The heterogeneity of verbal short-term memory impairment in aphasia

Author’s Accepted Manuscript The heterogeneity of verbal short-term memory impairment in aphasia Steve Majerus, Lucie Attout, Marie-Amélie Artielle, Marie-Anne Van der Kaa www.elsevier.com/locate/neuropsychologia

PII: DOI: Reference:

S0028-3932(15)30126-3 http://dx.doi.org/10.1016/j.neuropsychologia.2015.08.010 NSY5694

To appear in: Neuropsychologia Received date: 8 May 2015 Revised date: 9 August 2015 Accepted date: 10 August 2015 Cite this article as: Steve Majerus, Lucie Attout, Marie-Amélie Artielle and Marie-Anne Van der Kaa, The heterogeneity of verbal short-term memory impairment in aphasia, Neuropsychologia, http://dx.doi.org/10.1016/j.neuropsychologia.2015.08.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

THE HETEROGENEITY OF VERBAL SHORT-TERM MEMORY IMPAIRMENT IN APHASIA

Short title: VERBAL STM IN APHASIA

Steve Majerus123, Lucie Attout1, Marie-Amélie Artielle1, & Marie-Anne Van der Kaa4

1

Department of Psychology, Cognition & Behavior, Université de Liège 2

Cyclotron Research Center, Université de Liège

3

Fonds de la Recherche Scientifique FNRS, Belgium

4

Centre Hospitalier Universitaire, Université de Liège

Address for correspendence: Dr. Steve Majerus Department of Psychology – Cognition & Behavior, Université de Liège Boulevard du Rectorat, B33, 4000 Liège – Belgium Tel : 0032 4 3664656 Email : [email protected]

ABSTRACT Verbal short-term memory (STM) impairment represents a frequent and long-lasting deficit in aphasia, and it will prevent patients from recovering fully functional language abilities. The aim of this study was to obtain a more precise understanding of the nature of verbal STM impairment in aphasia, by determining whether verbal STM impairment is merely a consequence of underlying language impairment, as suggested by linguistic accounts of verbal STM, or whether verbal STM impairment reflects an additional, specific deficit. We investigated this question by contrasting item-based STM measures, supposed to depend 1

strongly upon language activation, and order-based STM measures, supposed to reflect the operation of specific, serial order maintenance mechanisms, in a sample of patients with single-word processing deficits at the phonological and/or lexical level. A group-level analysis showed robust impairment for both item and serial order STM aspects in the aphasic group relative to an age-matched control group. An analysis of individual profiles revealed an important heterogeneity of verbal STM profiles, with patients presenting either selective item STM deficits, selective order STM deficits, generalized item and serial order STM deficits or no significant STM impairment. Item but not serial order STM impairment correlated with the severity of phonological impairment. These results disconfirm a strong version of the linguistic account of verbal STM impairment in aphasia, by showing variable impairment to both item and serial order processing aspects of verbal STM.

Keywords: aphasia; short-term memory

INTRODUCTION Verbal short-term memory (STM) impairment is frequently observed in patients with aphasia (e.g., Caramazza, Basili, Koller, & Berndt, 1981; Francis, Clark, & Humphreys; 2003; Koenig-Bruhin & Studer-Eichenberger, 2007; Murray, Ballard, & Karcher, 2004). Furthermore, verbal STM deficits are long-lasting as they tend to persist even when singleword processing difficulties have resolved (Majerus, 2009; N. Martin & Saffran, 1996). Residual verbal STM impairment represents an important source of difficulties in these patients, as it will prevent efficient processing in situations where multiple verbal information has to be maintained for a limited amount of time, such as during oral conversations, sentence comprehension and planning, or processing of unfamiliar or degraded verbal information. It is 2

thus critical to understand the nature of verbal STM deficits in order to develop targeted intervention procedures. There are two possible hypotheses explaining verbal STM deficits in aphasic patients. A first one considers that the verbal STM deficits are a consequence of residual impairment in the language network, while a second hypothesis considers that verbal STM deficits are deficits on their own, caused by additional impairment to specific verbal STM processes. The present study aims at contrasting these two hypotheses in aphasic patients with single word processing difficulties, by using both group-level and case-level study designs. The first hypothesis, tying verbal STM impairment to residual language processing difficulties, is grounded in language-based accounts of verbal STM (Acheson & MacDonald, 2009; N. Martin & Ayala, 2004; N. Martin & Saffran, 1992). In these accounts, verbal STM is considered to be an emergent property of language processing, with verbal STM capacity determined directly by temporary activation of the language network. This is most clearly illustrated by the interactive activation model of language and STM processing proposed by N. Martin and Saffran (1992). In this account, the duration of activation of phonological, lexical and semantic representations in the language network is a major determinant of verbal STM capacity. Many other recent accounts of verbal STM consider that activation of language representations is, at the least, an important component of verbal STM, and this especially for the representation of item information (the phonological, lexical and semantic features of the stimuli to be maintained in verbal STM task) (Burgess & Hitch, 1999, 2006; Baddeley, Gathercole, & Papagno, 1998; Baddeley, 2000; R. Martin & Lesch, 1996). This theoretical assumption is supported by a number of studies in healthy adults consistently showing the presence of psycholinguistic effects in verbal STM, with verbal STM performance being higher for verbal items that possess richer and more easy-to-activate representations in the language knowledge base. These psycholinguistic effects in verbal STM 3

include lexicality effects, lexical frequency effects, phonological neighborhood effects, imageability effects, semantic relatedness effects, and sublexical phonotactic frequency (Gathercole, Frankish, Pickering, & Peaker, 1999; Jefferies, Frankish & Lambon Ralph, 2006; Hulme, Maughan, & Brown, 1991; Majerus, Van der Linden, Mulder, Meulemans, & Peters, 2004; Majerus, Martinez Perez, & Oberauer, 2012; Nakayama, Tanida, & Saito, 2015; Thorn & Frankish, 2005; Walker & Hulme, 1999). Similarly, patients with impaired semantic knowledge show reduced verbal STM performance especially for those words they don’t know anymore (Patterson, Graham, & Hodges, 1994; Papagno, Vernice, & Cecchetto, 2013). Functional neuroimaging studies in healthy adults have also shown that superior temporal and inferior temporal regions involved in phonological and semantic processing, respectively, show sustained activation during verbal STM tasks (Fiebach, Friederici, Smith, & Swinney, 2007; Majerus et al., 2010). Given these data, single word processing difficulties in patients with aphasia should lead to reduced verbal STM performance. This is supported by a review of brain-injured patients presenting so-called selective verbal STM impairment (Majerus, 2009). The vast majority of these patients had presented aphasic symptoms in the first weeks or months post-injury and the severity of their verbal STM deficits strongly correlated with the severity of their initial or residual single word processing difficulties (word/nonword repetition, picture naming). However, the fact that verbal STM impairment in aphasic patients is likely to result from underlying language impairment does not rule out the possibility that there are additional deficits for other, more specific processes involved in verbal STM. It is critical here to make a distinction between item processing and serial order processing in verbal STM. As already noted, the representation of item information is considered to depend, at least partially, on access to underlying language representations (Acheson & MacDonald, 2009; Burgess & Hitch, 2006; N. Martin & Saffran, 1992). On the other hand, the representation of serial order 4

information is considered by many recent verbal STM accounts to depend on specific positional marking processes (Burgess & Hitch, 1999; Brown, Preece, & Hulme, 2000; Gupta & MacWhinney, 1997; Hurlstone, Hitch, & Baddeley, 2014). Behavioral and neuropsychological studies have shown that psycholinguistic effects in verbal STM primarily affect recall of item information, but less strongly recall of order information (Majerus & D’Argembeau, 2011; Majerus, Norris, & Patterson, 2007; Nairne & Kelley, 2004; Poirier & Saint-Aubin, 1996). Also, brain imaging studies have shown that serial order processing in verbal STM is associated with neural substrates outside the language processing networks, such as the left and right intraparietal and superior parietal cortex (Attout, Fias, Salmon, & Majerus, 2014; Henson, Burgess, & Frith, 2000; Majerus et al., 2006; Marshuetz, Smith, Jonides, DeGutis, & Chenevert, 2000). A separation of item and order STM processes is also supported by a double dissociation between item retention deficits and serial order retention deficits that has recently been shown in two brain injured patients (Attout, Van der Kaa, George, & Majerus, 2012). This raises the question of the status of verbal serial order STM capacities in patients with aphasia. If verbal STM impairment is merely a direct consequence of underlying language impairment in aphasic patients, then we should expect mainly item STM deficits in aphasic patients. If verbal STM deficits are due to additional impairment of specific serial order STM processes, then we should also observe serial order STM deficits, at least in some aphasic patients, and the severity of the verbal STM deficits for serial order information should not be directly associated with the severity of STM deficits for verbal item information. These predictions are furthermore complicated by recent observations of possible multiple representational layers for serial order information. Fischer-Baum and McCloskey (2015) showed that serial order STM failures can be accounted for by inefficient positional markers, but also by deficient inter-item associations, the latter process considering that serial 5

position information is directly encoded via the updating of inter-item connections. Given that items are considered by many models of STM to be coded by the language system, this would mean that serial order information can also be encoded directly within the language system. This view is also shared by the linguistic account proposed by N. Martin and Saffran (1992, 1997), which considers that primacy and recency effects during serial recall are supported by semantic and phonological levels of activation, respectively. This is supported by recent findings showing that serial order recall can also be influenced by semantic variables under some conditions (Acheson, MacDonald & Postle, 2011; Poirier, Saint-Aubin, Mair, Tehan, & Tolan, 2015). Martin and Saffran (1997) also showed that aphasic patients with phonological impairment tend to present reduced recency effects in word/nonword repetition, while patients with semantic impairment tend to present reduced primacy effects. A recent neuroimaging study by Kalm and Norris (2014) goes in the same direction, by showing that neural activation patterns in superior temporal and temporo-parietal language processing cortices can code serial position information of syllables in nonword strings. These data, while rendering more complex the theoretical understanding of the processes involved in serial order STM, further increase the likelihood of observing serial order STM deficits in left-hemisphere damaged patients with aphasia. Importantly, however, these data would predict that item and serial order verbal STM impairments are closely associated, since the same representational levels are used to code item and serial order information in STM. The question of serial order STM impairment in aphasia, whether resulting from linguistic or specific serial order maintenance deficits, is of further importance due to its functional implications for language learning. Several studies have shown that verbal serial order STM capacity is a particularly important predictor of lexical language learning abilities in children and adults, and this independently from item STM capacity (Majerus, Poncelet, Van der Linden, & Weekes, 2008; Majerus & Boukebza, 2013). It follows that verbal serial 6

order STM capacity, if impaired in aphasic patients, is likely to negatively impact on the patients’ language recovery and relearning abilities, further stressing the need to investigate the status of serial order STM capacity in these patients. Previous studies investigating the nature of STM disorders in aphasic patients have mainly contrasted verbal and visual STM tasks, showing that STM impairment is not always restricted to verbal STM, and that impairment in visual and visuo-spatial STM tasks can be associated with the severity of language deficits (Burgio & Basso, 1997; Kasselimis et al., 2013; Lang & Quitz, 2012; N. Martin & Ayala, 2004; Mayer & Murray, 2012; Potagas et al., 2011; Seniów et al., 2009). However, these studies did not specifically address the nature of impairment in terms of item or serial order STM processes. The associated impairment in visual STM tasks and the fact that typical visual STM such as the Corsi block tapping task also require sequential processing to some extent, raise at least the possibility of associated serial order STM impairment. At the same time, at a theoretical level, the distinction between item and serial order processing in the visuo-spatial STM domain is currently less clear, and the question of the similarity of mechanisms involved in maintaining serial order information in the verbal and visuo-spatial domains is still an open question (Hurlstone et al., 2015). The present study will focus on the distinction between item and serial order processing in the verbal STM domain exclusively. In sum, the aim of this study was to provide a comprehensive and robust test of item and serial order short-term retention abilities within verbal STM in patients with fluent aphasia. According to a strong version of the linguistic account of verbal STM deficits in patients with aphasia, item and serial order STM impairment should be closely associated, and both should stem from underlying language impairment. According to a weaker version of the linguistic account of verbal STM, distinguishing separate item and serial order representational levels and processes within verbal STM, dissociations between item and 7

serial order STM deficits should be observable within verbal STM tasks, with only item STM deficits being specifically associated with the severity of underlying language impairment. We selected patients with single-word processing deficits at the phonological or lexical level, by excluding patients with speech output difficulties in order to allow for reliable assessment of performance in verbal STM tasks requiring verbal output. We administered tasks that had been shown in previous studies to differentially assess item and serial order verbal STM processes. First, we assessed item and serial order STM abilities by distinguishing item recall and order recall measures in immediate serial recall tasks for word lists, based on the proportions of item and serial order errors during recall. This procedure has been shown to reliably measure item and serial order STM capacities (Attout et al., 2012; Nairne & Kelley, 2004). In order to obtain estimates of verbal STM capacities over different STM contexts, we administered two different immediate serial recall tasks, by sampling memoranda from either closed or open stimulus sets; when sampling memoranda from a closed stimulus pool, serial order retention capacities are relatively more challenged than item retention capacities since the same items reappear across the memory lists but in different serial positions; the reverse is true when sampling memoranda from an open stimulus pool, with each item being new over the entire task. Furthermore, for the open list immediate serial recall task we contrasted high and low word imageability conditions, allowing us to assess the expected differential impact of linguistic knowledge on item and serial order recall and hence to provide further validation of the item and serial order STM measures used here (Majerus & D’Argembeau, 2011; Nairne & Kelley, 2004). A third type of task assessed serial order STM capacities in a task-specific manner, by using a serial order reconstruction task. In this task, the participants had to reconstruct the serial order of sequentially presented items; in order to reduce item processing requirements as far as possible, the items used were predictable and highly familiar (digits) and item information was available at recall (by providing cards on which the digits presented

8

in the memory list were printed) (Majerus et al., 2008). In order to provide a sensitive and exhaustive test of item and serial order STM impairment in aphasia, we used both group-level and individual case-level analysis designs. Studies contrasting different STM tasks in aphasic patients typically used a group-level analysis approach, which may have hidden important individual differences and dissociations of impairment as a function of STM tasks. A strong version of the linguistic account of STM deficits in aphasia would predict reduced item STM in every patient with impaired lexical or phonological processing abilities, and this hypothesis can only be assessed via individual-level analyses. Therefore we first conducted group-level comparisons and then contrasted each individual patient’s performance to that of the control group. In order to conduct a robust test of our research questions, we recruited large control samples in order to have a reliable estimate of the variability of performance in the general population. This is particularly important when conducting case series studies in which deficits may be overestimated when using too small and too closely matched control samples.

METHODS Patients The patient sample was comprised of 14 French-speaking patients (5 male) with a primary diagnosis of fluent aphasia. Most patients had presented a left hemisphere stroke, except for patients F, L and M who suffered from traumatic brain injury and patient I who suffered from localized brain tumor. The patients were at a subacute or chronic stage and were outpatients of the neuropsychological rehabilitation unit of the University Hospital Center of the University of Liège, Belgium. All patients had been screened for bucco-linguo-facial apraxia using an in-house test requiring the patients to carry out motor commands involving their tongue, lips and face muscles; all patients showed preserved performance except for a very

9

mild reduction in speed of command execution in patient M. All patients also had normal performance on a single phoneme repetition task, consisting in the repetition of all the phonemes of the alphabet. The patients’ language processing abilities were characterized by administering in-house developed picture naming, single nonword repetition, speech perception tasks and semantic categorization tasks (see legend of Table 2 for details). As shown in Table 2, most patients showed mild-to-severe forms of anomia in an oral picture naming task for high-, medium- and low-frequency target nouns, except for patients C, I, L and N ; nonword repetition, for bisyllabic nonwords composed of high frequency or low frequency phonotactic patterns, was impaired in most patients except patients A, B, C and I; speech perception difficulties, as assessed by a nonword minimal pair discrimination task for consonant contrasts of varying articulatory trait differences (e.g., ba-pa versus ba – fa; all syllable pairs were nonwords), were observed in all patients except patients A, B, C and G. All patients showed preserved semantic processing, as assessed by semantic categorization of pictures or synonym judgment of auditorily presented words. Digit spans ranged between 4 and 6 for forward digit span, and between 2 and 5 for backward digit span (WAIS-III, Wechsler, 1997). Some of the patients (A, B, C, F, I, J, K, N) presented associated deficits at the level of attentional and/or executive processing, as established by the TAP computerized battery of attention and executive functions; a given domain was considered impaired if performance was at least 2 standard deviations below control mean (Zimmermann & Fimm, 2007). All patients presented normal visual recognition abilities as tested by an object picture matching task and preserved number recognition abilities as tested by a number naming task. The mean age of the patients was 51.36 years (range: 37-66). All patients showed a right hand dominance, except for patient I who showed a left hand dominance and presented speech perception difficulties as a result of a right temporal lesion. Control groups 10

The main control group was comprised of 107 healthy, French speaking adults (62 female) recruited in the general population and matched for chronological age to the patient group (mean age: 52.64; range: 37 – 69). The vast majority of control participants had 12 years of education except for 12 patients (years of education > 8) and came from the same type of professional backgrounds as the patients. Given that the control group had been assembled in different testing waves, not all tasks could be administered to all control participants. This was the case for the serial order reconstruction task (see below) which was administered to 88 out of the 107 control participants (mean age: 51.89 years; range: 37 – 69) and for the open list immediate serial recall task which was administered to 58 out of the 107 control participants (mean age: 50.88 years; range: 42 – 69). Tasks Closed list immediate serial recall task The closed stimulus pool was composed of eleven two-syllable words. The words were selected to be concrete and of high frequency in order to avoid difficulties with stimulus identification in our patients. The words contained four or five phonemes, they were all nouns and word frequency ranged between 64 and 104 (New, Pallier, Brysbaert, & Ferrand, 2004). Word sequences were generated by randomly sampling from the stimulus set. The stimuli were presented in 40 sequences of 5 or 6 words1 in the center of the screen of a mobile workstation, each word being presented for 1250 ms. After the final word of each sequence, a question mark appeared, requiring the participants to recall all the words in the same order as during presentation. The instructions given were: “Recall all the words in the same order as 1

For patients E, F, G and H as well as for a subsample of the control group (N=18), 6-word lists had been administered. These were the first patients included this study. Given the long duration of this task, we subsequently decided to shorten the lists to 5 items. We compared performance in control participants having received either the 6-word list or the 5-word list conditions and observed virtually identical item recall and order recall accuracies and therefore we decided to merge the control data. For the single subject analyses reported later in the manuscript, we also compared performance of patients E, F, G and H with control data obtained only on the 18 participants who had received the 6 items lists, leading to identical results in terms of patient impairment profile as those reported in the manuscript and based on the full control sample.

11

during presentation. If you do not remember a word in a given position, say ‘blank’.” The participants’ responses were recorded for later transcription and scoring. In order to distinguish item and order recall performance, we computed two different performance measures, based on the type of errors produced by the subjects. The item recall measure was obtained by determining the proportion of items correctly recalled independently of their serial position, relative to the total number of items to be recalled. The serial order recall measure was obtained by calculating the proportion of items recalled in correct serial position relative to the overall amount of items recalled, allowing us to obtain a serial order recall measure corrected for differences in item recall abilities. Open list immediate serial recall task Two sets of 108 words were constructed, differing at the level of word imageability. The high imageability words had a rating of >4, and the low imageability words had a rating of <3, on a rating scale ranging from 1 (low) to 6 (high) (Hogenraad & Orianne, 1981). Both sets were matched for word length and contained 1-, 2- and 3-syllable words; mean word length was 1.8 syllables in each list. Both sets were also matched for word frequency, t(214)=1.749, n.s. (Content, Mousty, & Radeau, 1990). The words of each set were randomly assigned to lists ranging from 2 to 7 items, with four lists per sequence length. An increasing list length design was used for this task, while a fixed list length design was used for the previous task, in order to obtain estimates of verbal STM capacity that are independent of a specific STM task structure. An increasing list length design, probing STM for different list difficulty levels, with all list lengths being administered to all patients and participants, also allowed us to avoid potential ceiling effects for high capacity control participants and hence to obtain a more sensitive estimate of the range of control performance. The lists were presented auditorily at the rate of one item every 1 second. The instructions given were: “Recall all the words in the same order as during presentation. If you do not remember a word in a given 12

position, say ‘blank’.” The high imageability lists were presented first. We computed the same item recall and order recall scores as for the closed list immediate serial task, and this for each imageability list condition and by pooling over the different list lengths.

13

Table 1. Patient general characteristics. Patient Age Gender

Years of education

Profession

Condition/most recent

Time post-

lesion (months post-onset)

onset (months)

A

56

f

12

office left temporal CVA / orbitoworker frontal medial (4) B 60 f 12 social nurse left sylvian CVA / left inferior frontal, Broca’s area (11) 37 f 12 graphical left thalamic & right designer vertebral CVA / bilateral C thalamus, cerebellum vermis, left inferior occipital (6) 47 f 10 assistant left fronto-parietal CVA / D cook left superior temporal and inferior parietal, left middle occipital (10) 54 f 8 cleaning left sylvian CVA / left E operative superior temporal and inferior parietal (1) 66 m 12 sales TBI /left anterior temporal F manager (5) 53 f 12 hairdresser left occipito-temporal G CVA / left temporooccipital (1) 46 m 12 factory left sylvian CVA / left H worker temporo-parietal-occipital (9) 47 f 12 social brain tumor / right posterior I worker temporal (11) 43 f 12 factory left sylvian CVA; left J worker fronto-parieto-temporal (8) 53 m 12 factory left sylvian CVA; left K worker superior temporal & postcentral, left internal capsula (15) 49 m 12 logistic TBI with left temporoL supervisor parietal hematoma / no visible cortical lesion (6) 52 m 12 postal TBI / left fronto-temporal M worker and fronto-parietal (1) 56 f 12 team leader left sylvian CVA ; left N temporo-parietal (1) Table 2. Patient language and cognitive profile.

6 8

12

10

20

8 31

20

13 17 17

25

15 14

14

Patient Digit span f/b

Corsi

minimal pair

nonword

picture

auditory

Other

visuo- discrimination1 repetition2 naming3

semantic

impaired

categorization4

cognitive

spatial

domains 5

span A

6/3

4

-0.06

-0.08

-15.45

/

B

6/5

/

-0.76

-0.35

-1.64

max

C

6/4

/

-0.27

-1.38

1.27

max

D

4/2

/

-5.56

-12.55

-9.64

max

response inhibition, flexibility, response inhibition processing speed /

E

6/3

5

/

-25.26

-3.03

max

/

4/3

6

-5.09

-10.00

-1.64

max

G

4/4

/

-1.11

-15.55

-1.67

max

response inhibition, processing speed /

H

4/3

/

-11.85

-28.74

-2.36

max

/

5/5

/

-28.39

-0.57

-0.31

max

5/3

/

-1.93

-5.34

-3.71

max

K

4/3

/

-22.04

-36.60

-6.78

max

L

5/5

/

-23.78

-5.34

0.88

max

sustained attention, selective attention response inhibition, flexibility selective attention /

M

6/4

/

/

-12.31

-5.07

max

/

N

4/4

/

-10.61

-38.22

-0.30

max

selective attention

F

I

J

1

2

Minimal pair discrimination for 200 nonsense syllables differing by a single consonant (e.g. [bada]); task from Majerus et al. (2005) and Attout et al. (2012). Non-word repetition for nonwords with high or low phonotactic frequency patterns, all nonwords having a CVCCVC structure (e.g., /kubtal/ vs. /gormyf/ ; 60 items per condition); task from Attout et al. (2012). 15

3

4

5

The picture naming task contains a total of 90 line drawings or photographs of objects, the target names varying in word frequency (high, medium or low frequency) and word length (1 syllable, 2 syllables, 3 syllables) (Language Rehabilitation Task Force, 1998). Semantic categorization task for 72 pictures depicting objects from six categories that were from very distinct semantic fields (animals, tools, clothes, vehicles, musical instruments, arms) and objects from six other categories from close semantic fields (washing tools, writing tools, cooking tools, flying vehicles, road vehicles, water vehicles) (Majerus et al., 2001). For patient C, this test was replaced by a synonym judgment task for 80, auditorily presented high or low imageability words (Majerus et al., 2001). TAP – Test zur Prüfung der Aufmerksamkeit (Zimmermann & Fimm, 2009); indicated domains are impaired at the Z ≤ -2.00 level according to the standardized norms of the TAP assessing attentional and executive functions For the different language tasks, each individual patient’s performance is expressed in Zscores relative to the performance of independent normative groups of healthy controls (minimal pair discrimination and nonword repetition: N = 45, age range 45-65; nonword repetition: N = 53, age range 40-65; picture naming: N = 60, age range 40-65)

Serial order reconstruction task The serial order reconstruction task consisted of the auditory presentation of digit lists of increasing length. The lists, containing 3–9 digits, were sampled from digits 1 to 9. For list length 3, only the digits 1, 2, and 3 were used. For list length 4, only the digits 1, 2, 3, and 4 were used, and so on for other list lengths. The lists were recorded by a female voice and stored on computer disk, with a 500-ms inter-stimulus interval between each item in the list (mean item duration: 540 (±139) ms). The sequences were presented via high quality loudspeakers connected to a PC. They were presented with increasing length, with six trials for each sequence length. At the end of each trial, the participants were given cards (size: 5 · 5 cm) on which the digits presented during the trial were printed in black font. The number of cards corresponded to the number of digits presented and were presented in numerical order to the participants. The participants were requested to arrange the cards on the desk horizontally following their order of presentation. The instructions were: “You will now hear a sequence of 3 (4, 5, 6, 7, 8, 9) digits, including the digits 1, 2 and 3 (4, 5, 6, 7, 8, 9). Immediately after, I will give you cards on which the digits are printed. You need to arrange

16

the cards in the same order as the order in which the digits you heard were presented. You are ready?" We determined the number of positions correctly reconstructed, by pooling over all trials and list lengths.

RESULTS Group-based analyses A first analysis assessed group differences in performance on the closed list immediate serial recall task. Although the item and order recall measures obtained for this task are considered to reflect independent item and serial order STM abilities, as outlined in the introduction (Attout et al., 2012; Nairne & Kelley, 2004), separate analyses were run for the item and serial order recall measures given that they are not obtained from two different data points, and hence do not reflect repeated measures stricto sensu. Two t-tests assessed the group effect for the item and order recall measures, as a function of item and order information recall performance. We observed a significant group effect for both the item and order recall measures, t(119)=3.49, p<.0001,  2p=.09, and t(119)=4.28, p<.001, 2p=.13, with similar effect sizes. As shown in Figure 1, the patient group performed more poorly than the control group for both the item recall and the order recall measure. At the same time, as shown in Figure 1, the variance of performance in the patient group was much larger than in the control group. Given that this situation violates the homogeneity of variance assumption, estimation of the group effect could be inflated; we therefore re-assessed the group effects using the Welch test correcting for potential estimation bias. By applying a Welch correction on all measures, the group effect was confirmed: item recall measure, t(14.54)=2.56, p<.05, order recall measure, t(14.25)=2.91, p<.05. In sum, a robust group effect was observed for both the item and order recall measures of the closed list immediate serial recall task, with, at the same time, a very large inter-individual variability of performance in the patient group. 17

Figure 1. Group-level performance (mean and standard deviation; arrows indicate minima and maxima) for the closed list immediate serial recall task. Next, an ANOVA analysis was conducted for exploring the group effect in the open list immediate serial recall task, with word imageability as a two-level between-subjects variable. Item recall and serial order recall measures were again analyzed in separate analyses. Note that data for this task were not available for patient A restricting the patient group to N = 13. For the item recall measure, we observed a strongly significant effect of group, F(1,69)=26.31, p<.00001, 2p=.28, a significant effect of word imageability, F(1,69)=90.04, p<.00001, 2p=.57 and a significant group-by-word imageability interaction, F(1,69)=8.03, p<.01, 2p=.10. Post-hoc comparisons (Tukey’s HSD for unequal sample sizes; p<.05) showed that the imageability effect was significant in both groups, but was substantially increased in the patient group (controls: 2p=.52; patients: 2p=.79). When running the same analysis for the serial order recall measure, we observed again a strongly significant effect of group, F(1,69)=18.31, p<.0001, 2p=.21, a small but significant effect of word imageability, F(1,69)=6.93, p<.05, 2p=.09 and no significant group-by-word imageability interaction, F(1,69)<1.00, p=.76, 2p=.01. As expected, the word imageability effect was much smaller for the serial order recall measure (2p=.09) than for the item recall measure (2p=.57) in line with 18

previous studies on semantic effects in STM (Majerus & D’Argembeau, 2011; Nairne & Kelley, 2004). Post-hoc comparisons showed that the word imageability for the serial order recall measure was only significant in the control group (controls: 2p=.17; patients: 2p=.07). Finally, as for the closed pool immediate serial recall task, performance was considerably more variable in the patient group than in the control group. Therefore the group effects were re-assessed using Welch tests, confirming robust group effects for each of the four measures of this analysis: high imageability item recall measure, F(1,13.32)=8.28, p<.05, low imageability item recall measure, F(1,14.28)=16.40, p<.01, high imageability order recall measure, F(1,14.33)=9.40, p<.01, low imageability order recall measure, F(1,13.74)=6.78, p<.05.

Figure 2. Group-level performance (mean and standard deviation; arrows indicate minima and maxima) for the serial order reconstruction task. Finally, a t-test assessed the group effect for the serial order reconstruction task. Note that data for this task were not available for patients G and E, restricting the patient group to N = 12. A group effect was observed, t(98) = 6.51, p<.00001, 2p=.30, but with again a considerably larger variance in the patient group. The group effect was confirmed using a Welch test, t(11.80)=4.01, p<.005. 19

Figure 3. Group-level performance (mean and standard deviation; arrows indicate minima and maxima) for the open list immediate serial recall. In sum, the group analyses and effect size estimates show robust group effects for all measures, suggesting equally impaired performance for item-based and order-based measures of STM performance in the patient group. The increased imageability effect observed in the patient group for the open list item recall measure indicates particularly pronounced verbal STM impairment when support coming from semantic knowledge is reduced (Attout et al., 2012; R. Martin & Lesch, 1996). At the same time, performance in the patient group shows considerable variability which will be explored in the next section.

Single subject analyses Given the large variability observed in the patient group, it is likely that not every patient presents significant impairment on the item or order STM tasks, and that the group analyses hide a considerable heterogeneity in individual STM profiles. In order to obtain an overview of individual STM profiles, we displayed individual patient z-scores for each STM task on a radar plot. The serial order STM measures were positioned on the right side of the radar plot, and included the closed list order recall measure, the serial order reconstruction measure, the 20

high imageability open list order recall measure and the low imageability open list order recall measure. The item STM measures were positioned on the left side of the radar plot, and included the high imageability open list item recall measure, the low imageability open list item recall measure, and the closed list item recall measure; we further included an estimate of the size of the imageability effect in the item recall measures given that the preceding analyses had shown that item but not order recall was substantially influenced by the word imageability status of the memoranda; this imageability estimate was obtained by calculating a difference score between high imageability and low imageability item recall measures.

21

Figure 4. Radar plots for patients with relative greater serial order (top panel) or item (bottom panel) STM impairment. Serial order STM measures (C_ISR_Order = closed list order recall; REC_ORDER=serial order reconstruction, O_ISR_HI_Order= open list high imageability 22

order recall; O_ISR_LI_Order= open list low imageability order recall) are indicated in green font and item STM measures (O_ISR_HI_Item= open list high imageability item recall; O_ISR_LI_Item= open list low imageability item recall; C_ISR_Item = closed list item recall; IMAG_Item=imageability effect for item recall on the open list recall task) are indicated in blue font. As shown in Figures 4 and 5, three different patient profiles could be distinguished. Figure 4, top panel, depicts patients A, D, F and M with a relative greater serial order than item STM impairment, signaled by stronger negative z scores for the serial order STM measures positioned on the right side of the radar plot. Figure 4, bottom panel, depicts patients G, H, I with a relative greater item STM impairment, as shown by stronger negative z-scores on the left side of the radar plot, where the item STM measures are positioned. Figure 5 depicts the remaining 7 patients which appear to show a more balanced profile, with negative z-scores for most item and serial order STM measures.

23

Figure 5. Radar plots for patients with similar serial order and item STM performance. Serial order STM measures (C_ISR_Order = closed list order recall; REC_ORDER=serial order reconstruction, O_ISR_HI_Order= open list high imageability order recall; 24

O_ISR_LI_Order= open list low imageability order recall) are indicated in green font and item STM measures (O_ISR_HI_Item= open list high imageability item recall; O_ISR_LI_Item= open list low imageability item recall; C_ISR_Item = closed list item recall; IMAG_Item=imageability effect for item recall on the open list recall task) are indicated in blue font. Next, we determined the strength and specificity of the individual STM profiles, by computing composite z-scores for the different item and serial order STM measures. These composite scores were calculated by averaging the different z-scores obtained for the individual item or serial order STM measures. Table 2 shows the item composite z-scores and the serial order composite z-scores for each patient. All patients showed a mean deficit of at least one standard deviation relative to control data for the item and/or order STM domains. Modified t-tests, by assuming a theoretical normal population of mean = 0, standard deviation =1, and N=100, were used to determine the significance of impairment as indicated by the composite z-scores (Crawford & Garthwaite, 2002). The item composite z-score was selectively impaired in only one patient, patient H (see Table 3). The serial order composite zscore was selectively impaired in four patients, A, D, L and M. Patients K and N showed significantly impaired composite z-scores for both serial order and item components. These results confirm a considerable heterogeneity of STM profiles between patients, in contrast to the group-based results which identified reliable impairment for both order and item STM measures.

25

Table 3. Composite z-scores for item and serial order STM variables.

Patient

Item composite

Order composite

Z-score

Z- score

A

0,23

-2,18*

B

-0,43

-1,42

C

-0,05

-1,04

D

-1,37

-3,20*

E

-1,07

-1,40

F

0,11

-1,23

G

-1,40

0,36

H

-1,88*

0,17

I

-1,08

0,26

J

-1,32

-1,33

K

-3,40*

-2,42*

L

-1,36

-1,86*

M

-1,27

-2,66*

N

-4,13*

-3,59*

* Z-score indicating significant impairment (modified t-test; pone-tailed<.05; Crawford & Garthwaite, 2002) Finally, we examined the discrepancy between item and serial order STM deficits, on the basis of the composite item and serial order composite z-scores. We computed a composite difference z-score: composite  z-score= (item composite z-score) – (order composite z-score). A composite  z-score of +1 indicates an average discrepancy of one 26

standard deviation between item and serial order STM performance, with stronger serial order than item STM impairment; in the same logic, a composite  z-score of -1 signals reflects stronger item STM impairment relative to serial order STM impairment. The values of these composite  z-scores are presented in Figure 6. We observed positive scores exceeding 1 standard deviation average discrepancy in patients A, D, F and M, while negative scores were observed in patients G, H and I. This further shows that item and serial order STM components are not impaired equally across the different patients, and that selective item STM impairment is only present in a minority of patients.

Figure 6. Composite  z-scores. Positive scores reflect relative greater serial order impairment and negative scores reflect relative greater item impairment.

27

Finally, we assessed the association between item STM impairment, serial order STM impairment and language profiles, by computing Spearman’s rho correlation coefficients between the patients’ item and serial order composite z-scores with their z-scores for the different language tests; confidence intervals were estimated using a bootstrap procedure based on 1000 bootstrap samples. For the item composite z-score, we observed correlations of rho=-.05, p=.86 [.95 C.I.: -.55 - .68], rho =.42, p=.18 [.95 C.I.: -.26 - .85], and rho =.79, p<.001 [.95 C.I.: .39 - 99] with picture naming, minimal pair discrimination, and nonword repetition, respectively. For the serial order composite z-score, these correlations were rho =.39, p=.17 [.95 C.I.: -.22 - .81], rho =.17, p=.60 [.95 C.I.: -55 - .77], and rho =.23, p=.42 [.95 C.I.: -.38 - .75]. The correlation between the serial order and item composite scores was rho =.23, p=.43 [.95 C.I.: -38 - .75]. Although these correlations need to be considered with caution given the small sample size and the resulting large confidence intervals, they nevertheless indicate a strong correlation between the item composite z-score and the nonword repetition measure, suggesting a close relationship between phonological processing abilities and the item STM component. Note that a strong correlation between verbal STM span size and nonword repetition measures had also been previously observed in a metaanalysis on published cases with verbal STM impairment (Majerus, 2009). The present results suggest that the correlation between phonological impairment and verbal STM impairment is mainly driven by item STM impairment.

DISCUSSION The aim of this study was to determine whether verbal STM impairment in aphasia is a mere consequence of underlying single word processing deficits, resulting in pronounced and selective difficulties for item STM components, or whether verbal STM is impaired more broadly by affecting also the retention of verbal serial order information. A group analysis 28

showed reliably poorer STM performance in the aphasic group relative to a large control group, and this for both item and serial order processing aspects of verbal STM. An analysis of individual profiles however revealed at the same time a large heterogeneity of verbal STM profiles, with four patients presenting selectively impaired serial order STM, and two patients presenting impairment at both serial order and item STM levels. Only one patient presented a selective item STM impairment. The remaining seven patients presented weak but not significantly impaired performance for the item or serial order verbal STM components. Group versus single case analyses One important aspect of this study is the comparison of group-based and individual profile analyses for establishing STM profiles in aphasia. Most previous studies either focused only on group-level data or on single case data (e.g., Attout et al., 2012; Burgio & Basso, 1997; Caramazza et al., 1981; Kasselimis et al., 2013; Lang & Quitz, 2012; N. Martin & Saffran, 1992; Mayer & Murray, 2012; Potagas et al., 2011). The present study shows the importance of considering results at both the group level and the individual level, as group results may hide an important variability in individual profiles. If restricting the analyses of the present data set to the group level, we would conclude that patients with single word processing deficits have both impaired serial order and item verbal STM components and that the deficits are of similar size in each domain. An analysis of individual profiles however leads to a different conclusion, with the different patients showing very heterogeneous profiles. These results also mirror those obtained by Seniów et al. (2009) comparing verbal and visuo-spatial STM abilities, revealing overall poor performance for visuo-spatial tasks in an aphasic group, but this deficit was only observed in about 50% of patients when considering results at the individual level. In the present study, these heterogeneous individual profiles were obtained despite the use of conservative analysis methods. First, we established verbal STM profiles on the basis of composite z-scores established over multiple tasks for a 29

given STM component, leading to robust, across-task estimates of verbal STM impairment. Single case studies typically establish impairment profiles from individual test results where a patient may show a selective impairment on some tasks but not all tasks measuring the same component (e.g., Attout et al., 2012). In the present case, in order to qualify for impairment in a specific verbal STM domain, a patient had to show an average deficit of at least -1.80 standard deviations over the different tasks measuring one STM component. Second, the control data used here were derived from large control populations in order to obtain a robust and valid estimate of performance variability in the general population, with controls having varying socio-economic backgrounds. In other words, when a deficit for an individual patient was observed in the present study, it was based on performance clearly outside performance range of the general population estimate, and not just outside performance range of a specific socio-economic subgroup. Single case studies typically use small to very small control groups (often N=10 or less) from a restricted socio-economic or educational subgroup, which may lead to underestimation of the actual variability in the general population, and hence to overestimation of the significance of a patient’s deficit in a given patient. The deficits and between-STM-component dissociations observed at an individual patient level in the present study can thus be considered to be conservative estimates, and yet a considerable heterogeneity in individual verbal STM profiles was observed. As such, this study is the first to show selective impairment for item or serial order verbal STM components in an unselected sample of aphasic patients, as previous reports either focused on selected, single case analyses or described simple dissociations (Attout et al., 2012; Majerus et al., 2007). A question that arises is the extent to which the serial order verbal STM deficits may have been the reflection of more general cognitive deficits at the level of attentional processing or executive functioning. Although this study was not designed to directly compare serial order STM deficits to other cognitive processing domains, all patients had 30

received testing of attentional and executive functioning via a standardized battery (Zimmermann & Fimm, 2001). Of the four patients presenting a more pronounced serial order verbal STM deficit, two had preserved attentional and executive functioning (D, M) while the two other patients (A, F) showed impaired response inhibition on a go-nogo task (see Table 2). Response inhibition difficulties could have interfered with performance on the serial order reconstruction task requiring classifying response cards as a function of serial position. However, if a response inhibition impairment was to account for the serial order STM impairment in patients A and F, the dissociation between item and order recall scores for the immediate serial recall tasks is difficult to explain, given that both scores are based on the same type of auditory-vocal response. Patients D and M further demonstrate that serial order STM impairment can be observed without any associated impairment at the level of executive or attentional functioning. When applying the same reasoning to patients showing more pronounced item STM impairment, a similar, heterogeneous situation was observed: Two patients (G, H) showed no associated attentional or executive function deficits, while patient I showed sustained and selective attention deficits. In sum, this analysis indicates that serial order verbal STM deficits are not systematically accompanied by attentional or executive processing difficulties, although the small number of patients under investigation here urges us to remain cautious when discussing the potential role of associated cognitive deficits. Theoretical implications The present results disconfirm the most extreme forms of linguistic accounts of verbal STM impairment in aphasia, which would have predicted verbal STM impairment to be most pronounced for the item STM component. Relatively greater item than serial order verbal STM impairment was observed only in three patients (G, H, I), only one patient (H) was significantly and selectively impaired for the item STM component, and 4 patients showed item STM performance within 1 standard deviation of control performance (A, B, C, F). One 31

may argue that the latter observation could be related to the fact that these patients may have had milder single word processing deficits or may have recovered from these deficits, especially at the lexical level given that all stimuli used in the different STM tasks required access to lexical content. This explanation could be valid for patients B, C and F who showed no or rather mild signs of anomia at the moment of this study (object naming, z=-1.64, z=1.27, z=-1.64, for patients B, C and F, respectively) but not for patient A who showed severe difficulties at the level of lexical access (z=-15.45). At the same time, the correlation analysis between the item STM composite z-score and the language measures revealed a strong correlation with the nonword repetition measure, indicating a robust relationship between phonological processing abilities and item STM memory capacities despite the use of lexical material in the STM tasks. Yet, this relationship needs to be considered with caution, given that, despite this strong correlation, there was one patient, patient F, who showed impaired nonword repetition (z=-10.00) but preserved item STM abilities. In sum, item verbal STM impairment in aphasic patients does not directly and only mirror their single word processing difficulties, even if our results indicate an association between phonological processing deficits and item STM abilities. It is interesting to note that none of the language measures showed a significant correlation with the severity of serial order verbal STM impairment, although again we need to remain cautious here given the small patient sample size. Martin and Saffran (1997) proposed that phonological and semantic processing levels contribute to the retention of serial order information. Given the small and non-significant correlation between the phonological processing measures and the serial order STM abilities in our patient sample, the present study does not provide evidence for an association between phonological impairment and serial order verbal STM deficits. As regards the role of semantic impairment in serial order STM difficulties, the present results also do not support such a role given that performance on 32

a demanding semantic categorization task was preserved for all patients with serial order verbal STM impairment who had been administered this task. At the same time, this does not rule out an influence of phonological and semantic representational levels on more specific aspects such as primacy and recency effects in serial recall tasks. Martin and Saffran (1997) showed that patients with phonological impairment tend to have stronger difficulties to recall items from the recency portion of a memory list, while patients with semantic difficulties tend to have stronger difficulties to recall items from the primacy portion of a memory list. In the framework adopted in this study, we consider that these patients actually have item verbal STM impairment, with phonological impairment preventing efficient coding of the phonological characteristics of items on which recency recall is based, and semantic impairment preventing efficient coding of semantic item characteristics on which primacy recall is based. These patients will thus omit items from specific portions of the STM list, but they should not have difficulties in maintaining and recalling the serial positon information of the remaining items. This is supported at least for phonological processing by the results of the present study, with no significant correlation being observed between the severity of phonological processing deficits and the ability to maintain and recall serial order information, but a significant correlation being observed between phonological processing deficits and item recall. Overall, the results of the present study provide further evidence for the separation of item STM and serial order verbal STM components, as implemented by many recent models of verbal STM (Burgess & Hitch, 1999; Page & Norris, 1998; Brown et al., 2000; Martin & Gupta, 2004; see Hurlstone & Hitch, 2014, for a review). At the same time, this raises the question of the nature of these serial order STM capacities, which is far from being resolved. A number of studies have linked serial order coding to non-linguistic codes shared with other cognitive domains, such as temporal codes (Brown et al., 2000), numerical codes (Botvinick 33

& Watanabe, 2007) or spatial codes (Ginsburg, van Dijck, Previtali, Fias, & Gevers, 2014; Van Dijck & Fias, 2011; Van Dijck, Abrahamse, Majerus, & Fias, 2013; Abrahamse, van Dijck, Majerus, & Fias, 2014). Neural regions outside the language cortex have been linked to serial order coding, with a consistent implication of dorsolateral prefrontal cortex, cerebellar and intraparietal cortex, and this more predominantly in the right hemisphere (Attout et al., 2014; Henson et al., 2000; Majerus et al., 2010; Marshuetz et al., 2000). Given that the patients of this study showed primarily left-hemisphere damage in temporal or frontotemporal cortex, the overall high frequency of serial order STM impairment may surprise at first hand. On the other hand, Kalm and Norris (2014) recently showed that superior temporal and inferior parietal cortex in the left hemisphere also code some aspects of serial order processing, as they demonstrated that multivariate neural activation patterns in left temporoparietal cortex allow to distinguish between different syllable sequences, independently of syllable identity. Given these findings, the observation of overall lower performance for the serial order STM tasks in the aphasic group relative to the control group may be less surprising. Recent theoretical work has suggested that serial order information in the verbal domain may be coded at multiple representational levels, including updating of bindings directly between language item representations, and linking of items to distinct, non-linguistic positional markers (Fischer-Baum & McCloskey, 2015). With respect to both theoretical accounts, and especially relative to the within-language domain binding account, the important implication of the present study is that processes involved in item and serial order retention reflect distinct cognitive capacities and hence are separable, even when operating within the language domain. This is also in line with the findings by Kalm and Norris who showed that even if regions involved in phonological language processing allow to decode serial order information, the representations involved in coding item information in these regions are different from those involved in coding serial order information. Finally,

34

according to recent accounts suggesting that domain-general mechanisms may support coding of serial order information (e.g., Abrahamse et al., 2014; Hurlstone et al., 2015), one should predict that patients presenting specific serial order STM deficits in the verbal modality should also present deficits in visuo-spatial STM tasks that require sequential processing and storage. This prediction remains to be tested in future studies. Conclusions This study is the first to provide a comprehensive assessment of item STM and serial order verbal STM components in a sample of fluent aphasic patients with single word processing difficulties, by using both group-based and individual profile analysis strategies. At the group level, robust impairment at both item and serial order processing levels is observed. At the individual level, three types of verbal STM profiles are observed, highlighting an important heterogeneity hidden by group analyses. These results rule out extreme linguistic accounts of verbal STM impairment in aphasia which would have predicted impairment mainly at the item processing level and a close association between item and serial order STM impairment. This heterogeneity shows that item and serial order verbal STM capacities can be selectively impaired in aphasic patients, calling for comprehensive and detailed assessment strategies of verbal STM deficits in aphasia.

35

REFERENCES Abrahamse, E., van Dijck, J. P., Majerus, S., & Fias, W. (2014). Finding the answer in space: the mental whiteboard hypothesis on serial order in working memory. Frontiers in Human Neuroscience, 8, 932. doi: 10.3389/fnhum.2014.00932 Acheson, D. J., & MacDonald, M. C. (2009). Verbal working memory and language production: Common approaches to the serial ordering of verbal information. Psychological Bulletin, 135, 50–68. Acheson, D. J., MacDonald,M. C., & Postle, B. R. (2011). The effect of concurrent semantic categorization on delayed serial recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37, 44–59. Attout, L., Fias, W., Salmon, E., & Majerus, S. (2014). Common Neural Substrates for Ordinal Representation in Short-Term Memory, Numerical and Alphabetical Cognition. PLoS ONE, 9, e92049-doi:92010.91371/journal.pone.0092049. Attout, L., Van der Kaa, M. A., George, M., & Majerus, S. (2012). Dissociating short-term memory and language impairment: The importance of item and serial order information. Aphasiology, 26, 355-382. Baddeley, A., Gathercole, S., & Papagno, C. (1998). The phonological loop as a language learning device. Psychological Review, 105, 158-173. Baddeley, A. D. (2000). The episodic buffer: a new component of working memory? Trends in Cognitive Sciences, 4, 417-423. Botvinick, M., & Watanabe, T. (2007). From numerosity to ordinal rank: a gain-field model of serial order representation in cortical working memory. Journal of Neuroscience, 27, 8636-8642. Brown, G. D. A., Preece, T., & Hulme, C. (2000). Oscillator-based memory for serial order. Psychological Review, 107, 127-181. 36

Burgess, N., & Hitch, G. J. (1999). Memory for serial order: A network model of the phonological loop and its timing. Psychological Review, 106, 551-581. Burgess, N., & Hitch, G. J. (2006). A revised model of short-term memory and long-term learning of verbal sequences. Journal of Memory and Language, 55, 627-652. Burgio, F., & Basso, A. (1997). Memory and aphasia. Neuropsychologia, 35, 759-766. Caramazza, A., Basili, A. G., Koller, J. J., & Berndt, R. S. (1981). An investigation of repetition and language processing in a case of conduction aphasia. Brain and Language, 14, 235-271. Content, A., Mousty, P., & Radeau, M. (1990). BRULEX. Une base de donnees lexicales informatisee pour le francais ecrit et parle. / BRULEX: A computerized lexical data base for the French language. Année Psychologique, 90, 551-566. Crawford, J. R., & Garthwaite, P. H. (2002). Investigation of the single case in neuropsychology: confidence limits on the abnormality of test scores and test score differences. Neuropsychologia, 40, 1196-1208. Fiebach, C. J., Friederici, A. D., Smith, E. E., & Swinney, D. (2007). Lateral Inferotemporal Cortex Maintains Conceptual–Semantic Representations in Verbal Working Memory. Journal of Cognitive Neuroscience, 19, 2035-2049. Fischer-Baum, S., & McCloskey, M. (2015). Representation of Item Position in Immediate Serial Recall: Evidence From Intrusion Errors. Journal of Experimental Psychology: Learning, Memory, and Cognition. doi: 10.1037/xlm0000102 Francis, D., Clark, N., & Humphreys, G. (2003). The treatment of an auditory working memory deficit and the implications for sentence comprehension abilities in mild "receptive" aphasia. Aphasiology, 17, 723-750. doi: 10.1080/02687030344000201 Gathercole, S. E., Frankish, C. R., Pickering, S. J., & Peaker, S. (1999). Phonotactic influences on short-term memory. Journal of Experimental Psychology: Human

37

Learning and Memory, 25, 84-95. Ginsburg, V., van Dijck, J.-P., Previtali, P., Fias, W., & Gevers, W. (2014). The impact of verbal working memory on number-space associations. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40, 976-986. Gupta, P., & MacWhinney, B. (1997). Vocabulary acquisition and verbal short-term memory: computational and neural bases. Brain and Language, 59, 267-333. Henson, R. N. A., Burgess, N., & Frith, C. D. (2000). Recoding, storage, rehearsal, and grouping in verbal short-term memory: an fMRI study. Neuropsychologia, 38, 426440. Hogenraad, R., & Orianne, F. (1981). Valence d'imagerie de 1130 noms de la langue française parlée. Psychologica Belgica, 21, 21-30. Hulme, C., Maughan, S., & Brown, G. D. (1991). Memory for familiar and unfamiliar words: Evidence for a long-term memory contribution to short-term memory span. Journal of Memory and Language, 30, 685-701. Hurlstone, M. J., & Hitch, G. J. (2015). How is the serial order of a spatial sequence represented? Insights from transposition latencies. Journal of Experimental Psychology, Learning Memory and Cognition, 41, 295-324. Hurlstone, M. J., Hitch, G. J., & Baddeley, A. D. (2014). Memory for serial order across domains: An overview of the literature and directions for future research. Psychological Bulletin, 140, 339-373. Jefferies, E., Frankish, C. R., & Lambon Ralph, M. A. (2006). Lexical and semantic binding in verbal short-term memory. Journal of Memory and Language, 54, 81-98. Kalm, K., & Norris, D. (2014). The representation of order information in auditory-verbal short-term memory. The Journal of Neuroscience, 34, 6879-6886. Kasselimis, D. S., Simos, P. G., Economou, A., Peppas, C., Evdokimidis, I., & Potagas, C.

38

(2013). Are memory deficits dependent on the presence of aphasia in left brain damaged patients? Neuropsychologia, 51, 1773-1776. Koenig-Bruhin, M., & Studer-Eichenberger, F. (2007). Therapy of short‐term memory disorders in fluent aphasia: A single case study. Aphasiology, 21, 448-458. Lang, C.J., & Quitz, A. (2012). Verbal and nonverbal memory impairment in aphasia. Journal of Neurological Sciences, 12, 1655-1661. Language Rehabilitation Task Force (1998). Examen long du langage. Centre Hospitalier Universitaire, Liège. Majerus, S. (2009). Verbal short-term memory and temporary activation of language representations: the importance of distinguishing item and order information. In A. S. Thorn & M. Page (Eds.), Interactions between short-term and long-term memory in the verbal domain (pp. 244-276). Hove, UK: Psychology Press. Majerus, S., & Boukebza, C. (2013). Short-term memory for serial order supports vocabulary development: New evidence from a novel word learning paradigm. Journal of Experimental Child Psychology, 116, 811-828. Majerus, S., & D'Argembeau, A. (2011). Verbal short-term memory reflects the organization of long-term memory. Further evidence from short-term memory for emotional words. Journal of Memory and Language, 64, 181-197. Majerus, S., D'Argembeau, A., Martinez, T., Belayachi, S., Van der Linden, M., Collette, F., . . . Maquet, P. (2010). The commonality of neural networks for verbal and visual shortterm memory. Journal of Cognitive Neuroscience, 22, 2570-2593. Majerus, S., Lekeu, F., Van der Linden, M., & Salmon, E. (2001). Deep dysphasia: further evidence on the relationship between phonological short-term memory and language processing impairments. Cognitive Neuropsychology, 18, 385-410. Majerus, S., Martinez Perez, T., & Oberauer, K. (2012). Two distinct origins of long-term

39

learning effects in verbal short-term memory. Journal of Memory and Language, 66, 38-51. Majerus, S., Norris, D., & Patterson, K. (2007). What do patients with semantic dementia remember in verbal short-term memory? Sounds and order but not words. Cognitive Neuropsychology, 24, 131-151. Majerus, S., Poncelet, M., Van der Linden, M., Albouy, G., Salmon, E., Sterpenich, V., . . . Maquet, P. (2006). The left intraparietal sulcus and verbal short-term memory: Focus of attention or serial order? NeuroImage, 32, 880-891. Majerus, S., Poncelet, M., Van der Linden, M., & Weekes, B. (2008). Lexical learning in bilingual adults: the relative importance of short-term memory for serial order and phonological knowledge. Cognition, 107, 395-419. Majerus, S., Van der Linden, M., Mulder, L., Meulemans, T., & Peters, F. (2004). Verbal short-term memory reflects the sublexical organization of the phonological language network: Evidence from an incidental phonotactic learning paradigm. Journal of Memory and Language, 51, 297-306. Majerus, S., Van der Linden, M., Poncelet, M., & Metz-Lutz, M.N. (2004). Can phonological and semantic short-term memory be dissociated ? Further evidence from LandauKleffner syndrome. Cognitive Neuropsychology, 21, 491-512. Marshuetz, C., Smith, E. E., Jonides, J., DeGutis, J., & Chenevert, T. L. (2000). Order information in working memory: fMRI evidence for parietal and prefrontal mechanisms. Journal of Cognitive Neuroscience, 12, 130-144. Martin, N., & Ayala, J. (2004). Measurements of auditory-verbal STM span in aphasia: Effects of item, task, and lexical impairment. Brain and Language, 89, 464-483. Martin, N., & Gupta, P. (2004). Exploring the relationship between word processing and verbal short-term memory: Evidence from associations and dissociations. Cognitive

40

Neuropsychology , 21, 213-228. Martin, N., & Saffran, E. M. (1992). A computational account of deep dysphasia: Evidence from a single case study. Brain and Language, 43, 240-274. Martin, N., & Saffran, E. (1997). Language and auditory-verbal STM impairments: Evidence for common underlying processes. Cognitive Neuropsychology, 14, 641–682. Martin, N., Saffran, E. M., & Dell, G. S. (1996). Recovery in deep dysphasia: Evidence for a relation between auditory-verbal STM capacity and lexical errors in repetition. Brain and Language, 52, 83-113. Martin, R. C., & Lesch, M. F. (1996). Associations and dissociations between language impairment and list recall: Implications for models of STM. In S. E. Gathercole (Ed.), Models of short-term memory (pp. 149-178). East Sussex, UK: Hove. Mayer, J. F., & Murray, L. L. (2012). Measuring working memory deficits in aphasia. Journal of Communication Disorders, 45, 325-339. Murray, L. L., Ballard, K., & Karcher, L. (2004). Linguistic specific treatment: just for Broca's aphasia? Aphasiology, 11, 993-1016. Nairne, J. S., & Kelley, M. R. (2004). Separating item and order information through process dissociation. Journal of Memory and Language, 50, 113-133. Nakayama, M., Tanida, Y., & Saito, S. (2015). Long-Term Phonological Knowledge Supports Serial Ordering in Working Memory. Journal of Experimental Psychology: Learning, Memory, and Cognition. doi: 10.1037/a0038825 New, B., Pallier, C., Brysbaert, M., & Ferrand, L. (2004). Lexique 2: A new French lexical database. Behavior Research Methods, Instruments, & Computers, 36, 516-524. Page, M. P. A., & Norris, D. (1998). The Primacy model: A new model of immediate serial recall. Psychological Review, 105, 761-781. Papagno, C., Vernice, M., & Cecchetto, C. (2013). Phonology without semantics? Good

41

enough for verbal short-term memory. Evidence from a patient with semantic dementia. Cortex, 49, 626-636. Patterson, K. E., Graham, N., & Hodges, J. R. (1994). The impact of semantic memory loss on phonological representations. Journal of Cognitive Neuroscience, 6, 57-69. Poirier, M., & Saint-Aubin, J. (1996). Immediate serial recall, word frequency, item identity and item position. Canadian Journal of Experimental Psychology, 50, 408-412. Poirier, M., Saint-Aubin, J., Mair, A., Tehan, G., & Tolan, A. (2015). Order recall in verbal short-term memory: The role of semantic networks. Memory & Cognition, 43, 489499. Potagas, C., Kasselimis, D., & Evdokimidis, I. (2011). Short-term and working memory impairments in aphasia. Neuropsychologia, 49(, 2874-2878. Seniów, J., Litwin, M., & Leśniak, M.S. (2009). The relationship between non-linguistic cognitive deficits and language recovery in patients with aphasia. Journal of Neurological Sciences, 15, 91-94. Thorn, A. S., & Frankish, C. R. (2005). Long-Term Knowledge Effects on Serial Recall of Nonwords Are Not Exclusively Lexical. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 729-735. van Dijck, J. P., Abrahamse, E. L., Majerus, S., & Fias, W. (2013). Spatial attention interacts with serial-order retrieval from verbal working memory. Psychological Science, 24, 1854-1859. van Dijck, J. P., & Fias, W. (2011). A working memory account for spatial-numerical associations. Cognition., 119, 114-119. Walker, I., & Hulme, C. (1999). Concrete words are easier to recall than abstract words: Evidence for a semantic contribution to short-term serial recall. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25, 1256-1271.

42

Wechsler, D. (1997). WAIS-III. Echelle d’intelligence de Wechsler pour adultes-version III. Paris: Editions du Centre de Psychologie Appliquée. Zimmermann, P., & Fimm, B. (2007). Testbatterie zur Aufmerksamkeitsprüfung Version 2.1. Herzogenrath,Germany: Vera Fimm - Psychologische Testsysteme.

ACKNOWLEDGMENTS This work was supported by grants F.R.S.-FNRS N°1.5.056.10 (Fund for Scientific Research, FNRS, Belgium), PAI-IUAP P7/11 (Belgian Science Policy) and ARC12/17/01REST (Université de Liège). We would like to thank the patients and control participants for their time and effort invested in this study, and Françoise Collette and Mathilde Guillet for assistance in data collection.

Highlights    

Verbal STM for item and serial order information impaired in patients with aphasia Individual profiles reveal an important heterogeneity and double dissociations Item STM deficits are associated with phonological impairment Item and serial order retention rely on distinct verbal STM processes

43