Academic year

(1)

Introducing the QUEST-NL, a short bedside assessment of language for awake brain surgery:

Design and testing in healthy individuals

Academic year 2019-2020/2020-2021

Course code LTS998M20

Supervisor Dr. A. Rofes

Name Vertregt, J. (Juliette)

Student number s2843412

Version Final version

(2)

2

Abstract

Introduction: Considerable research is conducted on brain tumour and awake craniotomy

effects on language functioning and how to assess language functioning pre-, intra- and postoperatively in individuals with brain tumours. They often experience mild, selective language difficulties only detectable by sensitive tasks. However, a lack of standardized tests specifically designed for these difficulties exists, especially in the acute postoperative stage. Therefore, tests for aphasia are administered that lack sensitivity and suitability for detecting mild difficulties in all stages of awake craniotomy. Aim: In an attempt to fill the gap, here a short bedside test called the QUEST-NL is introduced. The design is aimed to be highly sensitive for detecting mild difficulties in the acute postoperative stage in individuals with brain tumours, and to provide a quick, complete, general functioning overview of routes and levels needed for adequate language functioning. Method: A detailed description of the QUEST-NL is provided after which two studies are conducted: a picture-name agreement study in 4 Dutch and 1 Flemish healthy individual(s) and a pre-test study in 20 healthy individuals divided over two age ranges. Results: The findings of the picture-name agreement study provided valuable information for ameliorating and completing the final test design. The pre-test findings showed that each parallel version can be administered in 6 minutes. Therefore, the QUEST-NL seems highly suitable for the acute postoperative stage. In general the test complexity was found to be highly viable for healthy individuals and similar for both versions and age ranges. However, comparing expressive subtasks revealed significant differences between versions. Discussion: In light of limitations of existing neurolinguistic protocols, language assessment approaches in clinical practice and aphasia bedside tests it is elaborated how these are tried to overcome in the QUEST-NL. Furthermore, the error patterns found based on the significant differences between subtasks are discussed in light of providing guidance for future references. Lastly, by considering current limitations and future directions, it is concluded that standardizing the QUEST-NL is important to be continued. It has a high potential of becoming a pioneer in being a reliable, valid, short test sensitive enough to detect mild, selective language difficulties.

(3)

3

Acknowledgements

Hereby, I would like to wholeheartedly thank my thesis supervisor Dr. Adria Rofes for providing me the opportunity to work on this unique project in collaboration with the Donders Institute. I feel grateful and honoured that I could grow as a person and extend my knowledge within a field of linguistics that raised my interest from the first year of my bachelor in linguistics. Furthermore, I would like to express my appreciation for the continuous support I experienced from Dr. Adria Rofes during this period. Besides, I would like to thank my internship supervisor of the Donders Institute, Vitoria Piai, deeply for the valuable discussions and feedback sessions we shared, and the continuous support I experienced from her as well. Moreover, I would like to thank my fellow student Imke Wets who created the opportunity to collect information on the linguistic structure of Flemish and the illustrator Victor Xandri Antolin without whom it would have been impossible to create such clear and matching images for both versions of the test.

Additionally, I would like to express my deep appreciation for all my family members, friends and acquaintances willing to participate in my project, despite the outbreak of COVID-19, without whom it was otherwise not possible to gain initial insight in the suitability of the test. Following this, I would like to especially thank my partner and highly talented beginning scientist Dmitri Schelling for all his support, help and input throughout the project.

All of you contributed to the successful and valuable completion of my project of which I am proud, and which will hopefully be useful for future references within the research field of linguistics.

(4)

4

5

APPENDIX B ... 77 APPENDIX C ... 79

(6)

6

Chapter 1: Introduction

People with brain tumours often experience selective language difficulties that are most often mildly natured. Although the difficulties are generally mild, they can present themselves at every level of language processing during every stage of the awake craniotomy process. Therefore, there exists a general rationale in assessing language pre-, intra- and postoperatively in this population in order to maintain preserved language abilities, identify impaired language abilities and monitor improvements or deteriorations in these abilities. However, when considering the language assessment batteries currently used, some gaps can be identified in literature and clinical practice. Firstly, the language assessment batteries used are either not specifically designed for people with brain tumours but for people with aphasia or are specifically designed for people with brain tumours but lack standardization criteria. As a consequence, they are thought to lack the sensitivity to detect mild, selective language difficulties as present in people with brain tumours. Secondly, currently used assessment batteries are mostly suitable for pre- and intraoperative stages, but there does not yet seem to be a standardized test especially designed for language assessment during the direct postoperative stage in people with brain tumours.

The aim of the current study is to fill these gaps in literature and clinical practice by introducing the design and preliminary standardization data of a new bedside test called the Dutch edition of the QUick ESTimation of language impairment (QUEST-NL). The test is specifically developed for detecting mild, selective language impairments as present in people with brain tumours. Besides, it is specifically developed to be suitable for administration directly postoperatively by non-linguistic as well as linguistic professionals. To verify and prove the relevance of the QUEST-NL, not only the design is described but also one picture-name agreement study and one pre-test study in healthy controls are conducted. With regard to the pre-test study, additionally the equivalence of the two parallel test versions of the QUEST-NL is statistically tested by comparing the individual scores on all subtasks of version A with the individual scores on all subtasks of version B.

This thesis is structured as follows: chapter 1 provides insight in the underlying deficit and language difficulties in people with brain tumours, the protocols and specific batteries currently used in all stages of awake craniotomy, and the need for a new type of language battery. In chapter 2 the design of the QUEST-NL and the method sections of the picture-name agreement study and the pre-test study are discussed. In chapter 3, the results of both preliminary standardization studies are discussed. In chapter 4 the QUEST-NL is discussed in light of the limitations of earlier literature and language assessment approaches. Besides, the limitations and future directions of the current study are discussed. Finally, in chapter 5 the general conclusion is discussed.

(7)

7

The incidence of brain tumours in the Netherlands is 5 to 7 per 100.000 individuals of which 20% can be classified as being low-grade tumours (Houben et al., 2006). These low-grade tumours are often situated in eloquent areas for language. As a consequence, individuals often state to experience changes in language functioning. Based on the location of these tumours one would expect that their language functioning is significantly deteriorated. However, even when considering tumour location and size, evidence shows that in these individuals the probability of having language difficulties is significantly lower than in individuals with strokes. Besides, if difficulties are present, their severity is generally mild (Anderson, Damasio & Tranel, 1990; Davie et al., 2009; Whittle et al., 1998).

This can be explained by differences in underlying pathological processes in strokes versus tumours. Whereas strokes involve the neural tissue, tumours involve nonneural tissue and only cause neural deterioration if depressing or dislocating this tissue. Moreover, for strokes specific brain areas have a higher probability to be impacted than others, due to the network of arteries and veins supplying blood to the brain, whereas for brain tumours there is major variation in the areas impacted (Miceli et al., 2012). Lastly, strokes are characterized by an acute onset whereas tumours by a gradual onset. Due to the fact that low-grade tumours have a growth rate of only 5 millimetres a year, brain plasticity induces reorganization of at risk language functions. Therefore, individuals with low-grade tumours often experience only mild and selective language difficulties (Desmurget, Bonnetblanc & Duffau, 2007; Duffau, 2008; Mandonnet et al., 2003). Although the language difficulties in individuals with brain tumours are mostly mild and selective, they can interfere with communication in daily life. Therefore, it is important to gain insight in the levels of language functioning that are found and assumed to be regularly (and selectively) impaired in the preoperative, immediate postoperative and follow-up stage in this population, so these levels can be monitored closely throughout the process of awake craniotomy.

Preoperative language difficulties in individuals with a presumptive brain tumour are explicitly examined in Antonsson et al. (2018). Twenty-three Swedish individuals were tested by using standard aphasia tests, tests from the field of dyslexia to examine morphology and the ability to analyse sentences, a Swedish test to examine high level language functions (BeSS; Laakso, Brunnegård, Hartelius & Ahlsén, 2000), and a subjective interview. Significantly lower lexical retrieval abilities were found in individuals with brain tumours compared to controls, but difficulties were of mild severity. Regarding high level language functions, 6 individuals performed significantly lower than controls on at least one task of the BeSS, indicating that not only more basic language abilities can be impaired in people with brain tumours but also complex language abilities in which adequate, extensive (non-)linguistic processing abilities are essential (e.g. repetition of long, complex sentences or following spoken, complex commands). This number was expected to be higher due to the complexity of high level language tasks. However, the BeSS is only piloted and therefore not yet reliably examined regarding validity, sensitivity and specificity, although it is known to have limited power (Antonsson et al., 2018). This implies that the low number of individuals identified as performing significantly poorer might be

(8)

8

the consequence of an inability to reveal impairments actually present in some of the other individuals in the sample (Antonsson et al., 2018). From a subjective point of view, it became clear that word retrieval, spelling, motor speech and remembering names are difficulties experienced by individuals with brain tumours themselves of which some individuals could not be identified as having language difficulties with existing standard language tests. Following this, in Satoer et al. (2012) a subjective examination was conducted as well in which it was found that 56.5% of individuals with brain tumours state to experience word finding problems that complicate everyday language, which could not be identified with an aphasia screening battery. This inconsistency between test results and subjective interviews implies the complexity and unsuitability of detecting these and potentially other, less consciously experienced difficulties with standard assessments for aphasia (Antonsson et al., 2018).

Other studies also acknowledge the presence of (alternative) preoperative language difficulties. Firstly, in Satoer, Vincent, Smits, Dirven and Visch-Brink (2013) a higher frequency of incomplete sentences was found in pre- and postoperative spontaneous speech samples of 27 people with brain tumours in eloquent as well as non-eloquent language areas. The incomplete sentences were preoperatively characterized by the omission of content words. Furthermore, preoperatively a peculiar correlation was found in that higher scores on a naming task would lead to more incomplete sentences. It was stated that this correlation indicates a disability to build adequate sentence frames including correctly selected lexical items. Secondly, in Racine, Li, Molinaro, Butowski and Berger (2015) word retrieval difficulties were found in more than 40% of the individuals (N = 22) for fluency as well as naming tasks. Thirdly, in Santini et al. (2012) it was found that 50% of the sample (N = 22) showed difficulties in at least one task of the BADA (Miceli, Laudanna, Burani, & Capasso, 1994) (23% on naming, 32% on comprehension and 13% on reading tasks). Fourthly, in Satoer et al. (2014) it was stated that often difficulties with word finding, memorizing short-term facts and carrying out complex tasks are reported to be experienced. Furthermore, long term consequences of awake brain surgery were examined in 45 individuals. At preoperative baseline difficulties were observed with naming and fluency tasks. Lastly, in Faulkner et al. (2017) it was found that 61% of 49 individuals with great variability in tumour characteristics and sites showed impairments in at least one task of their own assessment protocol. Most difficulties were observed with verb generation, naming, verbal fluency and picture-word matching, indicating that verb retrieval, auditory picture-word recognition, accessing semantic knowledge, lexical selection and phonological encoding seem to be the impaired underlying language functions.

With regard to immediate postoperative language difficulties, in Antonsson et al. (2017) it was found that, at group-level, the lower preoperative performance of 32 individuals with brain tumours compared to controls on the BNT, category fluency task and verb fluency task was even further deteriorated in the immediate postoperative stage, which is in concurrence with earlier findings (Bello et al., 2007; Santini et al., 2012; Satoer et al., 2014). In alternative studies, other immediate postoperative difficulties concern reading, writing, sentence comprehension, auditory comprehension, repetition, sentence dictation, word to picture matching, image description, incomplete sentences and utterance

(9)

9

length (Santini et al., 2012; Satoer et al., 2016). With regard to incomplete sentences and utterance length, the study of Satoer et al. (2013) showed that omitting content words as well as obligatory sentence parts (making a sentence comprehensible for the conversationalist) stand out postoperatively in people with brain tumours. Furthermore, difficulties with utterance length were found to be characterized by lower mean length of utterances for words (MLUw). This indicates that there were less words present in each utterance compared to utterances of controls and that the inclusion of sentence tasks is highly important when evaluating language functioning in people with brain tumours.

Follow-up language difficulties were also examined within the study of Antonsson et al. (2017). The performance of the sample group was at preoperative level for all tasks except the category fluency task on which performance was still significantly below preoperative levels. In other studies it was stated that at follow-up naming abilities can also be significantly lower than preoperatively found. Besides, when considering subjective interviews 58% of the individuals reported to feel concerned having new difficulties in speech, comprehension, reading, spelling and/or word finding (Chainay et al., 2009; Papagno et al., 2011; Moritz-Gasser, Herbet, Maldonado & Duffau, 2012; Satoer et al., 2012; Satoer et al., 2013; Racine et al., 2015).

Interestingly, beside studying group-level patterns individual levels were also examined in Antonsson et al. (2017) revealing even other pre- and postoperative difficulties. Preoperatively, four individuals had mild aphasia and 10 individuals experienced difficulties in language comprehension and high level language functions beside the difficulties present at group-level discussed above. In the acute postoperative stage, 68.8% experienced mild lexical retrieval and/or comprehension deficits. At follow-up, 51.7% experienced mild lexical retrieval, language comprehension and/or high level language deficits and for 25% of them these difficulties were not present at earlier stages. Therefore, the concerns reported in the interviews by Racine et al. (2015) seem to be justified for at least some individuals.

Although the studies above show that generally all levels of language functioning can be (selectively) impaired in people with brain tumours in the pre- and postoperative stages of awake craniotomy, one might have noticed that there is one language level that seems to be ‘forgotten’ when studying pre- and postoperative language difficulties: writing. Antonsson et al. (2018) did examine this ability in people with brain tumours, because writing requires intact language and memory functioning and is therefore a highly complex task. In 20 individuals with presumptive brain tumours it was found that pre- and postoperatively mainly typing speed was affected compared to controls. Besides, individuals with brain tumours had a higher proportion of pauses within words, typed fewer characters between pauses, showed a larger time frame between keystrokes, and showed overall a lower production rate than expected based on the time invested in typing. These production rate and typing speed impairments were found to be even further declined in the postoperative stage. That writing difficulties can occur in this population is further supported by Tucha et al. (2000). In 22% of their sample (N = 139) postoperative language difficulties were observed when individuals had to write a few sentences and when they had to read a short text, which could not be explained by motor or visual impairments.

(10)

10

So, it can be concluded that writing is a level of language functioning that can be impaired in people with brain tumours. However, beside intact language and memory functioning, also intact visual abilities and literacy skills are necessary in order to write successfully (Flamand-Roze et al., 2011). The elements of intact vision and literacy skills seem to be two main reasons that writing is not often evaluated pre- and postoperatively in people with brain tumours, because one’s writing ability can be influenced significantly by illiteracy or visual impairments such as hemiplegia (Flamand-Roze et al., 2011). Therefore, when including writing tasks in pre- and postoperative language assessment it would be complicated to differentiate whether a low score would be due to difficulties with writing or alternative difficulties.

Generally, the above discussion shows that language difficulties present in individuals with brain tumours vary greatly and can reveal themselves at every level of language functioning in all stages of the surgical process. Because diagnosed and experienced difficulties are commonly of mild nature and each level of language functioning can be selectively impaired, it is of significant importance to attempt assessing each language processing route and level of language functioning thoroughly during all surgical stages. Thereby, it can be ascertained that no level of functioning is missed. Besides, individuals can be informed in detail about the consequences and course of the tumour and/or the surgical procedure.

1.2 Current neurolinguistic protocols and language tests for awake brain surgery

Due to the great variety in mild, selective language difficulties present in individuals with brain tumours and because they can reveal themselves at every level of language functioning during every stage of the surgical process, there exists a general rationale in literature and clinical practice to assess language within this population. Next, an overview is provided of currently available neurolinguistic protocols and language tests for assessing language functioning in individuals with brain tumours.

In Miceli, Capasso, Monti, Santini and Talacchi (2012) a general protocol is proposed in which it is preoperatively advised to assess: (1) sublexical processing, (2) semantic and lexical knowledge, (3) syntactic abilities and (4) verbal short term memory (STM). Perioperatively, the intraoperative task assembly should be determined based on consulting neuroimaging information and the preoperative language profile. Thereby, damage to language functions during surgery is minimized.

Intraoperatively, picture naming tasks together with other tasks mapping eloquent language areas accurately and rapidly are recommended (e.g. word repetition tasks). Besides, it is advised to determine the task assembly based on: (a) anatomical criteria or (b) functional criteria. In case of approach (a), the included tasks are known to be affected by impairments in the brain area in which the tumour is situated. In case of (b), the included tasks are known to play a role in a certain linguistic level of the functional frameworks of at risk language functions. In Miceli et al. (2012) approach (b) is recommended over (a) because the kind of impaired language functions and location of these functions for similar tumours is ambiguous. Lastly, according to Miceli et al. (2012) the task assembly should be

(11)

11

as specific and customized as possible regarding the individual with the brain tumour. That is, by considering the tumour location in the individual and reasoning which subtle, selective levels could be impaired in that area based on current knowledge regarding the neurofunctional organization of language functions, not only the general levels of language organization should be assessed (e.g. level of semantics), but also subtle, selective levels that can be impaired independently from each other (e.g. the semantic level of living concepts versus the level of non-living concepts). However, due to time-consuming and labour-intensive factors accompanying this approach future test batteries should include tasks with parallel versions that could be used pre- as well as intraoperatively.

Postoperatively, in Miceli et al. (2012) it is advised to conduct language assessment within days or at least within 2-3 weeks after surgery (with medical stability as a requirement). This test administration should fall within 20-25 minutes due to suboptimal workload endurance and frequent fluctuations in postoperative cognitive abilities. Regarding postoperative task selection, tasks with parallel versions are recommended, so improvements in language functioning can be differentiated from learning effects. At follow-up, the task assembly is recommended to include language tasks and tasks mapping other language related cognitive abilities.

In Coello et al. (2013) a minimal pre-, intra- and postoperative protocol is proposed, in which language and other cognitive tasks that should minimally be included during these stages are suggested. Beside this protocol, general parameters are proposed for selecting individually customized intraoperative tasks. The parameters are based on the idea that the role of tumour location in brain functioning and additionally that the relation of the location to other functional connections playing a role in brain functioning (e.g. subcortical white matter tracts) should be considered during task selection. The first general parameter suggested by Coello et al. (2013) is using a multidisciplinary approach. According to Coello et al. (2013), thereby ideally only sensitive tasks assessing several brain functions at once will be included. Here, the value of naming tasks is pointed out because these tasks assess all subprocesses (which are all separate brain functions) responsible for word retrieval. The second general parameter suggested is that task selection should be based on individual characteristics such as hemispheric dominancy and important personal life characteristics (e.g. career). This strategy is pointed out to be fundamental, because it enhances the chances of a comparable quality of life as before surgery. The third general parameter suggested is that tasks should be selected based on tumour characteristics following the hodotopic view on brain functioning. This hodotopic view entails that brain functions depend on several parallel, interactive, large neural networks (De Benedictus & Duffau, 2011). Assuming this view to be true, it would therefore be safer to include tasks related to for example tumour location. From there the tumour cannot only affect that specific area but a much wider range of connected networks. This third parameter is the basic idea behind the recommendations for intraoperative task selection suggested in Coello et al. (2013). If considering the recommended tasks per tumour location, the reoccurring naming task for each area and white matter tract connectivity and the often reoccurring STM double task (simultaneously move a limb while naming a picture) stand out.

(12)

12

In Połczyńska (2009) a threefold task classification protocol available for English is suggested consisting of newly designed tasks for intraoperative language mapping during DES. The first class of tasks assesses language abilities represented in the left hemisphere (LH) during DES. These tasks focus on morphological and syntactic abilities, because commonly used object naming and automatized verbal tasks (e.g. counting) are limited to accessing the semantic concept, lemma selection, phonological encoding and constructing an articulatory-motor plan. Therefore, they provide a too narrow image of all abilities needed for successful communication (Levelt, Roelofs & Meyer, 1999; Roelofs, 2004). The second class of tasks assesses language abilities represented in the right hemisphere (RH). RH damage can cause impairments in prosody, processing non-literal speech, inferential comprehension and pragmatic language (Ross & Mannot, 2008; Cutica, Bucciarelli & Bara, 2006; Fournier, Calverley, Wagner, Poock & Crossley, 2008; Tompkins, Scharp, Fassbinder, Meigh & Armstrong, 2006). Furthermore, the RH seems especially crucial in processing distant semantic relationships such as present in ambiguous sentences (Monetta, Ouellet-Plamondon & Joanette, 2006; Kacinik & Chiarello, 2007). The third class of tasks assesses language abilities managed by deeper structures of the subcortex (e.g. caudate nucleus, subcallosal fasciculus). These structures play significant roles in comprehension, lexical selection, planning, initiation, articulation and production in general (Van Lancker Sidtis, Pachana, Cummings & Sidtis, 2006).

In Papagno et al. (2012) a cognitive battery for the preoperative, immediate postoperative and follow-up stage called the Milano-Bicocca Battery (MIBIB) is proposed which is available for Italian. This battery is designed mainly based on the view that a sensitive assessment battery evaluating multiple cognitive domains should be developed, because simple evaluations such as the Boston Aphasia Severity Rating Scale that include tasks only superficially assessing a cognitive function are known to not reveal more subtle cognitive impairments actually present (Påhlson, Ek, Ahlström & Smits, 2003). Therefore, not only language tasks, but also tasks examining memory functioning, executive functioning, praxis and spatial cognition are included.

Besides, in Papagno et al. (2012) three other design parameters were taken into consideration when developing the MIBIB. Firstly, the importance of considering the effect of tumour site on cognitive abilities when selecting tasks is taken as a design parameter. This is, however, a complicated criterium. The generally slow growth rate of these lesions benefits brain plasticity and the accompanying reorganization of language functions interferes with common scientific knowledge regarding the ‘usual’ anatomical sites of language functions (Papagno et al., 2012). So, an individual with a tumour in Broca’s area could have adequate language functioning without experiencing difficulties such as apraxia of speech. Secondly, the frequent subtility and specificity of language difficulties in this population (e.g. only a certain semantic class can be affected) is taken as a design parameter. The reason for this, is that standard language examinations for aphasia are viewed to lack the sensitivity to detect mild difficulties due to the high concreteness and simplicity of tasks (Papagno et al., 2012). Thirdly, developing brief protocols taking less than 60 minutes to conduct is taken as a design parameter. However, this parameter

(13)

13

is not met, because according to Papagno et al. (2012) a sensitive and reliable evaluation of cognitive domains at least takes 1,5.

In De Witte et al. (2015) the standardized Dutch Linguistic Intraoperative Protocol (DuLIP) is proposed. It is available for Dutch and is based on the idea that each task’s sensitivity and validity to detect minor language difficulties during direct electrical stimulation (DES) is of great importance for eventual postoperative language functioning. The task assembly consists of tasks divided over each linguistic level. However, besides models are proposed by which these tasks can be selected based on (sub)cortical lesion site. The tasks are controlled for psycholinguistic variables known to influence linguistic processing at a specific level, such as frequency, imageability, word length and word form. Recently, the value of controlling for these variables is more and more acknowledged because they are attributed to a specific underlying language level and characterize the nature of lexical semantic representations. Hence, they have the ability to reveal and declare the cause of error patterns at a more detailed level than was currently possible (Whitworth, Webster, & Howard, 2014; Shallice, 1988). Besides, they can have a discriminative value in distinguishing healthy controls from people with a language impairment (Sandberg & Kiran, 2014; Rofes, De Aguiar, Ficek, Wendt, Webster, Tsapkini, 2019).

For some tasks included in the DuLIP, the disadvantages were pointed out (De Witte et al., 2015). First, a phonological odd word out task might be too simple due to visual cues fading the necessity of intact phonological abilities to respond correctly. Second, syntactic abilities should be tested with an action naming task rather than syntactic fluency and verb generation tasks. The latter two more strongly assess lexical semantic abilities simultaneously, while naming actions in sentence context specifically assesses syntactic processing because intact inflectional abilities are necessary. Lastly, a semantic association task is discouraged due to its high level of complexity even for healthy controls. Reading, semantic knowledge and language initiation interact, and all need to be intact. Besides, other cognitive processes interfere such as executive functioning and memory.

In Faulkner, Wilshire, Parker and Cunningham (2017) the Brief Language Assessment for Surgical Tumour Patients (BLAST) is described as a pre-, intra- and postoperative protocol available for English. It is designed to offer an alternative for aphasia batteries because according to Faulkner et al. (2017) they lack the sensitivity to reveal mild, selective difficulties. It is based on and therefore assessing eight cognitive abilities essential for adequate language functioning. To determine the cognitive abilities essential for adequate language functioning, current language processing theories were consulted. Eventually the following abilities were identified: (1) auditory word recognition (McClelland, Elman & Diego, 1986; Norris, McQueen & Cutler, 2000), (2) accessing semantic knowledge (Levelt et al., 1999), (3) lexical selection (Levelt et al., 1999; Roelofs, 1997, 2004), (4) phonological encoding (Levelt et al., 1999; Roelofs, 1997, 2004), (5) verbal STM (Biegler, Crowther & Martin, 2008; Martin & Friedman, 2001; Lauro, Reis, Cohen, Cecchetto & Papagno, 2010), (6) goal driven language selection (Chang, Dell & Bock, 2006; Dell, Oppenheim & Kittredge, 2008; Speer &

(14)

14

Wilshire, 2013), (7) verb retrieval (Levelt et al., 1999) and (8) articulatory motor planning (Levelt et al., 1999; Roelofs, 1997, 2004). Subsequently, data of healthy controls and people with language impairments were used to select tasks evaluating each ability in isolation. The selection process was approached based on two basic thoughts: (1) tasks had to have high sensitivity and specificity and (2) the battery should take minimal administration time. Eventually the following tasks were selected that evaluate each cognitive ability in isolation: (1) picture naming, (2) verb generation, (3) picture-word matching, (4) word repetition, (5) nonword repetition, (6) the STROOP task (MacLeod, 1991), (7) letter fluency, (8) category fluency, (9) articulatory agility. The protocol took preoperatively 20-30 minutes to administer. The administration time was not evaluated intra- and postoperatively.

All these proposed protocols provide guidance for approaching language assessment during the stages of awake craniotomy in people with brain tumours. However, they do not provide insight in the approaches currently used in clinical practice in terms of the specific tests used. The studies of De Witte and Mariën (2013) and Rofes et al. (2017) aimed to gain insight in current assessment approaches. In De Witte and Mariën (2013) assessment approaches in clinical practice were reviewed by comparing assessment data of 35 studies in the pre-, intra- and postoperative stage separately. In Rofes et al. (2017) a general survey concerning peri- and intraoperative cognitive assessments in low-grade gliomas was filled out by 21 centres affiliated with the European Low-Grade Glioma Network (ELGGN).

For the preoperative stage, in De Witte and Mariën (2013) the complete Boston Diagnostic Aphasia Examination (BDAE; Goodglass and Kaplan, 1972) and the Aachner Aphasia Test (AAT; Huber, Poeck, Weniger and Willmes, 1983), both available for many languages, were found to be frequently administered. Regarding specific abilities, phonological fluency, semantic fluency, verb generation, repetition and spontaneous speech are evaluated frequently. This is in part concurrence with the findings in Rofes et al. (2017). The centres affiliated with the ELGGN were found to commonly use fluency tasks in the perioperative stage. Intraoperatively, clear patterns of assessment approaches were found in Rofes et al. (2017). Object naming, reading, semantic association, repetition and double tasks were found to be commonly administered. Although no clear patterns were found in De Witte and Mariën (2013), often assessed abilities seem generally in concurrence with Rofes et al. (2017). It was found that tasks focussing on picture naming, counting, reading, spontaneous speech, comprehension and verb generation abilities are often used intraoperatively. Specifically, for naming the Boston Naming Test (available for many languages) (BNT; Kaplan, Goodglass and Weintraub, 1983), the Dénomination Orale 80 (available for French) (DO80; Metz-Lutz, 1991) and the naming task from the AAT (available for many languages) (Huber et al., 1983) are chosen frequently. However, regarding naming and counting tasks it is found that naming reveals brain activation in significantly more brain areas during DES than counting. Therefore, naming tasks are viewed to have a higher sensitivity and greater contribution for language mapping than automatized speech tasks (Brennan, Whalen, De Morales Branco, O’Shea, Norton & Golby, 2007). Postoperatively, in De Witte and Mariën (2013) a pattern was found in the use of short neurolinguistic examinations (e.g. the Minimal Mental State Examination

(15)

15

(MMSE; Folstein, Folstein & McHugh, 1975)) to evaluate changes in language functioning directly after surgery. Another pattern found was the use of the same comprehensive batteries as pre- and intraoperatively, such as the BDAE, AAT, BNT and DO80 (Kaplan et al., 1983; Metz-Lutz, 1991; Goodglass & Kaplan, 1972; Huber et al., 1983). In Rofes et al. (2017) for this stage no patterns of specific abilities or tasks are discussed.

1.3 The need to develop a bedside language test for people with brain tumours

1.3.1 Gaps in literature and clinical practice and implications of current approaches

What can be drawn from the protocols in literature and common assessment approaches in clinical practice, is that most materials used for language assessment during the stages of awake craniotomy are originally designed for aphasia and need to be taken by a linguistic professional. However, due to the differences in the underlying deficit and nature of language impairments in strokes compared to brain tumours, questions are raised regarding whether tasks for aphasia are sensitive enough to detect the mild and selective impairments present in people with brain tumours (Papagno et al., 2012; Faulkner et al., 2017; Pålson et al., 2003; Antonsson et al., 2018). Besides, it can be drawn from literature that although most protocols are claimed to be suitable for the pre-, intra- as well as the postoperative stage, the main focus seems to be on the intraoperative stage (Coello et al., 2013; Połczyńska, 2009; De Witte et al., 2015). This is understandable because of the direct impact of this stage on future functioning and quality of life. However, the counterpart of this focus is that few studies, and to our knowledge no studies for Dutch, focus on language functioning in the acute stage after awake craniotomy while in this stage language functioning often deteriorates and great variation is found in the levels of language functioning that are impaired (Antonsson et al., 2017; Satoer et al., 2016). Furthermore, in this stage individuals can only endure suboptimal workloads due to fatigue and medical instability (Miceli et al., 2012).

Therefore, the use of lengthy protocols and standard batteries for aphasia that can only be administered by a linguistic professional seem unsuitable, because these approaches do not seem in line with the individual’s capacities in this stage and do not seem suitable to monitor the frequent, rapid changes in language functioning occurring during this stage closely. Furthermore, in the acute stage not linguistic professionals but nurses and medical specialists are generally frequently visiting the individual to monitor their recovery. So, it seems important that these non-linguistic professions can monitor language functioning without having to consult a linguistic professional. Altogether, there seems to be a gap in literature and clinical practice regarding adequate assessment materials for the acute postoperative stage (within hours/days/a few weeks after surgery). It seems that short language assessment batteries (e.g. in the form of a bedside test) are needed that are specifically designed for detecting the mild, selective and widely varying language difficulties in people with brain tumours

(16)

16

during this stage. These batteries should be able to administer within a few minutes and several times shortly after the previous assessment time by non-linguistic professionals. Thereby, acute postoperative deteriorations and improvements in language functioning can be monitored closely, reliably and quickly without evoking pressure to perform, fatigue or stress, and postoperative impairments can be identified as early as possible.

Although there exists a gap in literature and clinical practice with regard to the immediate postoperative stage of awake craniotomy, the existing protocols and common assessment approaches seem to provide some implications that can be learned from when designing a short language assessment battery especially designed for individuals with brain tumours in the acute postoperative stage.Firstly, because currently questions are raised regarding the sensitivity of aphasia tests for people with brain tumours, there seems to be growing consensus that for language assessment in this population tests should be selected or designed that are specific, sensitive and standardized and especially designed for language impairments in people with brain tumours (Miceli et al., 2012; Papagno et al., 2012; Faulkner et al., 2017; Coello et al., 2013; Połczyńska, 2009; De Witte et al., 2015). With regard to designing a short assessment battery for the acute postoperative stage, it seems therefore important to include tasks for which the language impairments present in individuals with brain tumours are taken as a premise for the design.

Secondly, it seems agreed that administration times should be kept to a minimum to enlarge the feasibility for using the same tasks in each operative stage (Miceli et al., 2012; Papagno et al., 2012). This implication seems especially important to take into consideration when designing a short assessment battery for the acute postoperative stage, because of the suboptimal workload individuals can endure and the rapid and frequent changes in language functioning. Thirdly, there seems to grow general consensus that all operative stages are evenly important and that therefore every level of language functioning should be assessed at every stage (Miceli et al., 2012; De Witte et al., 2015). This implication seems to support the view that language assessment in the acute postoperative stage should not be neglected, but more importantly guides the design of future batteries for the acute stage in that every level of language functioning should be represented in the test. Fourthly, the key role of subcortical levels of the brain is more and more acknowledged and therefore evaluating language functioning managed by the subcortical level should become standard procedure in all operative stages (Połczyńska, 2009; Coello et al., 2013; De Witte et al., 2015). So, when designing a short assessment battery for the acute postoperative stage, including tasks for language functions that are represented at subcortical levels should be considered (e.g. reading).

Fifthly, it seems to be implied that: (1) tasks should be selected based on functional criteria as well as lesion characteristics provided that the hodotopic view of brain functioning is considered (Miceli et al., 2012; Coello et al., 2013; De Witte et al, 2015), (2) naming tasks are of great value in all operative stages (Miceli et al., 2012; Coello et al., 2013; De Witte et al., 2015; Faulkner et al., 2017), (3) tasks with parallel versions should be selected so that improvement or deterioration can be differentiated from

(17)

17

learning effects (Miceli et al., 2012), (4) commonly used aphasia batteries are ineffective due to highly concrete and simple items (Miceli et al., 2012; Papagno et al., 2012; Faulkner et al., 2017), and (5) tasks should be controlled for psycholinguistic variables because they can declare error patterns at a detailed level and have a diagnostic discriminative value (De Witte et al., 2015). These five implications give valuable considerations for designing a short assessment battery for the acute postoperative stage in that at least a naming task should be included, it should have parallel versions, it should consist of more abstract and complex items to detect mild, selective language difficulties and the items should be controlled for psycholinguistic variables so the deeper level of impairment can be identified.

Lastly, certain tasks are discouraged to include in future assessments batteries, among which tasks assessing phonological odd word out, letter, category and syntactic fluency, counting, verb generation, semantic association, and goal driven language selection abilities (De Witte et al., 2015; Faulkner et al., 2017). With regard to a short assessment battery for the acute postoperative stage this overview is valuable in the task selection phase of the test design, because it provides differentiation for tasks that are recommended and discouraged to include.

1.3.2 The future potential of aphasia screening test frames for people with brain tumours

Several tests exist that try to satisfy the need of obtaining a reliable, quick, valid overview of language functioning in the acute stages, as desired in people with brain tumours. In Hachioui et al. (2017) eight tests are systematically reviewed on accuracy, reliability and feasibility. The ScreeLing (Doesborgh et al., 2003) and the Language Screening Test (LAST; Flamand-Roze et al., 2011) were found to have the highest diagnostic value. Therefore, they are valid, sensitive and specific tests to diagnose aphasia in the acute stage. The ScreeLing is a bedside test available for Dutch consisting of 3 subtasks that evaluate respectively the linguistic level of phonology, semantics and syntax (Doesborgh et al., 2003). It was designed with the aim to create a test that provides insight in the underlying linguistic deficit. Thereby, it corresponds to and guides the current cognitive linguistic therapy approach in which therapy focusses on recovery of the affected language level. This approach seems more effective than focusing on aphasia type as was concentrated on previously in the field of aphasia (Howard & Patterson, 1989; Cicerone et al., 2000). Due to that the ScreeLing can be administered in the first 2-11 days post-stroke and takes approximately 15 minutes it seems suitable for the acute stage. Besides, it has a high sensitivity of 86% and a high specificity of 96%, indicating that 86% of the individuals with aphasia is also correctly identified as having aphasia and besides that 96% of the healthy controls is also correctly identified as not having aphasia.

The LAST (Flamand-Roze et al., 2011) is developed to overcome shortcomings of previous screening tests, such as the inclusion of written tasks (significantly influenced by e.g. hemiplegia), highly complex visual materials, tasks simultaneously measuring other non-linguistic cognitive functions, and inappropriate administration times and scoring procedures. Besides, limitations in sensitivity, reliability and validity of other tests were motives. Considering these shortcomings, the

(18)

18

LAST can be characterized by comprising no written subtasks, no highly complex visual materials and no tasks measuring other cognitive functions. Another important design consideration was that it could be easily administered at bedside by non-linguistic experts (e.g. nurses, surgeons). The LAST is available for French, English and in the near future for Dutch (D’Hondt, 2019), and consists of 5 subtasks divided over two indices (an expressive and receptive index, respectively). It can be administered in approximately 2 minutes and has an easy scoring procedure of classifying an answer as either 1 for correct or 0 for incorrect. Moreover, an individual has 5 seconds per item to answer and if no response is given then, it is classified as incorrect. The subtasks included are: (1) object naming, (2) word and sentence repetition, (3) automatized speech, (4) picture recognition and (5) following verbal instructions. Beside the administration time and easy scoring procedure, the LAST has several other advantages. Firstly, the availability of a parallel version that has an equivalence coefficient of .96 is highly valuable, in that individuals can be assessed several times during the acute stage. In this stage many fluctuations in language functioning occur. With the parallel version A test-retest bias and uncertainty whether progress is due to actual language functioning improvement or learning effects can be prevented. Secondly, the test can be administered within 24 hours post stroke meaning that previously missed individuals who recovered quickly can now be identified and monitored as well. Thirdly, the test has a higher sensitivity and specificity compared to previous tests, namely 98% and 100%, respectively.

1.3.3 The significance of verb processing over noun processing in future assessment

Further than considering the future potential of aphasia bedside tests for people with brain tumours, it can be drawn from literature that future test designs are desired to represent and assess verb processing to a higher degree than noun processing. This is relatively new to the field of designing language assessments, because most currently existing tests represent noun processing to a higher degree, which also applies to the current version of the LAST and the ScreeLing (Flamand-Roze et al., 2011; Doesborgh et al., 2003). However, if comparing verbs and nouns based on grammar it can be concluded that verbs carry more information than nouns and the grammatical interpretation of a sentence occurs through processing the verbs (Baker, 2003). Therefore, verbs, and not nouns, form the core part of a sentence (Baker, 2003).

Besides, if comparing verb and noun processing from a more clinical point of view, the higher significance of assessing verbs over nouns can also be concluded. In Vigliocco, Vinson, Druks, Barber and Cappa (2011) studies were reviewed in order to provide a clearer view on whether processing words of different grammatical classes appeal to different neural systems. It became clear that cross-linguistic consistency exists with regard to the left inferior frontal gyrus (IFG) being closely involved in verb as well as noun processing, which can even be found for nonalphabetic languages such as Chinese (Feng, Qi, Yang, Yu & Yang, 2020). However, for the syntactic processing of verbs higher brain activation patterns can be found than for nouns. For English, it was found that inflected verbs cause higher levels of activation in the left IFG than inflected nouns, which is further supported by studies investigating

(19)

19

other Indo-European languages (Longe, Randall, Stamatakis & Tyler, 2006; Shapiro et al., 2005; Palti Shachar, Hendler & Hadar, 2007). Furthermore, even for nonalphabetical and nonmorphological languages with completely different grammatical rules, the IFG showed higher activation patterns for verbs than nouns (Feng et al., 2020). Altogether, these studies show the cross-linguistic consistency in syntactic verb and noun processing in that greater neural involvement and processing demands are required for verbs.

On the other hand, verb and noun processing do not only occur at the level of syntax but also at the level of semantics (Feng et al., 2020). If considering semantic processing from the clinical point of view, converging results to syntactic processing can be found. More brain regions are found to be involved in semantic verb processing than noun processing among which the superior temporal gyrus, which further indicates the higher level of complexity of processing verbs than nouns. Besides, more brain activity within shared regions during noun and verb processing is found for verbs than nouns, such as in the left middle frontal gyrus and left temporal gyrus (Feng et al., 2020). Furthermore, in people with post-stroke aphasia, verbs are typically more impaired than nouns, even though the opposite pattern is also found (Jonkers & Bastiaanse, 2007; Kambanaros & Van Steenbrugge, 2006; Kim & Thompson, 2000; Laiacona & Caramazza, 2004).

Furthermore, if especially considering the population undergoing awake craniotomy it was investigated in Havas et al. (2015) whether electric stimulation causes disruptions selectively in noun or verb processing in certain frontal lobe sites. Generally, most stimulated areas were found to cause disruptions for verb as well as noun processing at the IFG. If only considering selective effects, it was found that verbs carried a higher sensitivity for disruptions than nouns at inferior and middle frontal regions. Overall, it was concluded that action naming tests seem more suitable for preventing postoperative language difficulties than the current commonly used object naming tests during awake craniotomies, because it was found that verbs within action naming tests have a higher sensitivity to disruptions due to DES than nouns within object naming tests (Havas et al., 2015).

Altogether, these clinical results show that verb processing is highly important to assess, because more brain regions and greater neural activation within shared brain regions are found to be involved in verb processing compared to noun processing. This implies that if verb processing is impaired, it has greater consequences for language functioning than impairments in noun processing. However, although verbs seem more demanding on everyday life language functioning, currently such correlations are only found between finite verbs and everyday life language abilities (Rofes, Capasso & Miceli, 2015). Furthermore, the higher processing demands required for verb processing imply the higher complexity of verb processing compared to noun processing. Therefore, together with the finding that verb processing is more sensitive for disruptions, in people with brain lesions it can be assumed that verb processing is more likely to be affected than noun processing, especially in individuals with mild difficulties.

(20)

20

1.3.4 The need for a new type of battery

What can be drawn from the discussion above, is that a bedside test with a design comparable to the LAST but with items controlled for psycholinguistic variables and a high sensitivity for the language difficulties present in individuals with brain tumours is desired in especially the peri- and acute postoperative stages. It should include tasks that solely assess language abilities at word and sentence level and no tasks assessing accompanying non-linguistic cognitive abilities simultaneously. These latter type of tasks (e.g. verb generation or semantic association) complicate diagnosing language difficulties due to the interference of other cognitive processes such as executive functioning and memory also needed to complete the task successfully (De Witte et al., 2015; Antonsson et al., 2017). When an impairment is identified based on these tasks, the cause of impairments is unclear because it cannot be deducted whether the cause concerns a language or a non-linguistic cognitive ability (Faulkner et al., 2017). A language ability can be defined as being crucial for adequate language functioning and being highly specialized for language. So, if impaired, communicative performance is significantly affected but if similar non-linguistic tasks are administered no significant communicative impairment is observed (Faulkner et al., 2017). Although consensus exists that non-linguistic cognitive abilities should be carefully assessed as well, this can be better administered with a separate assessment battery (e.g. the Neuropsychological Assessment Battery (Stern & White, 2003).

Furthermore, although the future bedside test should cover all language difficulties that can be present in people with brain tumours and writing can be impaired in the acute postoperative stage (Antonsson et al., 2018, Antonsson et al., 2017; Santini et al., 2012), it should not include a writing task. The reason for this is that if illiteracy and/or visual impairments such as hemiplegia are present, they significantly interfere with one’s writing ability (Flamand-Roze, 2011). Thereby, it would be complicated to identify the cause of a low score on a writing task. Besides, the bedside test should be mainly focussing on verb processing instead of noun processing. Lastly, this test should consist of tasks together covering all linguistic processing routes (except for writing), so a quick, general image of the total language functioning system can be obtained.

Chapter 2: The QUEST-NL

The QUEST-NL is a newly proposed Dutch beside test especially designed for monitoring language functioning in the acute postoperative stage in individuals that underwent awake craniotomy. It is based on the screening test frame of the LAST (Flamand-Roze et al., 2011; D’Hondt, 2019) but designed to have a higher level of complexity and greater sensitivity than currently used aphasia tests in individuals with brain tumours, so mild, selective language difficulties as present in these individuals can be detected. Additionally, the test is designed to be quick and easy to administer by non-linguistic professionals due to their main involvement in monitoring the individual in the acute postoperative stage.

(21)

21

However, to be able to design this newly proposed bedside test, first insight was needed in current universally accepted different linguistic processing models and routes, and the kind and severity of impairments present in individuals with brain tumours. Furthermore, more insight was needed in the effects of verb use on linguistic processing and the effects of psycholinguistic variables on certain linguistic levels within linguistic processing routes. Therefore, in the upcoming chapter these four main topics will be discussed in section 2.1 when describing the theoretical framework behind the design of the QUEST-NL. Beside the discussion of the theoretical framework, in this chapter a detailed description of the design of the QUEST-NL will be provided in section 2.2. Furthermore, in section 2.3 the methods of the picture-name agreement study in Dutch and Flemish healthy individuals will be discussed. Lastly, in section 2.4 the methods of pre-testing both parallel versions of the QUEST-NL in Dutch healthy individuals will be discussed.

2.1 Theoretical framework

2.1.1 Language processing models

When designing the QUEST-NL, two widely used language processing models were taken into consideration that will be further elaborated next. The first is a model specifically describing word processing, whereas the second clarifies the stages involved in sentence processing.

Model of Ellis and Young

The word processing model proposed in Ellis and Young (2014) consists of processing levels and connections that are each included based on data showing that individuals with aphasia can have (selective) impairments in these levels and/or connections. Figure 1 shows the model as visualized in Ellis and Young (2014). The model consists of two input modalities, auditory and visual comprehension, and two output modalities, articulation and written production. For each modality, several processing levels are involved that will be discussed next.

The main function of the auditory analysis system (AAS) is identifying speech patterns in incoming speech and extracting individual sounds from background noise (phonemes). It distinguishes individual speech sounds, recognizes them and determines their right order. Furthermore, it has a high level of flexibility due to great variabilities in speech rates, voices and accents in which phonemes should still be recognized as phonemes.

The auditory input lexicon (AIL) aims to identify familiar phoneme strings (words) in order to inform one that a word was already heard once or more and might probably have a corresponding meaning in the semantic system (SS). It therefore consists of the phonological representations of words that are stored based on phonological correspondence. After activating a phonological representation, the connection between the AIL and the SS is activated which initially has a key role in accessing and stimulating the activation of semantic representations in the SS.

(22)

22

The SS generally consists of the semantic representations of words (lemmas) and has a memory function in that it stores meanings of words. Despite that Ellis and Young (2014) classify the SS as one system, there is an ongoing debate whether or not two separate systems should be acknowledged within the SS (one for conceptual knowledge of a word (nonverbal) and one for word meanings) due to double dissociations observed in individuals with language impairments (Levelt, 1989). The SS is very systematically assembled in that words with corresponding meanings are memorized together in a connectionist network of spreading activation (Dell, 1989). If a word like ‘bike’ is activated, the lemmas of corresponding meanings will be activated as well (e.g. car, scooter) but to a lesser degree causing the target word to be the one comprehended or expressed. For nouns, the SS activates the syntactic structure consisting of gender, number, distribution. For verbs, the SS activates the syntactic structure consisting of number, distribution, argument structure and thematic roles.

The phonological output lexicon (POL) provides access to the spoken word form after activating a lemma in the SS or, in case of reading, after a combination of activating the visual word form in the visual input lexicon (VIL, see below) and the lemma in the SS.

The phonological output buffer (POB) comprises each individual speech sound in a distinctive manner and can be stimulated from three levels: the AAS in case repeating (un)familiar heard words, the POL in case of naming, reading aloud, speaking spontaneously or repeating familiar words, and from grapheme-phoneme conversion (G/P-conversion) in case of reading aloud unfamiliar words.

The visual analysis system (VAS) has comparable functions to the AAS although from a visual comprehension point of view. It sees to the identification of distinct letters in visual (non-)words or letter strings, the encoding of letter positioning within a word, and the perceptual arrangement of distinct letters together forming the elements of one single word.

The VIL has functions in visual comprehension comparable to the functions of the AIL in auditory comprehension. Letter strings together forming graphemic representations (words) that are already familiar to the individual are here identified and besides stored based on graphemic correspondence. If a letter string is identified as being a familiar graphemic representation, the VIL can initiate the process of accessing its semantic representation via the connection between the SS and the VIL, it can initiate the process of accessing the accurate spoken word form in the POL, or it can initiate both. However, if a letter string cannot be identified as familiar, the VIL can present itself in 3 ways: (1) VIL declares the string unfamiliar after which no further processing takes place at this level, (2) VIL allows the activation of an alternative but visually comparable graphemic representation or (3) VIL initiates an attempt to pronounce the letter string by relying on the pronunciation of a comparable familiar graphemic representation.

The graphemic output lexicon (GOL) is a storage facility for the spelled forms of familiar words and sees to the accessibility of these forms in order to continue the writing process. It can be stimulated from three levels: the SS in case of writing down one’s own conceptual thought, the AIL in case of

(23)

23

spelling to dictation, and the POL if pronouncing the word facilitates GOL-access in case of inaccessibility directly via the connection between the SS and the GOL.

The grapheme level (GL) comprises abstract grapheme representations and can comparable to the POB be stimulated from three levels: the VAS in case of copying familiar read words, the GOL in case of written naming, writing spontaneously, spelling to dictation for familiar words or copying familiar words, and phoneme-grapheme conversion (P/G-conversion) in case of spelling to dictation for unfamiliar (non)words.

The level of G/P-conversion is called upon when unfamiliar (non)words need to be read and the VIL can therefore not be relied on. Unfamiliar (non)words can only be read aloud by going directly from letter identification to speech output. Due to the fact that men is able to read aloud unfamiliar (non)words this level was included in the model. The same holds for the level of P/G-conversion in that men is able to formulate accurately spelled forms for regular but not irregular unfamiliar (non)words which is facilitated by mapping phonemic representations onto graphemic representations.

As can be observed in Figure 1, all processing levels are connected indicating that for certain language functions certain routes within the model are called upon. Next, these routes for specific language functions will be described briefly. Although some language functions can be adequately used by calling upon several routes, in designing a new test it seems important to assess at least one of those routes so that every language function is represented and eventually each processing level in the model is examined. However, regarding repetition abilities, it is important to assess at least two routes: one for words and a separate one for nonwords. Nonword repetition and word repetition rely on different underlying processing levels and otherwise certain levels are missed in the assessment.

The route for auditory word comprehension runs via AAS to AIL and finally to SS. The connections between AAS and AIL and AIL and SS are assumed to be interactive based on evidence that context information is used for early activation of a number of other possible targets and for interpreting a word despite poor hearing opportunities.

For verbal (non)word repetition several routes are available. Firstly, the lexical route that runs via AAS, AIL, SS, POL, POB and the speech output level. Secondly, the non-semantic lexical route that runs via AAS, AIL, POL, POB and the speech output level. In this case the SS is not crossed, but the route is still called lexical in that existing (ir)regular words are recognized by an individual despite that no semantic representation is available in the SS. Lastly, the sublexical route that runs via AAS, POB and the speech output level which is especially used for nonwords, new words and proper and geographical names. Furthermore, this route can be consulted to activate inner speech for monitoring one’s own utterances and retaining auditorily presented information.

Verbal word production is administered via SS, POL, POB and the speech output level. For visual word comprehension and reading aloud there are 3 routes available that show comparison to the routes available for verbal repetition. Firstly, the lexical route is available that runs via VAS, VIL and SS. For reading aloud it also runs via POL, POB and the speech output level.

(24)

24

Secondly, the non-semantic lexical route is available that runs via VAS, VIL, POL, POB and the speech output level. Again, this route does not cross the SS, but is still lexical in that existing words are recognized as having seen them before although a semantic representation is lacking. Finally, the sublexical route is available that runs via VAS, G/P-conversion, POB and the speech output level which can be used for reading aloud nonwords and new words.

Finally, to write words several options are available. First, this can be done via SS, (POL), GOL, GL and the writing level. However, if one wants to write heard nonwords the route via AAS, POB, P/G-conversion, GL and the writing level can be used. Furthermore, for writing auditorily presented existing words, the routes via AAS, AIL, SS, GOL, GL and the writing level, via AAS, AIL, POL, GOL, GL and the writing level or via AAS, AIL, SS, POL, POB, P/G-conversion, GL and the writing level can be used. If one wants to copy a written word, this is possible directly via VAS, GL and the writing level.

Model of Levelt

The model of Levelt (1989) is contrary to the model of Ellis and Young (2014) a word and sentence processing model combined (see Figure 2). Because word processing is commonly accepted

Academic year

Introducing the QUEST-NL, a short bedside assessment of language for awake brain surgery:

Design and testing in healthy individuals

2

3

4

TABLE OF CONTENTS

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24