indicative of a processing slowdown associated with the cognitive impairment in the MCI population. Besides, differences in processing times in lexical decision tasks between agematched and MCI individuals have been reported elsewhere in the literature (e.g. Taler & Jarema, 2006). A third option to explore is that MCI participants are not slower in detecting violations, but they are slower in decomposing morphologically complex items2. Given that this is the first linguistic study in MCI to use complex lexical items, there is no evidence from previous investigations on this topic. However, since morpheme recognition tends to be done automatically (e.g. Rastle & Davis, 2008) and at earlier stages of lexical access (e.g. Morris & Stockall, 2013) in healthy populations, there is no reason to assume that this

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

15

fundamental property of the human processor is seriously compromised in a condition which is characterized by mild language problems, such as MCI. Yet, this is an issue that calls for further investigation. Moreover, if the only difficulty was with decomposition, then we would not observe the same pattern as controls in their RTs (see Figure 2), and most importantly we would not observe significantly higher RTs for the aspectual violations and also faster RTs for the real words. Instead, we would observe flat RTs for all stimuli. Thus, we are left with the conclusion that MCI individuals are slower at detecting violations and dealing with the multidimensional processing of complex words violating various combinatorial properties of stem+affix than healthy-matched controls. Given that the unacceptability of the pseudo-words used in the current study relies on syntactic, thematic (argument structure) and aspectual properties, we cannot make any claims about any of these domains being more impaired than the others. However, an interesting finding concerns aspectual violations (*preplavalec), the condition that elicited significantly higher RTs. Aspect is a verbal property which describes how an event unfolds in time and distinguishes between perfective and imperfective events (Comrie, 1976). A perfective event description reflects the speaker’s choice to see the event as completed, e.g. the water boiled while an imperfective event description reflects a speaker’s choice to see the event as in progress, e.g. the water was boiling. Thus, on these grounds, aspect is considered to be a subjective category (Comrie, 1976; Smith, 1997). This becomes clear when compared to tense which calls for objectivity, since an event either takes place in the present (the water is boiling) the future (the water will boil) or the past (the water boiled), regardless of the speaker’s perspective on the event. Given its particularities, the processing of aspect requires the integration of linguistic and extra linguistic knowledge, and it has been found to be impaired in other cognitively impaired populations, such as Alzheimer’s patients (Fyndanis et al., 2013). The fact that aspectual violations are by far the slowest and least accurate category of violations detected in the current experiments could indicate a difficulty by MCI individuals to detect this subtle feature of the verbal base. In conclusion, the study provides evidence for MCI patients’ ability to detect violations of word formation in Slovenian which require knowledge of multiple linguistic levels, namely the syntactic category of the stem, the argument structure properties of base verb and the event aspectual specifications. Thus, the evidence suggests that morphological rule knowledge is preserved in MCI. However, the processing times of these violations are indicative of a deficit which appears to rely on the extra-linguistic cognitive system and points to an executive dysfunction, a common problem with MCI patients. Even though, thus far, morphological processing has not been used in the literature as requiring the involvement of executive functions, our initial assumption that MCI individuals might have difficulties resolving the conflict that arises from the inappropriate combination of stem and affix in pseudo-words due to an executive dysfunction appears to be supported by the data and by correlation analyses. If this is indeed the case, then we are dealing with another case of individuals with cognitive decline where an executive dysfunction is responsible for compromised linguistic behaviour, in our case morphological processing. The present results are consistent with the theory that executive dysfunction in MCI can affect conscious semantic decisions (Duong et al., 2006). Finally, the different performance in the grammaticality judgment task and the lexical decision task calls for a task effect

16

C. MANOUILIDOU ET AL.

which is suggestive for the potential that on-line studies provide for the detection of risk groups for dementia.

Acknowledgements Preliminary results were presented at the 14th International Conference of the Science of Aphasia in Brussels, 2013. The authors wish to thank the audience of the meeting, Linnaea Stockall as well as two anonymous reviewers for useful comments and suggestions. Many thanks are also due to Anka Slana and Maja Kraigher for help with testing patients and age-matched controls.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Declaration of interest The authors report no conflict of interest.

Funding The research reported here was partly funded by a grant from ARRS (L3-4255) – Slovenian Research Agency awarded to Zvezdan Pirtošek. Notes 1. Possible words are well-formed, albeit non-attested. 2. We are grateful to an anonymous referee for this suggestion.

References Albert, M., Blacker, D., Moss, M. B., Tanzi, R., & McArdle, J. J. (2007). Longitudinal change in cognitive performance among individuals with mild cognitive impairment. Neuropsychology, 21, 158–169. Azuma, T., Sabbagh, M. N., & Connor, D. J. (2013). The effect of healthy aging and mild cognitive impairment on semantic ambiguity detection. Journal of Neurolinguistics, 26, 271–282. Bedny, M., Hulbert, J. C., & Thompson-Schill, S. L. (2007). Understanding words in context: The role of Broca’s area in word comprehension. Brain Research Special Issue Mysteries of Meaning, 1146, 101–114. Belanger, S., & Belleville, S. (2009). Semantic inhibition impairment in mild cognitive impairment: A distinctive feature of upcoming cognitive decline? Neuropsychology, 23.5, 592. Belleville, S., Bherer, L., Lepage, E., Chertkow, H., & Gauthier, S. (2008). Task switching capacities in persons with Alzheimer’s disease and mild cognitive impairment. Neuropsychologia, 46, 2225– 2233. Brandt, J., Aretouli, E., Neijstrom, E., Samek, J., Manning, K., Albert, M. S., et al. (2009). Selectivity of executive function deficits in mild cognitive impairment. Neuropsychology, 23, 607–618. Burani, C., Dovetto, F. M., Spuntarelli, A., & Thornton, A. M. (1999). Morpholexical access and naming: The semantic interpretability of new root–suffix combinations. Brain and Language, 68, 333–339. Chertkow, H. (2002). Mild cognitive impairment. Current Opinion in Neurology, 15, 401–407. Collie, A., Maruff, P., & Currie, J. (2002). Behavioral characterization of mild cognitive impairment. Journal of Clinical and Experimental Neuropsychology, 24, 721–733. Comrie, B. (1976). Aspect. New York, NY: Cambridge University Press.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

CLINICAL LINGUISTICS & PHONETICS

17

Davie, J. E., Azuma, T., Goldinger, S. D., Connor, D. J., Sabbagh, M. N., & Silverberg, N. B. (2004). Sensitivity to expectancy violations in healthy aging and mild cognitive impairment. Neuropsychology, 18, 269–275. De Jager, C. A., Hogervorst, E., Combrinck, M., & Budge, M. M. (2003). Sensitivity and specificity of neuropsychological tests for mild cognitive impairment, vascular cognitive impairment and Alzheimer’s disease. Psychological Medicine, 33, 1039–1050. Delis, D. C., Kramer, J. H., Kaplan, E., & Ober, B. A. (1987). California verbal learning test psychological corporation. San Antonio, TX: Psychological Corporation. Duong, A., Whitehead, V., Hanratty, K., & Chertkow, H. (2006). The nature of lexico-semantic processing deficits in mild cognitive impairment. Neuropsychologia, 44, 1928–1935. Elliot, R. (2003). Executive functions and their disorders. British Medical Bulletin, 65, 49–59. Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). ‘‘Mini-mental state’’: A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12, 189–198. Fruchter, J., & Marantz, A. (2015). Decomposition, lookup, and recombination: MEG evidence for the full decomposition model of complex visual word recognition. Brain and Language, 143, 81–96. Fruchter, J., Stockall, L., & Marantz, A. (2013). MEG masked priming evidence for form-based decomposition of irregular verbs. Frontiers in Human Neuroscience, 22(7), 798. Fyndanis, V., Manouilidou, C., Koufou, E., & Tsapakis, E.-M. (2013). Agrammatic patterns in Alzheimer’s disease: Evidence from tense, agreement and aspect. Aphasiology, 27(2), 178–200. Hodges, J. R., Erzinçlioglu, S., & Patterson, K. (2006). Evolution of cognitive deficits and conversion to dementia in patients with mild cognitive impairment: A very long-term follow-up study. Dementia and Geriatric Cognitive Disorders, 21, 380–391. Hoffman, P., Jefferies, E., and Lambon Ralph, M.A. (2010). Ventrolateral prefrontal cortex plays an executive regulation role in comprehension of abstract words: Convergent neuropsychological and rTMS evidence. Journal of Neuroscience, 46, 15450–15456. Hussey, E. K., & Novick, J. M. (2012). The benefits of executive control training and the implications for language processing. Frontiers in Cognition, 3(158), 1–14. Hussey, E., Teubner-Rhodes, S., Dougherty, M., Bunting M., & Novick, J. (2010). Improving gardenpath recovery in healthy adults through cognitive control training. Talk Presented at the 16th Annual Conference on Architectures and Mechanisms for Language Processing, York, UK. Joubert, S., Brambati, S. M., Ansado, J., Barbeau, E. J., Felician, O., Didic, M., et al. (2010). The cognitive and neural expression of semantic memory impairment in mild cognitive impairment and early Alzheimer’s disease. Neuropsychologia, 48, 978–988. Kan, I. P., & Thompson-Schill, S. L. (2004). Effect of name agreement on prefrontal activity during overt and covert picture naming. Cognitive, Affective, & Behavioral Neuroscience, 4, 43–57. Korpus slovenskega jezika FidaPLUS. Retrieved: 30.11.2012, from http://www.fidaplus.net/. Lambon Ralph, M. A., Patterson, K., Graham, N., Dawson, K., & Hodges, J. R. (2003). Homogeneity and heterogeneity in mild cognitive impairment and Alzheimer’s disease: A cross-sectional and longitudinal study of 55 cases. Brain, 126, 2350–2362. Manouilidou, C. (2006). On the Processing of Thematic Features of Deverbal Nominals. PhD Thesis, University of Ottawa, Canada. Manouilidou, C. (2007). Thematic constraints in deverbal word formation: Psycholinguistic evidence from pseudo-words. Proceedings of the 7th International Conference on Greek Linguistics, University of York, UK. On-line publication, available at: http://icgl7.icte.uowm.gr/english_ papers.htm Manouilidou, C., Kordouli, K., Papanagiotou, A., Messinis, L., & Papathanassopoulos, P. (2014). Lexical-semantic deficits in mild cognitive impairment: The case of abstract vs. concrete nouns. Stem-, Spraak-en Taalpathologie, 19(S01), 92–95. Manouilidou, C., & Stockall, L. (2014). Teasing apart syntactic category vs. argument structure information in deverbal word formation: A comparative psycholinguistic study. Italian Journal of Linguistics, 26(2), 71–98.

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

18

C. MANOUILIDOU ET AL.

Marjanovič, K., Manouilidou, C., & Marvin, T. (2013). Word-formation rules in Slovenian agentive deverbal nominalization: A psycholinguistic study based on pseudo-words. Slovenski jezik – Slovene Linguistic Studies, 9, 93–109. Marvin, T. (2002). Topics in the stress and syntax of words. Cambridge, MA: MIT Working Papers in Linguistics. Miller, E. K., & Cohen, J. D. (2001). An integrative theory of prefrontal cortex function. Annual Review of Neuroscience, 24, 167–202. Meyers, J. E., & Meyers, K. R. (1995). Rey complex figure test and recognition trial: Professional manual. Lutz, FL: Psychological Assessment Resources. Morris, J., & Stockall, L. (2013). Early, equivalent ERP masked priming effects for regular and irregular morphology. Brain and Language, 123(2), 81–93. Murphy, K. J., Rich, J. B., & Troyer, A. K. (2006). Verbal fluency patterns in amnestic mild cognitive impairment are characteristic of Alzheimer’s type dementia. Journal of the International Neuropsychological Society, 12, 570–574. Nasreddine, Z. S., Phillips, N. S., Bédirian, V., Charbonneau, S., Whitehead, V., Collin, I., et al. (2005). The Montreal cognitive assessment, MoCA: A brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society, 53(4), 695–699. Novick, J. M., Trueswell, J. C., & Thompson-Schill, S. L. (2005). Cognitive control and parsing: Reexamining the role of Broca’s area in sentence comprehension. Cognitive, Affective, & Behavioral Neuroscience, 5, 263–281. Novick, J. M., Hussey, E., Teubner-Rhodes, S., Harbison, J. I., & Bunting, M. F. (2014). Clearing the garden-path: Improving sentence processing through cognitive control training. Language, Cognition and Neuroscience, 29(2), 186–217. Novick, J. M., Kan, I. P., Trueswell, J. C., & Thompson-Schill, S. L. (2009). A case for conflict across multiple domains: Memory and language impairments follow damage to ventrolateral prefrontal cortex. Cognitive Neuropsychology, 26(6), 527–567. Olichney, J. M., Morris, S. K., Ochoa, C., Salmon, D. P., Thal, L. J., Kutas, M., et al. (2002). Abnormal verbal event related potentials in mild cognitive impairment and incipient Alzheimer’s disease. Journal of Neurology, Neurosurgery and Psychiatry, 73, 377–384. Petersen, R. C., Smith, G. E., Waring, S. C., Ivnik, R. J., Tangalos, E. G., & Kokmen, E. (1999). Mild cognitive impairment: Clinical characterization and outcome. Archives of Neurology, 56, 303–308. Petersen, R. C., Doody, R., Kurtz, A., Mohs, R. C., Morris, J. C., Rabins, P. V., et al. (2001). Current concepts in mild cognitive impairment. Archives of Neurology, 58, 1985–1992. Petersen, R. C. (2003). Mild cognitive impairment. New York: Oxford University Press. Puregger, E., Walla, P., Deecke, L., & Dal-Bianco, P. (2003). Magnetoencephalographic features related to mild cognitive impairment. NeuroImage, 20, 2235–2244. Rastle, K., Davis, M. H., & New, B. (2004). The broth in my brother’s brothel: Morpho-orthographic segmentation in visual word recognition. Psychonomic Bulletin & Review, 11, 1090–1098. Rastle, K., & Davis, M. H. (2008). Morphological decomposition based on the analysis of orthography. Language and Cognitive Processes, 23, 942–971. Reitan, R. M., and D. Wolfson. (1993). Theoretical, Methodological and Validational Bases of the Halstead-Reitan Neuropsychological Test Battery. Reitan Neuropsychology Laboratory. Robinson, G., Blair, J., & Cipolotti, L. (1998). Dynamic aphasia: An inability to select between competing verbal responses? Brain, 121, 77–89. Royle, P., Drury, J. E., Bourguignon, N., & Steinhauer, K. (2012). The temporal dynamics of inflected word recognition: A masked ERP priming study of French verbs. Neuropsychologia, 50(14), 3542–3553. Schnur, T. T., Schwartz, M. F., Kimberg, D. Y., Hirshorn, E., Coslett, H. B., & Thompson-Schill, S. L. (2008). Localizing interference during naming: Convergent neuroimaging and neuropsychological evidence for the function of Broca’s area. Proceedings of the National Academy of Sciences, 106(1), 322–327. Smith, C. S. (1997). The parameter of aspect (2nd ed.). Dordrecht: Kluwer Academic Publishers. Stramljič Breznik, I. (1999). Prispevki iz slovenskega besedoslovja. Maribor: Slavistično društvo Maribor.

CLINICAL LINGUISTICS & PHONETICS

19

Downloaded by [Christina Manouilidou] at 09:42 24 November 2015

Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology 18.6, 643. Taler, V., & Jarema, G. (2004). Processing of mass/count information in Alzheimer’s disease and mild cognitive impairment. Brain and Language, 90, 262–275. Taler, V., & Jarema, G. (2006). On-line lexical processing in AD and MCI: An early measure of cognitive impairment? Journal of Neurolinguistics, 19, 38–55. Taler, V., & Phillips, N. A. (2008). Language performance in Alzheimer’s disease and mild cognitive impairment: A comparative review. Journal of Clinical and Experimental Neuropsychology, 30, 501–556. Traykov, L., Raoux, N., Latour, F., Gallo, L., Hanon, O., Baudic, S., et al. (2007). Executive functions deficit in mild cognitive impairment. Cognitive and Behavioral Neurology, 20, 219–224. Wang, P. N., Lirng, J. F., Lin, K. N., Chang, F. C., & Liu, H. C. (2006). Prediction of Alzheimer’s disease in mild cognitive impairment: A prospective study in Taiwan. Neurobiology of Aging, 27, 1797–1806. Wechsler, D. (2008). Wechsler adult intelligence scale - IV. San Antonio: Pearson, Inc.