Using the Barnes Language Assessment with Older Ethnic Minority Groups

ABSTRACT (word count 227)

Objective

There are many issues concerning the assessment of older people from ethnic minority groups, the most significant being the language barrier experienced by those whose English is an additional language (EAL). This study aimed to test the hypothesis that EAL participants would score less well than those with English as a first language (EFL) on the subtests of the Barnes Language Assessment (BLA), elucidate the reasons for any such differences and discuss the implications

Methods

The Barnes Language Assessment (BLA) is an accurate tool providing information about expected patterns of language in different dementia syndromes. This study compares the performance of EAL participants with EFL participants. The BLA was administered to 144 participants, divided into sub-groups with respect to age, gender and educational background, none of whom had a working diagnosis of dementia.

Results

Results suggest that EAL speakers performed less well compared to EFL speakers when other variables were matched. Significantly better BLA scores, at the one per cent level, were found in both EAL and EFL groups with higher educational achievement for eight of the fifteen sub-tests.

Conclusion

Differences were found in performance on the BLA between EAL and EFL participants. The degree of difference between EAL and EFL speakers decreased as educational achievement rose. The consequences of these findings for service delivery and the problems of recruitment of older EAL participants are discussed.


INTRODUCTION

The number of older people from minority ethnic communities in the United Kingdom has risen sharply over recent decades (Odutoye and Shah, 1999) and are predicted to continue to do so. It is recognised that social research has not yet fully explored the issues facing black and minority ethnic elders (Lindesay et al 1997; Richards et al 2000). The results of such research may have implications for policy development and service provision.

Studies suggest that ethnic minority populations have a greater incidence of dementia, the cause attributed to unidentified genes or other risk factors (Tang et al 1998). Dementia is diagnosed later in these groups (Ayalon and Arean 2004), so ethnic elders are less likely to access important early interventions and support. A number of explanations are given for this delay in diagnosis, such as a reluctance to access services (Eolas 1999), difficulty getting appropriate information, and lack of knowledge about services, together with a lack of appropriate test materials (Richards et al 2000). Even when interpreters or bilingual interviewers are used, there seem to be particular problems around the reliability of the interview process in ethnic minority groups (Rait et al 2000) (Shah1999).

Finally, there are differences in culture and belief, with some traditional approaches to health and well-being conflicting with western values (Butt and O’Neil 2004). The effects of ageing may be experienced and treated in a different way in minority cultures.

Language, age and education

Ethnicity and bilingualism are not the same; although many people in ethnic minority groups are bilingual. However, language barriers are often cited as the cause of difficulties when assessing ethnic elders.

Age is significant when considering language barriers in ethnic communities. At present ethnic individuals of over 85 years are more likely to be monolingual than younger people. In the future, although this “older old” group will expand, the proportion with some knowledge of English may increase. However barriers in the use of English language assessments may continue to exist, as may the effect of lack of formal education.

Education level is known to be a significant influence on ethnic elders’ performance on cognitive testing (Richards et al 2000; Lindesay et al 1997). However, education level alone is an inadequate indicator of performance when considering a group with such varied ‘education experience’. This is due to an absence of internationally standardised educational levels.

The Barnes Language Assessment

The initial research leading to the development of the BLA was carried out by a group of speech and language therapists, specialized in working with older people. The test provides a valuable screen and baseline measurement of language skills (Bryan et al 2001). Feedback from participants and clinicians was that the project highlighted the need to distinguish between EAL and EFL speakers, as well as to control for variables other than age and gender (Bryan et al 2006). These were social class, occupation and education levels. There are indications, based on the literature, that EAL speakers will do significantly worse on language tests when compared to EFL speakers. The second project, carried out between 2002-2005, established robust normative data for EFL elders, and norms for EAL elders. This study aimed to test the hypothesis that EAL participants would score less well than EFL participants, elucidate the reasons for any such differences and discuss the implications.

METHOD

Sampling

144 healthy community living participants over the age of 50 were recruited from a variety of settings including day centres, community groups, work-places and via word of mouth. The project was advertised by A4 posters, letters and personal contact. Interviews were carried out in participants’ homes, central locations (eg. day centres) and at the University of Surrey. Criteria for inclusion in the study included no evidence of a dementia, no significant current depression, no significant neurological history, no reading or writing impairment. The criteria were applied using screening tools described below. The 144 participants were divided into EFL (89) and EAL (55).

During the analysis stage, the EAL participants were grouped according to ethnic background; European, Asian (including Chinese and Japanese), Other e.g. Creole and Guyanese, and level of education (see Table one)

The aim was to identify any patterns emerging in the data, particularly error patterns, although it is acknowledged that such broad groupings will need to be refined. Discussion of emerging error patterns particular to ethnic groups is the subject of a future paper.

Table 1 here

Participants were reassured throughout the process that the project was to collect data about ‘normal’ functioning and not in order to identify any pathological process. However, if participants expressed concern about their performance they were advised to contact their GP. Only one individual had a significantly impaired performance and it was found that she was being monitored by her GP; her data was eliminated.

Tools

Before the BLA was administered, a pre-assessment screen was carried out to eliminate any significant conditions e.g. dyslexia or depression. The pre-assessment contained the following screens; Mini Mental State Examination (Folstein et al,1975) to exclude cognitive impairment; BASDEC (Brief Assessment Schedule Depression Cards) (Adshead et al, 1992) to exclude depression; WRAT (Wide Range Achievement Test) (Jastak and Jastak 1978) to exclude reading and writing problems; Health Screen to establish the presence of any neurological conditions.

The BLA is made up of 15 subtests. The subtests are grouped into expression (5 subtests), comprehension (3 subtests), reading and writing (3 subtests), memory (2 subtests) and executive function (1 subtest). Key areas of language functioning are included in each modality e.g. word fluency, naming, word and sentence comprehension, word and sentence reading and writing.

Data Analysis

Analysis was undertaken on two levels. Using SPSS version 15, first summary statistics were extracted and secondly bivariate analysis was conducted on the BLA sub-tests with respect to EFL /EAL. Analysis of education levels, age and gender were also conducted. Where necessary, the variables were tested for normality (Shapiro-Wilk) and approrialte tests (t-tests, Mann Whitmey U, Spearman’s rank) were performed.

RESULTS

The results confirmed the hypothesis that EAL speakers would perform less well on the BLA compared to EFL speakers. Gender was not found to be a significant variable at the one per cent level in relation to performance on the BLA, and age was only significant at the one per cent level for two sub-tests (verbal fluency animals and trail time). Both of these are timed tests. However, significantly better BLA scores at the one per cent level were found in both EAL and ESL groups with higher educational achievement for eight out of the fifteen subtests (spoken word to picture matching, verbal fluency S and animals, word definition, TROG, forward digit span, sentence writing errors, story re-telling) two further sub-tests were significant at the five per cent level (spelling to dictation and following commands).

The table below presents summary statistics across langugae background and education groups

Table 2 here

Although EAL speakers’ mean scores were usually lower than scores of the EFL participants (see Table 2), there were some subtests where EFL/EAL scores in the higher education sub-groups (O/A levels and diploma/degree) were similar e.g. Following Commands and Sentence Writing, Picture Description and Word-Picture Matching.

Generally, scores increased for both EFL and EAL when the subject’s education levels were higher. Differences between EFL and EAL tended to reduce as the education level rose i.e. there was a greater difference between EFL and EAL in the no qualifications sub-group.

The no qualification EAL sub-group did poorly on verbal production sub-tests (Picture Naming, Word Fluency, Word Definition ) when compared with their EFL counterparts and other EAL speakers in higher education groups. On a test of Picture Description, this sub-group’s score differed from all the other groups (2.5, compared to 4 for other groups) and Single Word Spelling was significantly lower than all other groups. Also, in the no qualification EAL sub-group, Trail Test time was much higher than other groups (a higher score indicating less ability).

DISCUSSION

The findings show that in order to effectively assess a wide population of older people that includes ethnic elders, adapted norms on formal language tests are needed. This finding is supported by the differences found between the EAL and EFL groups tested on the BLA. The EAL group did less well on a language test than their contemporaries in the EFL group. The results show that education is a significant factor in EAL performance, and that the higher the level of education, the less the difference between EAL and EFL performance.

The sub-group that did consistently worse was the EAL group with no formal qualifications. This could be explained by a number of factors. The first may be unfamiliarity with the test format, so poor Trail Test scores (time and error) could be due to unfamiliarity with the test concept (a type of puzzle) that may have strong cultural bias. The fact that tests of verbal output (expression) were more poorly performed by EAL sub-groups lends support to the view that some ethnic older people will have limited functional use of English, poor vocabulary and reduced use and understanding of more complicated syntax. Those with a higher levels of education are more likely to have worked in the UK, have required English as part of their job specification, and/or have been exposed to some formal English teaching.

There are therefore two types of EAL individuals, for whom different approaches to assessment and management are required, one of whom has some functional English, the other, very limited or no functional English.

For those with functional English, consideration of error profiles is also an important finding from the study (to be expanded in a future paper). It is likely that some ‘errors’ found on subtests such as reading aloud may be based on regular patterns of pronunciation within certain language cultures. Understanding this would reduce the possibilities of “false positive” diagnoses, but further analysis is needed to provide a framework within which to understand and measure such error patterns. This work could be subject of future research.

Surprisingly age alone was not a significant variable. The ‘older old’ in our study tended to fall into the EAL no qualifications group and therefore, as described above, were the worst performers. However, data analysis suggested this was due to lack of knowledge of the English language rather than age per se. This does not negate the fact that there is a group of older ethnic people for whom accessing services remains problematic due to lack of English and for whom using a tool like the BLA would not be appropriate.

Another finding was the difficulty in recruiting participants from minority ethnic groups. The problem increased with age, the 80+ group being the most difficult to recruit. The method of recruitment varied from contacting agencies such as local day centres, facilities for older people and exercise groups, to word of mouth via friends and work contacts. It was assumed at the beginning of the research that older age recruitment would be easier than it was. The result was that the target number of participants (5) in some EAL sub-groups was not achieved, notably the 80+ group.

One factor that could have contributed is a suspicion amongst older ethnic elders about participating in research and answering questions. This may be due to different cultural attitudes to ageing, and resistance to pressure to conform to a different culture not fully understood. Some participants may have felt being tested meant being “checked on”, while others were unwilling to disclose frailties. Some may have felt that because you are old there is no value in you being tested. Elsewhere, a lack of respect or understanding of culture has been attributed to the poor development of relationship between researchers and minority elder participants (Brangman 1995; Bedolla 1995.)

A functional level of English is required to complete the BLA. Many older people from an ethnic background may have limited English, relying on family to translate where necessary, and mixing only with those from the same ethnic, and therefore language, background. This was particularly true of the Chinese community, and is supported by evidence (Yu 2000).

Also, as noted earlier, there may be cultural differences amongst ethnic groups, not only with regard to attitude to research but also in terms of the type of tasks that they were asked to do. The differences applied not only to linguistic tasks such as reading and writing, but also to other tasks e.g. picture recognition or the Trail Test (Carter et al 2005). Therefore, the very act of being involved in a formal test may be a factor that prevented people from taking part in the study. A lack of familiarity with the test situation per se may cause difficulties in participating in structured tests or interviews and has been cited as one cause of misdiagnosis in minority elders (Espino et al 1998).