Menu
Menu
Menu
Menu
The dearth of locally developed measures of language makes it difficult to detect language and communication problems among school-age children in sub-Saharan African settings. We sought to describe variability in vocabulary acquisition as an important element of global cognitive functioning. Our primary aims were to establish the psychometric properties of an expressive vocabulary measure, examine sources of variability, and investigate the measure’s associations with non-verbal reasoning and educational achievement. The study included 308 boys and girls living in a predominantly rural district in Kenya. The developed measure, the Kilifi Naming Test (KNT), had excellent reliability and acceptable convergent validity. However, concurrent validity was not adequately demonstrated. In the final regression model, significant effects of schooling and area of residence were recorded. Contextual factors should be taken into account in the interpretation of test scores. There is need for future studies to explore the concurrent validity of the KNT further.
The data in the current study were derived from a larger cross-sectional study on the development of appropriate methodologies to assess executive functions, motor skills and the home environment in school-age children in a rural district. To satisfy the inclusion criteria for the main study, children had to speak one of the local dialects or Kiswahili as their first language; demonstrate physical ability to perform the tasks; and, be resident within the study area. In the following section, we first describe the development of the confrontation naming test: the Kilifi Naming Test (KNT). We then explain how we established the psychometric properties of the KNT in terms of its reliability, validity, and sensitivity. The study was conducted in the former Kilifi District (currently known as Kilifi County) that constituted the former Coast Province in Kenya. Over 80% of the county’s inhabitants belong to the Mijikenda ethnic group, which comprises nine sub-groups with similar dialects closely related to Kiswahili, the lingua franca and country’s national language. Approximately 50% of the labor force is engaged in subsistence farming mainly growing maize, cassava, and cowpeas. A few engage in livestock farming of cattle, sheep, and goats. Coconuts, cashew nuts, and mangoes are the main cash crops (Kahuthu, Muchoki, & Nyaga, 2005). The majority of the population (70.8%) experiences absolute poverty (Commission on Revenue Allocation, 2013), manifest as limited access to basic needs, and an inability to meet the minimum cost of food and nonfood items essential to sustain life (Kahuthu et al., 2005). High poverty levels are also associated with school drop-out, deteriorating health conditions and poor literacy levels (Kahuthu et al., 2005). The average literacy level in the county is 68.2% (Commission on Revenue Allocation, 2013), which compares poorly with the national rate of 83.9% (UNESCO Institute for Statistics, 2011). About 21% of the population has never attended school, with the greater proportion being women. Family structure is largely collectivist, with extended families including members of several generations, living together in homesteads comprising four to ten mud-walled thatched houses. More recently brick-walled and iron-roofed houses have become more common. It is imperative to point out some salient features of the setting within which the study was conducted as these may have had a covert influence on the vocabulary development of children. Boys have more unstructured time than girls and most of their time is spent unsupervised outside the household. They thus have more opportunities for social play with same-gender peers (Awiti, 2011; Wenger, 1989). Girls on the other hand spend more time with older females within the homestead engaged in chores such as looking after infants and toddlers (Wenger, 1989). Such activities are unlikely to promote vocabulary development. It is not unusual for a child growing up in this context to learn three different languages: their mother tongue, Kiswahili the national language, and English, the language of instruction in elementary schools. Traditionally as in most African societies, a child is expected to be obedient, quiet, and undemanding in the presence of adults and talkativeness is frowned upon (Nyasani, 1997). Children could not initiate conversations and were taught to avoid asking adults questions as it would seem as though they were challenging them and attempting displays of superior knowledge. Adults rarely engage in any play activities with children (Mbise & Kysela, 1990) as most adult–child communication is for the purpose of giving instructions (Wenger, 1989). We recruited children from the catchment area of five local schools distributed across neighborhoods in the district ranging from sparsely populated (64 persons/sq.km compared to the district average of 114 persons/sq.km) to densely populated (325 persons/sq.km) semi-urban areas. Children in school and out of school were included in the sample. A description of the full study sample is presented in Table 1. A total of 308 children (51% girls, n = 160) were included in the tool development (n = 100) and validation process (n = 208) of the current study. Their ages ranged between 5 and 13 years (M, SD = 9.08, 1.12). Nearly one quarter (n = 74) of the sample had linear growth retardation (or stunting) while 11.4% (n = 35) had no schooling experience. Within household status, more than one third (39.9%, n = 123) fell into the category of “least wealthy” (Level 1). The derivation of the household wealth index is explained in a later section. The Kenya Medical Research Institute/National Ethics Review Committee (KEMRI/NERC) provided ethical clearance for the study. Permission to visit schools was obtained from the District Education Office. We explained the purpose of the study to the head teachers of selected schools and then sought their permission to recruit children. We also held meetings with community leaders, elders, and parents (and guardians) of selected pupils to explain the purpose of the study. We presented information regarding the study in the language with which parents were most familiar. After each meeting, a screening questionnaire was administered to parents/guardians to establish if selected children met the study’s eligibility criteria. We then obtained written informed consent for their children’s participation. We explained the nature of the assessments to the children prior to test administration. All the selected children assented to their participation in the study. We developed a confrontation naming test similar to the BNT in terms of structure, administration, and scoring that would be appropriate for school-age children (eight years and above). The BNT also provided an appropriate framework for length, and was used to suggest possible categories of words. In designing the test, we had a number of objectives. The test needed to: be simple and quick to administer; require no specialized equipment; and, elicit clear, responses that are easy to record. To develop the test, we followed the 4-step systematic test adaptation procedure outlined by Holding, Abubakar, and Kitsao-Wekulo (2009). Step 1: Construct definition. An extensive review of existing literature did not reveal any studies reporting the concept of confrontation naming within the sub-Saharan African context. We therefore obtained the following simple definition from a reference book: the ability of children to name common words depicted through pictures presented to them. Step 2: Item pool creation. We identified a list of words that would be suitable for creating a measure of vocabulary development in a rural community of school-age children. Figure 1 summarizes the procedures followed in creating the item pool and provides details of the number of participants included at each stage. We supplemented words from existing measures of child development based on previous research (Carter et al., 2006; Carter et al., 2005; Holding et al., 2004) with words obtained from surveys of children’s language. In one of the surveys, 176 schooling and out-of-school children whose ages ranged between 5 and 17 years were asked to list as many words as they knew in any language (the local language, Kiswahili, or English). We also considered the input of 54 community members and an “expert panel” comprising a psychologist, a nurse, an educationist, and a linguistics professional. Steps in creation of KNT item pool. Step 3: Developing the procedure and training examiners. In the next stage, we formulated semantic stimulus cues. These were phrases used to prompt the respondent to produce a correct response if they were not able to name a picture accurately. For example, if the child misperceived a saucepan as a cup, s/he would be given the cue that the item was used “for cooking”. The appropriateness and utility of these stimulus cues was then tested on a group of children by presenting the pictures first, with and then, without stimulus cues. Some of the stimulus cues were rewritten to improve clarity. In most cases, the frequency of correct responses was higher when the items were presented with stimulus cues. We then developed a list of acceptable responses for each item to reduce ambiguity in scoring as multiple possible names would make it difficult to score an item reliably. We trained four local child development assessors to administer and score the KNT. All the assessors had completed secondary school education and were familiar with the local dialect. At the start of the training, we explained the purpose of the study, and the importance of adhering to the tool administration protocol. This training ensured that the test was consistently administered in a standardized manner by all assessors. The assessors were provided opportunities for practice among themselves, and with nonstudy children under the direct supervision of the first two authors. They received feedback on how to improve their test administration techniques until they got to a level where they were well-versed with the procedures. As explained in the procedures followed for the larger study (Kitsao-Wekulo, Holding, Taylor, Abubakar, & Connolly, 2013), observations on tool administration continued until more than 90% agreement with any one of the trainers was reached by each assessor. Step 4: Evaluation of developed schedule. Sixty items were tested on 75 nonstudy children and then ordered according to frequency of correct responses. These items were then administered to the first 100 children (according to the identification numbers assigned to them) enrolled in the current study. Fifteen items that were found to be too easy (more than 90% correct responses) were taken out of the list. Slightly more difficult items (N = 19) were then tested on another 16 nonstudy children. The easy items that were discarded from the original list were replaced with sixteen of these difficult items. In total, 191 children participated in this evaluation. The final version of the KNT had 61 items ordered according to their difficulty level. The names of the objects range in difficulty from simple, high frequency words (easy) to rare words with low frequency (difficult) of occurrence. The 61 items of the KNT were administered as part of a neuropsychological battery to 208 children from the main sample. (These children were not included in the process of developing the KNT as described in Step 4 of the adaptation procedures). The full battery (See Appendix 1 for a brief description of the tests) comprised tests of executive function, verbal and working memory, verbal/visual selective reminding, learning, auditory, and visual sustained and selective attention, and nonverbal reasoning. The battery did not include any other measures of expressive language. The tests were modified for cultural and linguistic appropriateness (See Kitsao-Wekulo et al., 2013, for a detailed description of test modifications made and test administration procedures followed) and the battery was on average administered in a single session of approximately 2-hour duration, including two 10-minute breaks. In order to maximize participation, testing was conducted using an interactive play-like style which has been found suitable for children having minimal experience with standardized testing or where a significant proportion does not attend school (Alcock et al., 2008; Holding et al., 2004; Kitsao-Wekulo et al., 2013; Nampijja et al., 2010). Administration of the KNT. In the KNT, the child is asked to spontaneously give one-word responses when presented with a black and white line drawing of a familiar object. The assessor pointed to a picture or part of a picture which the child was required to name. Testing was conducted within a room or in a quiet outdoor location at a school near the child’s home. All children were tested individually but within sight of other children to minimize test anxiety. The items were administered to children in a standard order beginning with item 1. A stimulus cue was provided when no response was given, the child stated that s/he did not know the name or the item was incorrectly perceived. No time limits were imposed for responding. Similarly to the cut-offs used in the original procedure for the administration of the BNT (Kaplan et al., 1983), if a child failed to correctly name any objects on six consecutive trials, the test was discontinued. Several children (n = 167; 80.3%) met the criteria for discontinuation. The test took between 10 and 20 minutes to administer. A second administration of the KNT was completed within 6–8 weeks of the first assessment. Scoring. If a child provided the correct response, that is, the name of the item as indicated on the record sheet, the assessor recorded “C” on the record sheet. An erroneous response that was spontaneously corrected before any cue was provided was also scored as correct. In addition, a score was awarded when the child responded correctly after the cue was provided. Credit was given for a correct answer in any one of three languages: English, Kiswahili, and the local dialect. Making provisions for responses in different languages is especially important in settings where children grow up using several languages (Alcock et al., 2008) as they are likely to pick up vocabulary terms in more than one language. A score of “1” was awarded for all correct responses. If provision of a stimulus cue did not result in a correct answer, that is, the child’s response differed from the names specified on the record sheet, the word that was produced by the child was recorded verbatim as a nontarget word response. All scoring was checked by the assessor who administered the test and then cross-checked by a second assessor. Any disagreements were resolved through discussions. The final score was calculated by summing the number of spontaneously correct items with the number of correct items following a stimulus cue. The maximum score was 61. Other variables. Information on child gender, age, school experience (number of years that child has attended school), and household wealth was collected using a structured interview form. Birth records were used, where available, to confirm the child’s date of birth. For the purpose of the current study, an age variable in 6-month increments was created. School experience was classified according to three categories: no school (non-schoolers), between one and two years of school (recent attenders), and more than 2 years of school (longer-term attenders). A composite index of household wealth that divided the sample into three approximately equal groups—least wealthy (Level 1), moderately wealthy (Level 2), and the most wealthy (Level 3)—was derived from six socioeconomic indicators: maternal and paternal education, maternal, and paternal occupation, type of windows in the child’s dwelling and ownership of small livestock. The wealth index score was calculated by summing the values assigned to each of these indicators as detailed by Kitsao-Wekulo and colleagues (2013) in an earlier study. Using a stadiometer, we measured children’s heights to the nearest centimeter to calculate their nutritional status designated as with or without linear growth retardation. Linear growth retardation, a marker for nutritional status, was defined as height that was more than two standard deviations below levels predicted for age according to the World Health Organization (WHO) reference curves for school-aged children (World Health Organization, 2007). Area of residence was characterized as rural or urban according to the most common settlement within the school catchment area. To quantify reading skills, we administered reading (letters, words and sentences) tests (Bhargava, Jukes, Ngorosho, Khilma, & Bundy, 2005) to a subset of 135 children in our study sample. In the reading task, children were required to select real letters, words and sentences from lists which included fake forms. This test was designed to measure children’s comprehension and the scores were adjusted for incorrect answers. We summed the scores across reading tests to obtain the reading score. A modified version of the Colored Progressive Matrices (CPM: Raven, Court, & Raven, 1998) test was administered as a measure of nonverbal reasoning. In the CPM, the child is required to complete a pattern by selecting the appropriate missing symbol from a set of six alternatives. A detailed description of this test is provided in the Appendix 1. A descriptive analysis of the background characteristics and distribution of scores was conducted. Item difficulty, defined as the percentage of correct responses for each item, was assessed to determine whether the items included on the test had appropriate difficulty levels (easy, medium, hard). Internal consistency and test–retest reliability of the KNT were quantified using the intraclass correlation coefficient (ICC). The Pearson product moment correlation coefficients (Pearson r) were computed to examine the relationship between the KNT and nonverbal reasoning (CPM), and the KNT and reading skills. These associations were used as measures of convergent and concurrent validity, respectively. As there were significant effects of maturational changes on performance (Figure 2), the KNT scores were regressed against age to produce age-corrected scores. Using the standardized KNT scores as the dependent variable, hierarchical linear regression analyses were conducted to understand which among the independent (background) variables (age, gender, nutritional status, household wealth, school experience, and area of residence) explained the observed variance in KNT scores, and to explore the forms of these relationships. In the first step of the hierarchical regression, we inserted school experience and area of residence in a stepwise fashion as the main predictors. In the second step, we inserted household-level variables (household wealth and nutritional status) as these have been shown to have an influence on vocabulary scores. The third step involved entering the child-level characteristic (gender) as a predictor. Our justification for entering the variables in this order was based on the strength of the associations between the variables and KNT scores. The proportion of variance in naming performance accounted for by each of the background variables was quantified using R 2. Alpha levels were set at <.05 for statistical significance. Age effects on KNT performance.
N/A
