The Dutch Telephone-based Language-Battery:
evaluation of language in brain tumour patients after surgery
Tessa Ruijs 11110856
Thesis supervisor: dr. N. Vasic Second reader: dr. J. Rispens Project Supervisor: V. Piai, PhD
20 - 01 - 2018 University of Amsterdam Ma Thesis General Linguistics
Acknowledgement
First of all, I would like to thank Vitoria Piai from the Radboudumc for helping me with this project. Thank you for helping me with the information about the test(s), the search for the right statistical analysis and for bringing me in contact with the patients I could test for this study. In addition, I would like to thank those patients for their time and feedback, which helped me a lot. Also, a big thank you to all 30 participants who I could call twice with the test.
Finally, I would like to thank Nada Vasic, my supervisor from the University of Amsterdam for the feedback and help.
Abstract
Introduction: we created the Dutch Telephone-based Language-Battery (DTLB) to evaluate language function in brain tumour patients after surgery. We created this test because, at the Radboudumc, currently, language testing post-surgery is limited to the Boston Naming Test (Kaplan et al., 1983), counting and spontaneous speech. These tests are done only six weeks and six months post-surgery when a patient does not show a severe language impairment immediately.
Background: as new surgical invention, such as awake surgery in combination with direct electrical stimulation, improve the outlook for brain tumour patients, more attention is paid to the quality of life after surgery. This can be done by more language testing before, during and after surgery. When a patient shows language difficulties, speech therapy should be offered fast so that the difficulties could partly or fully resolve. However, in spite of the language testing, aphasia could still occur and remain in the patient, even with speech therapy. To contribute to the evaluation of language, we created the DTLB, a test which could be administered over telephone. The test consists of two test versions and six language production tests, which we tested on a healthy control group and 6 (former) brain tumour patients.
Research Objectives: based on control group scores, our aim is to show 1) that the subtests of our test battery do not differ significantly from each other; 2) that the overall tests do not differ significantly from each other; 3) a descriptive analysis of the test items. Based on the patient group scores, we want to 1) compare the DTLB scores to the Neuro-Psychological screening scores to see how the two tests relate to each other; 2) show feedback from the patient group; 3) corroborate the purpose of the test-battery.
Method: we administered the test on 30 healthy participants and 6 brain tumour patients who had brain surgery in the last 12 months. Besides administering the test, we also asked the patient group for feedback on the test.
Results: the results show that the total tests and most subtests do not differ significantly from each other, except for the repetition test. In this test, most errors are made in the non-word repetitions, as the descriptive analysis shows. This description also shows that in the word retrieval test, a lot of inappropriate answers were given, due to the multiple correct answers to the questions. The results based on the patient group present that the test indeed can show the differences between pre- and post-surgery language testing were expected and the feedback about the telephone-based test was overall very positive. Lastly, the purpose of the test was shown by comparing the patients’ scores and the control group scores, which showed the differences that were expected.
Conclusion: we created a valid language-test battery that could be administered over telephone. However, given the limitations of this test, we recommend future studies to focus on adjusting this test and on the telephone-based concept.
Table of Content Acknowledgement... 2 Abstract ... 3 Table of Content……….4 List of Tables ... 5 1. Introduction ... 6 2. Background ... 8
2.1 Importance of language evaluation before, during and after surgery ... 8
2.2 Radboudumc Protocol ... 10
3. The Dutch Telephone-based Language Battery ... 12
3.1 The Test-battery ... 12 3.2 Research objective ... 14 4. Methodology ... 16 4.1 Materials ... 16 4.1.1 Comprehension screening ... 16 4.1.2 Word Retrieval ... 16 4.1.3 Repetition ... 17 4.1.4 Phonological Fluency ... 17 4.1.5 Semantic odd-word-out ... 17 4.1.6 Semantic Fluency ... 18 4.1.7 Sentence Completion ... 18 4.2 Subjects ... 19
4.3 Procedure and Analysis ... 19
5. Results ... 21
5.1 Control group results... 21
5.2 Patient group results ... 26
6. Discussion ... 31
6.1 Control group & patient group results ... 31
6.2 Limitations and recommendations ... 33
7. Conclusion... 35 References... 36 Appendix A ... 39 Appendix B ... 40 Appendix C ... 50 Appendix D ... 51
List of Tables Main text:
Table 1. Language Difficulties post-surgery in the patient group Table 2. Mean, SD and range of test version A and B of the DTLB
Table 3. Wilcoxon p-values outcome of the subtests and total test scores DTLB Table 4. Errors in the word retrieval test items
Table 5. Errors in repetition test items
Table 6. Errors in semantic odd-word-out test items Table 7. Errors in sentence completion test items Table 8. NPS and DTLB scores patient group Table 9. Z-scores patient group
Appendix:
Table 1. Subtests Description of the DuLIP
Table 2. Control group and patient group information Table 3: Results control group
1. Introduction
For the current study, the Dutch Telephone-based Language-Battery (DTLB) was created in order to evaluate language and to discover possible language difficulties after brain surgery (craniotomy). The study was done in collaboration with Vitoria Piai from the Radboud university medical centre (Radboudumc) and the Donders Institute for Brain Cognition and Behaviour in Nijmegen. At the moment, language testing at the Radboudumc is limited to only six weeks and six months after the brain surgery, with a small number of language tests: language testing starts sooner, only when a patient shows signs of a severe impairment, the patient is then tested more often. Up until June 2017, only the Boston Naming Test (Kaplan, Goodglass and Weintraub, 1983) was used to evaluate language, together with counting and spontaneous speech. In June 2017, the short Dutch Intraoperative Linguistic Protocol (De Witte et al, 2015) was added to the BNT for language testing before, during and after awake surgery. Both the limited test opportunities as well as the limited scope of language tests is unfortunate, as Satoer et al. (2014) point out, because language recovery after brain surgery takes longer than six months. It could even take place up to one year post-surgery, due to brain plasticity.
Furthermore, the lack of language testing after surgery has two consequences: first, if a patient does have a mild language deficit and it goes unnoticed, the patient will not receive treatment. Secondly, there is no information available for future patients about potential language problems, severe or mild, during the year after surgery. For these reasons, and as a brain tumour or damage done during surgery (e.g. damage to the brain tissue surrounding the tumour) could lead to a language impairment such as aphasia, language evaluation after surgery would be of an advantage to the patient. In other words, early detection of a language impairment post-surgery can lead to faster treatment and a better chance of full recovery. Moreover, with more and consistent language testing after a craniotomy, more information about the potential language deficits will be available for both future brain tumour patients, who will undergo a(n) (awake) craniotomy, and for those persons involved in brain tumour surgeries, such as the neurosurgeon, psychologist and speech therapist. With these patients, and the problems with the current protocol at the Radboudumc in mind, the DTLB was created.
The DTLB is designed to be administered over telephone and consists of six language production subtests and a pre-test comprehension screening. The comprehension screening was added in order to evaluate the general verbal and auditory comprehension of a patient, which is important as a comprehension deficit could be a possible outcome of the surgery (Wernicke’s aphasia). The six language production tests each test a different aspect of language, as they focus on word retrieval, phonology (with a repetition test and a phonological fluency test), semantics (with an odd-word-out test and a semantic fluency test) and syntax (with a sentence completion test). For the test-battery, two test versions were designed, version A and B, to avoid a training effect. These versions consist of the same subtests, but with different test-items.
With the current study, we want to put the DTLB through its first validation phase and aim to achieve six research goals. The first three focus on the control group results: 1) to show that the subtests of the test versions do not differ significantly from each other; 2) to show that the total test versions do not differ significantly from each other, and 3) based on individual results, to give a descriptive analysis of all test-items to get an insight into potential improvements to the test items. The last three goals focus on patient group scores: 1) to compare the individual patient test scores of the DTLB with their neuro-psychological screening scores, to see how
they relate to each other; 2) to show the validation of the procedure by evaluating feedback and 3) to corroborate the purpose of the test battery by comparing the patient scores with the control group scores.
The structure of the study is as follows: in chapter 2, we discuss the background of the current study. In chapter 3 we present the research objective. In chapter 4, we discuss the methodology. In chapter 5, we show the results of the current study. In chapter 6 we discuss the analysis of the results and share some limitations and recommendations for future studies and finally, in chapter 7, we draw a conclusion based on the results.
2. Background
In the current section, the background of the current study will be presented. We discuss the importance of language-testing before, during and after surgery in brain tumour patients. In addition, we present the possible language impairments due to either a brain tumour or tumour resection (see 2.1). Afterwards, we discuss the current language testing protocol of the Radboudumc and the short comings of this testing protocol (see 2.2)
2.1 Importance of language evaluation before, during and after surgery
In recent decades, the outcome for a brain tumour patient has significantly improved. One of the reasons is that new surgical procedures were invented, which improved the (awake) brain surgery techniques (Faulkner, Wilshire, Parker & Cunningham, 2017), such as Direct Electrical Stimulation (DES). Awake craniotomy1 is the preferred approach when a tumour is located in the eloquent cortex that controls language and motor function (Korkmaz Dilmen, Akcil, Oguz, Vehid and Tunali, 2017). Nevertheless, it depends on the hospital, neurosurgeon, the kind of tumour and the patient as to whether an awake procedure is going to be performed. DES is the procedure through which during awake surgery the neurosurgeon stimulates the cortical surface with a bipolar electrode in order to provoke reproducible, temporary changes in behaviour and performance, while the patient performs different language (which is referred to as ‘language mapping’), motor and cognition tests (Borchers, Himmelback, Logothetis & Karnath, 2012). As De Witte and Mariën (2013) describe it, DES is currently put into practice as the standard intraoperative approach for identifying critical language areas and its pathways when a space-taking lesion occurs. With DES, the neurosurgeon gets a better idea of the anatomy of the language functions in the patient’s brain, which could lead to better language preservation.
The current knowledge of the anatomy of the brain and pre-operative language mapping is often insufficient to predict the post-operative language status. With an awake craniotomy and language mapping during surgery, the neurosurgeon thus gets a better idea of the location of the language functions in a patient’s brain. Awake craniotomies improve the extent of tumour resection or removal and increase the quality of life in the long term after surgery. This is because sensorimotor functions (sensory, perceptual and motor skills) and cognitive functions (memory, attention, concentration and language) are more likely to be preserved (Miceli, Capasso, Monti, Santini & Talacchi, 2012). Now that more attention is paid the patients’ quality of life after a craniotomy and to the preservation of language (Faulkner et al., 2017), language testing before and after awake craniotomy is just as important as intra-operative language testing and language mapping. The comparison between the patient’s language level before and after surgery gives an indication of language changes due to the tumour resection. Whenever a deficit is detected, the right rehabilitative therapy (speech therapy) can be offered to the patient as soon as possible (Miceli et al., 2012). Also, long-term language testing after surgery could serve as a baseline for the (cognitive) sequelae (a condition which is the consequence of a previous disease or injury) of the therapeutic procedures that follow the craniotomy and for monitoring language progression (Miceli et al., 2012).
Unfortunately, even with language testing before, during and after awake craniotomies, patients can still be left with a language impairment, most often in the form of aphasia. Simply
1 Depending on the hospital, a patient can be awake during the whole procedure of the surgery (conscious sedation), or only
put, aphasia is an impairment of the ability to comprehend, produce and/or repeat language (Kemmerer, 2015). Bastiaanse (2010) emphasizes that aphasia is a language disorder and not specifically a speech disorder, as it does not necessarily affect articulation. Kemmerer (2015) described eight classic aphasic syndromes, all with their own set of symptoms: Anomic aphasia, Broca’s aphasia, Wernicke aphasia, Conduction aphasia, Transcortical Motor aphasia, Transcortical Sensory aphasia, Mixed Transcortical aphasia and Global aphasia. We have listed the main symptoms of these types of aphasia below:
Anomic aphasia: people with anomic aphasia have predominantly word-finding problems: naming objects/activities is their most impaired ability. In their word finding problems, some (semantic) categories could be more impaired than others. The diagnosis of anomic aphasia is given when difficulties with accessing lexical items is a predominant feature (Edwards, 2005). Besides being a type of aphasia, anomia is the most common symptom of language dysfunction as one can find it in most aphasic patients.
Broca’s aphasia: the speech of someone with Broca’s aphasia is nonfluent (slow, but effortful) and mostly with verbal paraphasias (explained below). Patients mostly have difficulties with producing grammatical morphemes, but their understanding of language relatively preserved. The repetition of words and sentences is usually better than normal speech, but can also be impaired, as Ardila and Rosselli (1992) mention in their study: they conclude that repetition difficulties can also occur in Broca’s aphasia and Wernicke’s aphasia, not just in conduction aphasia (discussed below).
Wernicke’s aphasia: people with Wernicke’s aphasia have fluent speech, which is, however, mostly semantically unclear. One of the biggest consequences of Wernicke’s aphasia is the impaired comprehension of sentences, phrases and mostly single words. Repetition of words and sentences is often disrupted.
Conduction aphasia: when someone has conduction aphasia, their speech is often more fluent than Broca’s aphasic patients, but less fluent than Wernicke’s aphasic patients. They produce frequent phonemic paraphasias and their comprehension is relatively preserved. The biggest characteristic is the impaired ability to repeat verbal stimuli, particularly low-frequency, polysyllabic words and non-words (Sidiropoulos, Ackermann, Wannke & Hertrich, 2010).
Transcortical motor aphasia: the symptoms of Transcortical motor aphasia (TCMA) are nonfluent speech and poor production planning and initiation. However, comprehension is relatively preserved and also repetition is good.
Transcortical sensory aphasia: people with Transcortical sensory aphasia (TCSA) have fluent speech, but with phonemic paraphasias, neologisms and semantic substitutions. The content of their speech is relatively empty. Their comprehension is impaired, while repetition is well preserved too.
Mixed Transcortical aphasia: Mixed transcortical aphasia (MTCA) is, as the name already says, a mixture of TCMA & TCSA. The speech production of patients with MTCA is nonfluent and comprehension is severely impaired. Repetition of words and sentences is still intact.
Global aphasia: people with Global aphasia have impaired speech production, speech comprehension and also their repetition is severely impaired. This aphasia type has a lower occurrence because of language reorganization in the brain during tumour growth, the use of (F)MRI scans before surgery and the cortical mapping procedures during awake surgery that
allow surgeons to minimize damage to critical language areas during the resection (Davie, Hutcheson, Barringer, Weinberg and Lewin, 2009).
As these symptoms show, word finding, repetition of words, comprehension and paraphasias (explained below) are the first language aspects one should test in order to indicate whether a person has a language problem and to specify which type a of aphasia this person has.
Also, and not unimportant, paraphasias often occur in aphasic patients. Paraphasias are unintended errors of syllables, words or phrases (Adams, Bedrick, Fergadiotis, Gorman & Van Santen, 2017)2. As Bastiaanse (2010) explains, paraphasias occur at different levels: verbal paraphasias and phonemic paraphasias. Verbal paraphasias can be divided into two categories, that is ‘semantic paraphasias’ which means that there is still a relationship between the produced word and target word (e.g. ‘table’ instead of ‘chair’) and ‘irrelevant paraphasias’, which means that there is no relationship between the produced word and the target word (e.g. ‘mountain’ instead of ‘chair’). Phonemic paraphasias are errors on articulation, which are mostly caused by either a phonological impairment or an articulatory deficit.
Finally, two other common impairments are apraxia of speech and dysarthria. Apraxia of speech (AoS) is a production disorder that involves an impairment of the articulatory planning mechanisms (Kemmerer, 2015). Dysarthria is a motor speech impairment, which is an impairment of motor control over the muscles of the vocal apparatus (Kemmerer, 2015). These last two disorders could appear on their own, but also in combination with any type of aphasia. Now, with these symptoms of different aphasic syndromes in mind, we created the DTLB with different subtest which each focus on a different symptom. Errors made in these subtests could give an indication to the type of aphasic symptom a patient could have (as will be explained in section 3.1 about the test-battery and elaborated on in section 4.1 when discussing the different subtests in more detail).
2.2 Radboudumc Protocol
In the Radboudumc, before an awake craniotomy, a long neuro-psychological screening (NPS) is executed. It takes approximately two to three hours to administer and it consists of different cognitive tests, including tests for memory, attention and language, for which, a phonological and semantical fluency test, the WAIS Similarities test (Wechsler, 2012) and the Boston Naming Test (BNT; Kaplan et al., 1983) are used. This means that language was tested before, during and after surgery only by means of the word retrieval tests, counting and spontaneous speech. The word retrieval tests focus on word finding in the mental lexicon, counting is considered to be automatic speech, or ‘overlearned’ serial speech and it is a simple way to elicit continuous and fluent motor speech (Talacchi et al., 2013), and spontaneous speech, which mirrors conversational speech in daily life (De Witte et al., 2015b) shows the interaction of different linguistic levels, such as phonology, word finding, and morphosyntax in discourse (spontaneous speech) (Prins & Bastiaanse, 2004). According to De Witte et al (2015a), testing language with only these tests is restricted because these tests do not cover the essential variety of expressive and receptive language functions which are important and necessary for sufficient communication. Fortunately, a short version of the Dutch Intraoperative Linguistic Protocol
2Adams et al. (2017) note that individuals without a language disorder also sometimes produce errors in their speech. Though, those with a language impairment such as aphasia, are tend to make them more often.
(DuLIP; De Witte et al., 2015a) was added to the NPS and is now used for language testing during the awake craniotomy.
The DuLIP, a Dutch language test-battery, was created for intraoperative language testing during cortical stimulation (DES) and language mapping, as it gives a better evaluation of the location of different language functions in the brain than only the use of word retrieval tests. The DuLIP covers a variety of articulatory, phonological, semantic and syntactic tests and it takes approximately 1.5 hours to administer. Prior to the surgery, an individual-tailored selection of the subtests is made, based on the patient’s language level before the surgery. Specifically, all test-items of the selected subtests which the patient fails to answer correctly, are removed from these subtests. This is done to ensure that all errors the patient makes during surgery are as a result of the cortical stimulation and are not caused by a pre-existing language deficit (De Witte et al., 2015). In other words, that the errors are not made because of brain damage that already exists due to the tumour, but because of DES. Table 1 in Appendix A shows all subtests of the DuLIP and how these subtests work. The short version of the DuLIP used in the Radboudumc consist of the following tests: a repetition task with words and sentences; a naming task in a semantic picture-out task; a sentence completion task and finally, a sentence-evaluation task, based on semantic, syntactical and phonological violations3.
After the surgery, language is only tested by examining spontaneous speech for language difficulties. After that, language testing is very limited: the patient will get more specific language testing immediately, only if they show a severe impairment the week after the surgery. Otherwise, language will only be tested six weeks and six months post-surgery with the short and adapted version of the DuLIP, while a patient is in the hospital for an MRI (Magnetic Resonance Image). This is unfortunate, as language recovery after brain surgery could take place until one year post-surgery, as a result of brain plasticity (Satoer et al., 2014).
Moreover, the lack of language testing after t surgery has two consequences for patients: first, when a patient has a mild language impairment, and this is undetected, the patient will not receive the treatment (s)he needs. Second, the hospital cannot give information about potential language problems caused by surgery to future brain tumour patients. Since future patients get no information, those involved in the brain tumour resection, such as the neurosurgeon and speech therapist, also don’t receive information about the state and progress of the patients’ possible language impairment.
With these (new) patients and the problems with the current testing protocol in mind, we created the Dutch Telephone-based Language-Battery. Notice that we do not intend the DTLB to replace the currently used language test, but we are positive that the DTLB will complement these currently used tests.
3In the naming task in a semantic picture-out task, the examiner will show the patients a couple of pictures, which (s)he needs to name and tell which of the pictures does not match the other pictures, based on semantics. In the sentence- evaluation task, the participant must evaluate a sentence which has either semantic, syntactical or phonological errors.
3. The Dutch Telephone-based Language Battery
In the current section, we present the Dutch Telephone-based Language-Battery. The test was created, as mentioned in section 2.2, because of the current lack of language testing after an awake craniotomy in the Radboudumc. In the following section, we discuss the concept and content of the test-battery in section 3.1. In section 3.2 we present the research objective of the current study.
3.1 The Test-battery
The DTLB was created with the goal of determining possible language deficits after brain surgery, either severe or mild. In contrast to the current language tests the Radboudumc uses, the DTLB covers more linguistic levels than just word retrieval, such as phonology, semantics and syntax. The test-battery is designed to be administered multiple times over the course of one year after surgery to keep track of possible language changes. As it is a short test (with approximately 10 test items per subtests), it does not serve the purpose of diagnosing a patient with a certain type of aphasia, but it should rather be used to establish a baseline by which language change/difficulties can be marked and which could give an indication of the possible presence of a symptom of aphasia.
The English telephone-based-Language-Battery4 was the starting point for the current study. It was created by De Witte and colleagues (In Preparation) to advance language testing after surgery, and to be less time consuming for the patient, as physical presence not being needed (the patient does not have to travel to the hospital). In the DTLB, we used the same subtests as in the English telephone-based-Language-Battery. These different subtests tap into different aphasic symptoms, which is why they were added to the English telephone-based-Language-Battery. We will explain below which aphasic symptoms or aphasic syndrome these subtests relate to, where we discuss why the DTLB consists of the same subtests and what they look like.
First of all, the DTLB consists of six language production tests and one pre-test comprehension screening. This pre-test comprehension screening indicates whether the general verbal and auditory comprehension screening of the participant is intact. This is important as a deficit in language comprehension (e.g. Wernicke’s aphasia, as mentioned in section 2.1) could be a possible outcome of the surgery. In this case, the participant will not be able to understand the language production tests. The language production tests consist of the following tests: a word retrieval test; two phonological tests: a repetition test and a phonological fluency test; two semantic tests: a semantic odd-word-out test and a semantic fluency test and a sentence completion test. With these subtests, different language aspects will be assessed. These aspects relate to possible symptoms of aphasia, for example, the repetition task was included to see whether a patient has repetition problems, which is a symptom of conduction aphasia and (possibly) Broca’s aphasia. We explain this in greater detail below, when we discuss all subtests of this test-battery.
First, the word retrieval test is included as the evaluation of word retrieval (word finding) of a patient is of great importance in the diagnosis of neuro-cognitive disorders, such as aphasia (Yochim et al., 2015). As mentioned in section 2.1, anomia is the most common symptom of
4 As neither the test-battery itself nor the article is published, we will refer to this test-battery as the English
language dysfunction as it occurs in most aphasic patients. As a word retrieval deficit is the only aphasic symptom a patient has, (s)he could be diagnosed with Anomic aphasia. Therefore, it is important to test whether a patient has such difficulties in both spontaneous speech as in an object- or action-naming task. The latter is specifically tested in this test-battery5. A simple naming task, which is mostly a picture naming task (see for example the BNT or the Aachen Aphasia Test; Huber, Poeck, Weniger, and Willmes, 1983) can already reveal mild anomia (Laine & Martin, 2006). In order to test word retrieval over the telephone, we created a Dutch non-visual test for this test-battery, which was inspired by the Verbal Naming Test created by Yochim et al. (2015). In addition to this, a naming task allows the examiner to evaluate more than one function: with simple naming task, one tests the semantic processing, lexical access, phonological encoding and speech production at the same (Fernández-Coello et al, 2013)
Furthermore, two phonological tests are also put in the test: a repetition task and a phonological fluency task. The repetition task was included as repetition difficulties can occur in multiple types of aphasia, such as conduction aphasia, Broca’s aphasia or Wernicke’s aphasia. It is worth noting that, we already ruled this last one out with the pre-test comprehension screening. Repetition is currently not tested at the Radboudumc, though, as described in section 2.1, it is one of the first symptoms one looks at when diagnosing someone with aphasia. In this test, the participant must repeat some (existing) words, but also a couple of non-word test items (e.g. frimótika) were included as well, despite the fact that this task is also slightly difficult for healthy people, as Sierpowska et al. (2017) describe in their research. They found that the participants of the study showed significant differences between correct word repetition and correct non-word repetition. In the same study, Sierpowska and colleagues discuss that the non-word repetition task is even more sensitive in evaluating possible repetition difficulties. In addition, we put the repetition tests under the phonological test-label, as repetition difficulties are due to a deficit the phonological short-term memory (STM, i.e. phonological loop). The phonological STM is the momentary storage and deals with processing of spoken material and of which its capacity is limited (Conti-Ramsden & Durkin, 2007). Similar tests can also be found in the AAT (Huber et al., 1983), the English Aachener Aphasie-test (EAAT; Miller et al., 2000) and the Comprehensive Aphasia Test-NL (CAT-NL; Visch-Brink, Vandenborre, De Smet and Mariën, 2014).
The phonological fluency test is added, because in a verbal fluency test a number of cognitive processes take part, such as working memory (remembering the assignment and keeping track of the words said; Schmand et al., 2008), self-monitoring and cognitive flexibility (Baldo et al., 2006). Baldo et al., (2006) also argue that phonological (letter) fluency requires to leek strategic through lexical or phonological memory and thus word finding is also tested with this fluency test. Such tests are often used in language-batteries, such as the DuLIP.
The two semantic tests of the DTLB are a semantic odd-word-out test and a semantic fluency test. The semantic odd-word-out test is incorporated, because it tests both semantic knowledge, semantic comprehension and semantic processing. The semantic fluency test is included as it tests semantic-based word retrieval, as word retrieval difficulties are the most common
5 Spontaneous speech is not specifically included in this language-test-battery, it could always be tested by having a small
symptom of language dysfunction, which could also be focused on naming nouns of a specific semantic category.
Finally, sentence completion is included in the test-battery as this measure some of the subtests administered previously but within one test: repetition, word retrieval and production of semantically and syntactically correct speech (De Witte et al., 2015). It also measures language dynamics; a sentence is potentially infinite and changeable as there is not just one way to complete a sentence correctly. In addition, as Bastiaanse et al. (2003) explain, producing a sentence does not just entail retrieving lexical items, but also linking the thematic roles to grammatical roles6. In section 4.1 a detailed description of all subtests can be found as well as examples of the test items.
To our knowledge, a telephone-based language-battery has not been used to assess language in brain tumour patients in The Netherlands. However, it has been used in identifying mild cognitive impairments and dementia, for example, the telephone Montreal Cognitive Assessment (t-MOCA; Pendleburry & Rothwell, 2009) and the Telephone Interview for Cognitive Status (TICS; Brandt, Spectre and Folstein, 1988). An advantage of the TICS is that the test can also be administered to individuals with (severe) visual and/or motor impairments (Fong et al., 2009). We think it is convenient for brain tumour patients too, to have a telephone-based language test, as they do not have to travel to the hospital anymore to participate in this short test. With the English Telephone Language-Battery in mind, and because we know a telephone-based test works for patients with possible cognitive impairments as well as the advantages for patients with visual and/or motor impairments and because the patient does not have to travel to the hospital, we decided to create a telephone-based language battery for brain tumour patients too.
The rest of this thesis is dedicated to the first validation phase of the DTLB: in section 3.2 we present the research objective of the current study, followed by the methodology and a description of the tests of this test-battery in section 4. In section 5 we discuss the results of the current study and in section 6 and 7, we discuss the results and draw a conclusion based on this analysis.
3.2 Research objective
Section 3.1 highlighted the importance of designing a telephone test battery to evaluate language in brain tumour patients of the Radboudumc. In this first validation phase, we first test a healthy control group with the following three goals:
1. Based on the group results, we want to show whether each of the subtests of the test- versions differ significantly from each other;
Prediction: the subtests do not differ significantly from each other.
6As an example, Bastiaanse et al. (2003:53-54) use ‘The girl eats a biscuit’ in which ‘The girl’ gets the thematic role of agent, which needs to be mapped onto the grammatical role of subject and ‘a biscuit’ gets the thematic role of theme which has to be mapped onto the grammatical role of (direct) object.
2. Based on the total group results, we want to show whether the total results of the tests of the test-versions differ significantly from each other;
Prediction: the total results in the two test versions do not differ significantly from each other.
3. Based on the individual results, we perform a descriptive analysis of the test items to get an insight into the errors of the test, in order to improve the current test version.
Secondly, we test six former brain tumour patients. With their scores, we want to show the purpose of the test battery: does the test show the difference in scores between someone with and without language problems. Based on the scores of these patients, we have three goals:
1. We want to compare the individual patient test scores of the DTLB to the NPS scores, to see how they relate to each other.
Prediction: we expect to see the scores of those who had language difficulties before the surgery increase when comparing the NPS and DTLB scores and the scores of those who had language difficulties after surgery decrease.
2. To evaluate feedback given by the patient group to see what they think about the DTLB and to validate the test.
3. To corroborate the purpose of the test battery by comparing the individual patient scores to the total control group scores: does the test show the difference in language function between someone with and without language problems?
Prediction: we expect to see a difference in the control group scores and the patient scores of those who experience language difficulties after surgery.
4. Methodology
In this section, the methodology of the current study is discussed. In the following section, we present the materials of the DTLB (see section 4.1), followed by the subjects of the current study (see section 4.2) and finally we present the test procedure and analysis of the test results (see section 4.3).
4.1 Materials
The DTLB consists of two test versions, A and B, to avoid the training-effect between pre- and post-operative language testing. That is, there is a chance someone will perform better the second time, due to “training”, as that person has already completed the same test before, which we try to avoid. Both versions consist of the same six language production subtests and a pre-test comprehension screening, but with different pre-test items. The pre-pre-test comprehension screening is administered first. If the participant passes this screening, the examiner will continue with the language production tests. These subtests are a word retrieval test; two phonological tests, namely a repetition test and a phonological fluency test; two semantic tests, which are a semantic odd-word-out test and a semantical fluency test and finally, a sentence completion test, which focusses on syntax.
All test items are based on either the DuLIP (De Witte et al., 2015) or the Diagnostic Instrument for Mild Aphasia (DIMA; Forthcoming), with the exception of the word retrieval test items, which we designed ourselves.
The test items of the subtests of both test versions are matched based on the following factors: the word retrieval items are based on frequency, those of repetition are based on syllable-count and stress pattern and the sentences used in the sentence completion task are matched based on syntactical similarities.
Below, the different subtests of the test-battery are discussed. First, in 4.1.1 the comprehension screening is presented; in 4.1.2 the word-retrieval test; in 4.1.3 the phonological fluency test, followed by the semantic odd-word-out test in 4.1.4. In 4.1.5 the semantic fluency test is discussed and finally, in 4.1.6 the sentence completion test is described. The test-items of the DTLB can be found on the score-forms in Appendix B.
4.1.1 Comprehension screening
The comprehension screening was included to evaluate the general verbal and auditory comprehension of the participant. As was mentioned above, this is important as a participant with a comprehension deficit will not be able to do the language tests, simply due to them not understanding what is being said and asked. The screening consists of five yes/no questions, divided into some simple questions (e.g. ben je een man ‘are you a man?’) and some more difficult questions (e.g. is een vlinder een vogel ‘is a butterfly a bird?’). All questions must be answered correctly by the participant in order to continue with the language tests of the test-battery. When one or more questions are answered incorrectly, one stops testing.
4.1.2 Word Retrieval
The aim of this language production test is to test word retrieval. The test consists of 20 items (fifteen nouns and five verbs). The participant is presented with a descriptive question about a
noun or verb and must produce the definition that fits the description. For example, a question could be wat doe je met een boek? ‘what do you do with a book?’ to which the participant must give a one-word answer, in this example lezen ‘to read’/’you read it’.
The inspiration for this test came from the English Telephone Language-Battery, which contains 25 test items. We chose to use only 20 items in the DTLB, to keep the test-battery short and the word retrieval test not much longer than the other subtests. The target-answers of this subtest are found in the SUBTLEX-NL database (Keuleers & Brysbaert, 2010), which is a database of Dutch word frequencies based on 44 million words from television and movie subtitles. The reason this database was used instead of Corpus Gesproken Nederlands (Taalunie, 2004), is that the SUBTLEX-NL has been updated more recently and therefore has a more accurate representation of the frequency of spoken words. The target-answers of the two test-versions are matched with each other based on frequencies and are administered in order of high-frequency to a lower frequency.
4.1.3 Repetition
The repetition test is the first phonology test. With this test, we want to see whether a patient has repetition problems of (non-)words and/or sentences. This test consists of six words and four sentences. The words can be divided into three categories: the first two words are three-syllable words with different stress patterns; the next two words are compound words; and the last two are non-words, with three or four ‘syllables’. Before these test-items, one word (legende ‘legend’) was added as a practice item.
The sentence test-items also start with a practice item (hij krijgt een geschenk ‘he gets a gift’). The following four test-item sentences are similar in length, but different in structure and tenses. For example a past-tense sentence de jongen vergat zijn vissen te voeren ‘the boy forgot to feed his fish’ (version A) and a question: waarom winnen wij nooit een wedstrijd ‘why do we never win a game?’ (version B) must be repeated. All (non-)words and sentences used in this test-battery are selected from the repetition task of the DIMA.
4.1.4 Phonological Fluency
The phonological fluency test is the second phonological test that is administered. To summarize the explanation given in section 3.1 on what this subtest tests: the appropriate items (words) the participant produces in a phonological fluency task are selected based on phonological knowledge (Baldo et al. 2006, reported in Biesbroek et al., 2016).
Participants are instructed to produce as many words as possible, in one minute, starting with phonemes /d/ (version A) and /k/ (version B). The phonemes were chosen according to the use of these phonemes in the NPS the patients must do before an awake surgery.
The control data for these phonemes was collected by Schmand et al. (2008), during the making of a phonological fluency test based on the Controlled Oral Word Association Test (COWAT) from the Multilingual Aphasia Examination (Benton & Hamsher, 1989).
4.1.5 Semantic odd-word-out
The first semantic test is the semantic odd-word-out test. The aim of this test is to test word retrieval of a specific semantic category. With this test, the participant must find the semantic
relation between words (e.g. that a car and bicycle are both in the same semantic category of means of transport). The test consists of 10 test items. The participant is presented with three words verbally. One of these words does not match the other two words semantically. The participant listens to the three words and is asked to repeat the word that is not semantically related to the other two words. (S)he is also instructed to name the reason why the word does not match the other two words.
An example of the words the participant can hear is been, arm, raam ‘leg, arm, window’, with which we expect the participant to answer raam hoort er niet bij, want dat is geen lichaamsdeel/ want dat is een deel van een gebouw ‘window does not match the other two, because that is not a body part / because it is part of a building’.
The test items of this test are selected from the semantic odd-word-out subtest of the DuLIP. However, we changed the number of words per item from four words (in the DuLIP presented on a screen) to three words which are verbally presented. We did this to simplify the task and really focus on the test to find the semantic relation between words and not specifically test working memory.
4.1.6 Semantic Fluency
The second semantic test is a semantic fluency test, also called a category fluency test. It tests the semantic-based word retrieval (Baldo et al., 2006). The current subtest only contains one semantic category. The participant is instructed to produce as many words as possible in a certain semantic category in one minute. Version A has the semantic category professions and version B has the semantic category animals.
The categories were taken from the DuLIP as these are the categories in Dutch which are used often, for example in the Groninger Intelligentie Test I (Snijders & Verhage, 1983) and II (Luteijn & Barelds, 2004). Furthermore, Van der Elst et al. (2006) provided normative data for naming animals and professions in a semantic fluency test for Dutch speaking participants and also tested the effects of age, education, and gender. Only one semantic category was included in this test, to keep the test-battery as short as possible.
4.1.7 Sentence Completion
The last test is the Sentence Completion test. This test focuses on multiple language aspects, which are tested in the other subtests, such as repetition and word retrieval, but also on language dynamics as a sentence is infinite and changeable: there is not just one way to finish a sentence correctly.
This test consists of 10 test-items in each version. The participant is asked to repeat the beginning of a sentence that is given by the examiner and to finish this sentence. The test-item sentences can be completed by either a single noun, multiple nouns or a subordinate clause, depending on the sentence-onset. An example of a sentence which must be finished is ik snij met… ‘I cut with…’, to which could be answered with ik snij met een mes ‘I cut with a knife’. The test items for this subtest are selected from the DIMA.
4.2 Subjects
For this study, two groups of participants are tested. A full overview of all the participants can be found in Table 2 in Appendix C.
The first group consists of 30 healthy adults between the ages 40 and 70. These subjects were organized into three subgroups, consisting of 10 participants each. The youngest group (40-49 years old, mean age 46,4) consisted of five males and five females, all with different educational levels. The next age group (50-59 years old, mean age 54,5) also consisted of five males and five females with different education levels. The last group (60-70 years old, mean age 63,3) counted six males and four females, again with different education levels. All of them had Dutch as their native language.
The second group, referred to as the patient group, consists of six male adults (ages 50-65, mean age 57,0) who, in the last year, underwent an awake or non-awake craniotomy due to a brain tumour. These patients had their language tested before and (a couple of months) after surgery, always in a face-to-face situation. Whether these participants experience(d) language difficulties before or after the surgery differed per participant, as is described in Table 1 below.
Table 1. Language difficulties post-surgery in the patient group
The control group participates in both test versions, which we administer at least one week apart from each other. The order in which version A and B are administered is counterbalanced (also shown in Table 2 in Appendix C). The patient-group participants all participate in one test version (B) and are asked whether they want to participate a second time in the other test version after the first test is completed. In the end, three of the six patient-participants agreed to participate in the other test version as well. We initially test all patient participants with the same test version to make sure they all participate in the same test as we do not know how many patient participants want to participate in the other test version. We use test version B instead of test version A first, as B seems to contain less difficult test items.
4.3 Procedure and Analysis
The test is administered over telephone. We first phoned the participants to inform them about the goal of this study, the reason why the DTLB was designed and what is expected of them. We also make an appointment to administer the language test and get verbal permission for the use of the test-results for this thesis project during this conversation. During the second (and third) phone call(s), the participant receives instructions about the test and is assured that (s)he
Patient
Language difficulties
pre-surgery Language difficulties post-surgery
P1 None Only the first week after surgery, he had some word-retrieval
difficulties, but does not experiences this anymore
P2 Yes (word finding) No difficulties after surgery
P3 None Yes, P3 experiences word-retrieval difficulties since the surgery
P4 None No difficulties after surgery
P5 None No difficulties after surgery
P6 None Every now and then he experiences some word-retrieval problems
can always skip a test item, ask more questions during testing or stop the testing during administration without giving an explanation. Importantly, the examiner always asks if the participant can find a comfortable and non-distracting environment to do the test. After testing, only the patient group was asked to give feedback on the test itself and on the test being administered over telephone. Note that we did not ask the patient participants whether they had or have language difficulties until after language testing, so that we did not know about this during administering the test, unless the patient participant told us himself during testing. The test takes 15 to 20 minutes to administer. A phone-call lasts approximately 20 to 30 minutes due to the instructions. Because the patient-group are also asked to give feedback on the test, their phone-call(s) are approximately 45 minutes.
The answers are written on a score form. The score forms of all tests can be found in Appendix B. These score forms also contain the instructions for each test and how the scoring system works.
The analysis of the results is as follows: first, we analyse the control group scores; in order to see whether the subtests of the test-versions differ significantly from each other we present the mean, standard deviation (SD) and range of the results of each subtest. We use the Wilcoxon for a statistic analysis of these results. The same steps are made to see whether the total test results differ significantly from each other. Then, based on the individual results, we make a descriptive evaluation of the test items to get an insight about the errors of the test.
Secondly, we focus on the patient scores: first, we compare the individual patients’ scores of the DTLB with their NPS scores, to see how these scores relate to each other. Then, we evaluate the procedure of the test-battery by describing the feedback we got from the patient group. Finally, in order to corroborate the purpose of test, we compare the range of the control-group and the patient-group. To support this comparison, we calculate the z-score of the patients’ results to see how many SDs their scores deviate from the control group scores.
5. Results
In the current section, the results addressing each research goal will be discussed separately. In section 5.1, we start with the control group results, through which we consider whether the subtests of the test-versions are significantly different from each other. We do the same with the total results of the two test versions. We then show the descriptive analysis of test-items. In section 5.2, we discuss the results based on the patient group scores. We present the results of the comparison of the NPS- and DTLB scores, discuss the feedback on the test-battery the patient group gave and finally, show the results of the speculation of the purpose of the test by comparing the control group scores with the individual patient scores. All test scores of both the control group and the patient group can be found in appendix D.
5.1 Control group results
In this section, we present the control group scores results. We start with showing whether the subtests of the two test-versions of the DTLB differ significantly from each other or not. We also show whether the total test scores of the two test versions are significantly different or not. Then, we give a descriptive analysis of the test-items.
Research objective 1 & 2: control group (sub)test results correlations
To see if the subtests of the DTLB differ significantly from each other, we first look at the mean, range and standard deviation (SD), which are shown in Table 2. Though it is not a statistical analysis, we can see that most means, SD’s and ranges are very similar to each other. The only exceptions are the SD’s of the word retrieval tests, which differ 2 SD’s.
Table 2. Mean, SD and range of test version A and B of the DTLB
In order to check the differences between the individual tests in each test version and the difference between the two tests versions, we used the Wilcoxon test. This particular test was used as we first performed the Spearman Correlation, a non-parametric test, which resulted in a negative correlation between the subtests of each test version and the overall test. Of course, all subtests and total tests could indeed all differ significantly from each other, but the negative
Test Mean Standard Deviation Range
Word retrieval A 38.4 3.58 20 - 40 Word retrieval B 38.66 1.67 34 - 40 Repetition A 17.77 0.43 17 - 18 Repetition B 17.47 0.63 16 - 18 Phonological fluency A 2.88 0.76 1,5 - 4,5 Phonological fluency B 2.93 0.61 2 - 5 Semantic odd-word-out A 19.8 0.66 17 - 20 Semantic odd-word-out B 19.57 0.94 16 - 20 Semantic fluency A 5.92 0.93 3.5 - 7 Semantic fluency B 6.07 1.03 3 - 7 Sentence completion A 19.5 0.86 17 - 20 Sentence completion B 19.63 0.56 18 – 20
Total Score Version A 104.26 4.19 101 - 110
Total Score Version B 104.33 3.27 91.5 - 109
outcome could be for a reason: the negative correlation could be caused by the lack of data points (test items) or the small number of participants. Therefore, we decided to leave the Spearman Correlation out and focus on the Wilcoxon test. We accept the tests not to differ significantly from each other if the scores across patients do not significantly differ from each other at an alpha-level of 0.5. The significance levels (P-values) of the Wilcoxon test can be found in Table 3.
Table 3. Wilcoxon p-values outcome of the subtests and total test scores DTLB
As table 3 shows, the word retrieval test, the phonological fluency test, the semantic odd-word-out test, semantic fluency test and sentence completion test show P > .05, which makes it possible to conclude that these subtests of both test versions do not differ significantly from each other. However, the repetition tests had a p-value of .04, which means that the two repetition tests do significantly differ from each other. Furthermore, the comparison of both complete test versions with the Wilcoxon tests shows a p-value of .48, which means that also the overall tests are not statistically different.
Research objective 3: descriptive analysis of the test items
The descriptive analysis is presented per subtest. The test items of each test are discussed in the same order as they are administered, starting with word retrieval, followed by repetition, semantic odd-word-out and sentence completion. The fluency tests are left out of this analysis as they consist of only one test item each. The pre-test comprehension screening is also not discussed, given that all 30 control group participants answered all questions correctly.
The test item analysis is structured as follows: the test items which were not answered with the target answer, once or more, are shown and analyzed.
Word retrieval: before the evaluation of the test items, we need to explain more about the scoring procedure. Initially, we scored the control group responses as follows: zero points for an inappropriate answer, one point for the target-answer given after four seconds and two points for the target-answer given within four seconds. Afterwards, however, we decided to change the scoring system only for the current study as we discovered that some test items could have more than one correct response. Now, the participants got 2 points for a target-answer or a synonym of this target-answer given within four seconds (e.g. kasteel ‘castle’ and paleis ‘palace’). One point was still given for a correct target-answer given after four seconds, but also when another correct answer was given which was neither the target-answer nor a synonym of the target-answer (e.g. eend ‘duck’ instead of vissen ‘fish’). Zero points were given when an answer was incorrect or not given. Table 4 shows the inappropriate answers for both word retrieval tests.
Subtest P-value
Word retrieval A & B .31
Repetition .04
Phonological Fluency .81
Odd-Word-Out .33
Semantic Fluency .44
Sentence Completion .51
Table 4. Errors in the word retrieval test items
In test version A, the test items that really stand out are vissen ‘fish’, which got an inappropriate response thirteen times and kasteel ‘palace’, which got an inappropriate response fifteen times. For the inappropriate target-answers given instead of vissen, all participants still got one point, as their answers were not incorrect. All participants who said paleis instead of kasteel got two points, because paleis is a synonym of castle. The only test item that appears to be not easy to answer at all is kledingstuk ‘piece of clothing’, as two participants gave a description of what a kledingstuk is, one responded with ‘I don’t know’ and two participants took longer than 4 seconds to answer. We decided to give zero points to a description of the target answer, as the instructions clearly say we only need to hear one word as an answer. All other inappropriate answers were given only by one, two or three participants. In version B,
Target-answer answered inappropriately Responses Number of inappropriate responsesa Points given (max: 2 points) Version A School Lokaal 1 1 Wonen Slapen 1 1 Leven 2 1 Alles 1 0 Vissen Kikkervis 3 1 Eend 6 1 Kikker 4 1 Kasteel Paleis 15 2
Krijt Whiteboard stift 1 1
Kledingstuk Explanation given 2 0
No answer 1 0
(Improved) answer after four seconds 2 1
Duinen Improved answer after four seconds 1 1
Version B Geld Pinpas 2 1 Politie Agent 1 1 Stad (Hoofd)stad 7 2 Dokter Huisarts 1 1 Verpleegster 1 1
Personal answer given 2 0
Brug Improved answer after four seconds 1 1
Strand Improved answer after four seconds 1 1
Kust 1 1 Maasvlakte 1 1 Duinen 1 1 Ochtend In de morgen 1 0 2e dagdeel 1 0 Sigaret Sigaar 2 1
Koffers Tas/ rugzak 4 1
Zinken Stranden 1 1
Improved answer after four seconds 1 1
Potlood Gum 2 0
Kieuwen Schubben 1 0
Vinnen 2 0
Mond 1 0
Kitten Answer after four seconds 1 1
Wrong answer given 1 0
more participants gave an inappropriate response to a test item, though these are spread over more target-words. The test items that really stand out are the answers given instead of stad ‘city’, which seven participants answered with hoofdstad ‘capital’. They got two points for this answer, as it is not incorrect (Amsterdam is the Netherlands’ capital). Another test item that was answered inappropriately four times is koffer ‘suitcase’, which was answered with either tas (bag) or rugtas (backpack). Participants got one point for this answer. All other test items that got an inappropriate response, only got one or two non-target word answers.
Repetition: in both test versions, all (existing) words were repeated correctly. Most errors were made in the repetition of the non-words. In Table 5 only the errors made in this test are shown, including the number of times the target answers were repeated incorrectly.
As table 5 shows, in version A, only the non-words were repeated incorrectly: ánamo was repeated incorrectly twice, while frimótica was repeated incorrectly five times. The (existing) words and sentences of this test-version were repeated correctly by all thirty participants.
In version B, the Repetition test shows a higher number of errors. Again, all (existing) words were repeated correctly by all thirty participants. Both non-words were repeated incorrectly a couple of times as ínima was repeated incorrectly three times and slanugári eleven times. The mistake we heard most when slanugári was repeated incorrectly was the absence of the /l/ in the onset consonant cluster /sl/. Also, in this version, the question-sentence was not repeated correctly twice. However, it was only one word that was not correct (once someone said hij ‘he’ instead of jij ‘you’ and someone said bestek ‘cutlery’ instead of lepel ‘spoon’.
Table 5. Errors in repetition test items
Version A
Non-words Number of incorrect repetitionsa
ánamo 2
frimótika 5
Version B
Non-words Number of incorrect repetitionsa
ínima 3
slanugári 11
Sentences
In welke la leg jij de lepel? 2
a: Number of inappropriate target answers is out of 30 responses.
Semantic odd-word-out: this test did not show many errors, which resulted in an almost perfect score. In Table 6 all inappropriate responses can be found.
Most of the points not given were because of the explanation being missing. Land ‘country’, ijs ‘ice’ and lepel ‘spoon’ were all given without an explanation by the same participant. Only one point was given for these responses. Brief ‘letter’ was given as an answer once too, but also without explanation as to why it does not fit the other two words. Hond ‘dog’ was given once as an incorrect answer by another participant who did not make any other errors in this test. All other test items got a correct response from all thirty participants.
In version B, target-answer bal ‘ball’ gave the most errors: one participant gave the right target-answer without an explanation, while four participants could not give the right answer and got zero points for that test-item. For three of the four this test-item was the only one they
answered incorrectly. The fourth participant also got a half incorrect answer, because she could say which word did not fit with the other two but could not give an explanation for the words hoed ‘hat’ and maand ‘month’. In this version, someone changed his full answer to another word, as he first answered that potlood ‘pencil’ was wrong (without an explanation). Only when I asked him again why pen was the odd-word-out of the three, he corrected himself and said appel ‘apple’ with a correct explanation, for which I gave him 1 point. All other test items got a correct response from all thirty participants.
Table 6. Errors in semantic odd-word-out test items
Word 1 Word 2 Word 3 Target Response Number of
inappropriate responses a:
Points Given (max: 2 points) Version A
Kerk Hond Paard Kerk Dog, because it is not a
building 1 0
Peer Appel Land Land Land; no explanation 1 1
IJs Kachel Oven IJs IJs; no explanation 1 1
Emmer Brief Gieter Brief Brief; no explanation 1
Maan Vork Lepel Maan Lepel; no explanation 1 1
Version B
Bal Pop Zeep Zeep Bal; because it is
round. 1 1
No answer 4 0
Mond Oor School School School; no explanation 1 1
Hoed Straat Weg Hoed Hoed; no explanation 1 1
Roos Maand Tulp Maand Maand: no explanation 1 1
Pen Potlood Appel Appel Corrected after four
seconds
1 1
a: Number of inappropriate target answers is out of 30 responses.
Sentence completion: finally, Table 7 shows the sentences which were hardest to finish for some participants, for which fewer points were given.
In version A, interestingly, only one sentence could not be completed by two participants, while all the other sentences were completed (whether this was with or without hesitation). As can be seen in the table below, four of the five sentences which were not finished perfectly, were finished correctly but with a lot of hesitation. Only Zij belde niet, omdat… ‘she did not call, because…’ was not finished at all by two participants. All the other sentences were finished perfectly and without hesitation by all participants.
Version B shows a similar pattern as version A, although all sentences were completed, but not all without hesitation. In addition, two sentences were completed correctly first, but the participant changed the words afterwards, which happened, for example with de auto staat in de straat…. nee in de garage ‘The car is in the street… no.. in the garage’. In this case, one point was given instead of 2 as it was seen as a form of hesitation.
Overall, what stands out in term of errors is the number of inappropriate answers in the word retrieval test and the number of incorrect repetitions of slanugári in the repetition test. The semantic odd-word-out test and sentence completion are well made.
Table 7. Errors in sentence completion test items
5.2 Patient group results
In this section, we focus on the individual patient results. We start with the comparison of three patients’ DTLB scores and compare them to their NPS scores, followed by an overview of the feedback that was given. Finally, we will compare the control group scores and the individual patient scores to corroborate the purpose of the battery, that is to see whether the test-battery scores show a distinction between those with and without language problems.
Research objective 1: NPS- and DTLB scores comparison
We compare the individual patient test scores of the DTLB to their NPS scores, to see how they relate to each other. Only three patients’ scores can be compared as only these three patients did the NPS due to an awake craniotomy. The time between the test administration of the NPS and DTLB varies from one year to six months. Note that the NPS is administered face-to-face and the scores are from before the surgery and the DTLB is administered over telephone and the scores are from after the surgery. The scores of the patients that we use are those of P1, P2 and P6 (see Table 1 in section 4.2 for their language status before and after surgery).
The tests that we compare are the following ones: the Token Test (De Renzi & Vignolo, 1962) of the NSP and the comprehension screening of the DTLB, which both indicate whether the participant has a comprehension difficulty. The Token test was created to provide a sensitive and specific measure of auditory-verbal comprehension (Swihart, Panisset, Becker, Beyer & Boller, 1989). In the Token test, the participant has some circles of different colours and sizes in front of him to which he gets related questions, such as ‘point at the big, blue circle’. Then, the short version of the BNT (Kaplan et al., 1983) of the NPS and the word retrieval test of the DTLB are compared. Both tests test whether the participant has problems in word finding. The semantic fluency tests of both the NPS and DTLB as in both test-batteries professions and animals are used as semantic categories. Finally, the NPS WAIS Similarities (Wechsler, 2012) test-scores will be compared with the semantic odd-word-out test. In the WAIS Similarityies test, the participant is shown two everyday objects or concepts and the participant must indicate
Sentence
Number of inappropriate
completions a: Reason (max: 2 points) Points given
Version A
De vrouw probeerde… 1 Hesitation 1
Elke week… 1 Hesitation 1
Zij belde niet, omdat… 2 No answer given 0
Toen de vaas brak… 3 Hesitation 1
Het meisje hamert erop dat… 5 Hesitation 1
Version B
De auto staat in 1 Change of words 1
De buurman denkt dat 1 Hesitation 1
De vrouw op het strand 1 Change of words 1
Iedere dag 3 Hesitation 1
Hij kwam niet omdat 2 Hesitation 1
Toen de boom omviel 2 Hesitation 1
