Chapter 5: Methodology
5.4 Instruments used to assess language exposure and proficiency
5.4.1 LBQ and UBiLEC
5.4.1.3 Use of UBiLEC Excel spreadsheet to quantify LBQ data
The UBiLEC (Unsworth, 2011a, 2011b, 2013b) Excel spreadsheet was employed for the purposes of capturing and analysing the exposure data elicited through means of the LBQ that
157 was designed specifically for the purposes of the present study. The UBiLEC is suitable for use with bilingual and trilingual populations aged between 2 and 18 years. As mentioned above, it consists of a LBQ and an accompanying Microsoft Excel document that contains the necessary algorithms to calculate values for a number of different exposure variables, based on the data captured via the LBQ. Unsworth (2011a) provides a detailed description of how the algorithms in the Excel spreadsheet work, instructions on how to enter the data collected via the LBQ into the Excel spreadsheet and a copy of the UBiLEC LBQ. Only a brief description of the type of data that is used to calculate certain variables (rather than the exact algorithms), and the resultant output of a completed UBiLEC Excel spreadsheet will be supplied below.
Recall that the output of the spreadsheet, once all exposure data have been entered, is (i) CAoE to each language as percentage of the child’s waking hours in a typical week; (ii) CLoE to each language in years; (iii) TLoE to each language in years; and (iv) quality (in terms of “nativeness”) of the input the child has in each of her three languages at the current time, given as a value between zero and five.
The data entered into the spreadsheet to enable a calculation of CAoE firstly includes information on every person with whom the child has regular contact at home and outside of the home environment (specifically, at daycare, school and out-of-school care, with a distinction between the person offering the instruction in this context and the other children in this context). This information includes a rating of the ability every person/group of people has to understand and speak each of the three languages, the percentage of the speech every person/group of people directs at the child that is in language A, B and/or C, and the percentage of the speech that the child in turn directs at this person/group of people that is in language A, B and/or C.
Next, the data collected via the LBQ in description of a typical weekday and Saturday/Sunday in the child’s life is entered into the relevant table, each row in this table representing a period of thirty minutes in the case of a week day and an hour in the case of a day at the weekend. Finally, extra information on other sources of language exposure that the child is perhaps not exposed to on a daily basis, is captured in a separate table. Such sources include, but are not limited to, sports/clubs, friends, reading, television and the computer. The output of the spreadsheet includes a differentiation between quantity of exposure calculated
158 using only the home and school context data, and quantity of exposure including these extra sources of exposure.
In the majority of the cases in the current study, cousins and good friends were listed as siblings in order for exposure from these peers to be captured in the calculation that includes only exposure at home and school, and not only in the calculation that includes exposure at home, school and from other sources. This is because, in the low SES areas the participants were sourced from, there are often a number of families residing together in one house or on one property (usually including a house, with a shack or other form of informal dwelling beside it). Also, because of parents’ long working hours and resultant absence or fatigue, children spend a lot of time playing with neighbouring children, either outside in the street or in each other’s homes. As such, the participants in the current study often spend as much time on a daily basis with cousins and/or friends as with siblings.
As illustration of how the UBiLEC spreadsheet calculates CAoE, consider the following example, based on a child who spends 10 waking hours per weekday at a crèche where English and Afrikaans are used equally as the MoIs, and where the children in the classroom use English, Afrikaans and isiXhosa in equal amounts (exposure via instruction being considered to constitute two thirds, i.e. 67%, of the total exposure in the crèche context, and exposure via classmates one third, i.e. 33%): 10 hours x 0.67 = 6.7 hours language exposure per day via instruction; 6.7 x 0.5 = 3.35 hours exposure per day to English and to Afrikaans, respectively, via instruction; 10 hours x 0.33 = 3.3 hours language exposure via classmates; 3.3 x 0.33 = 1.1 hours exposure per day to English, Afrikaans and isiXhosa, respectively, via classmates; 3.35 + 1.1 = 4.45 hours total exposure per day to English and Afrikaans in the crèche context; 0 + 1.1 = 1.1 hours total exposure per day to isiXhosa in the crèche context; 4.45 x 5 = 22.25 hours exposure per week to English and Afrikaans, respectively, in the crèche context; 1.1 x 5 = 5.5 hours exposure per week to isiXhosa in the crèche context. In the case of each language, the aforementioned weekly total number of hours is added to the number of hours of exposure to that language that is received in all other contexts that the child is immersed in during the work week, the latter number being calculated in a similar fashion as the number for exposure in the crèche context. This total number of hours of exposure to a given language during the work week is then added to the number of hours of exposure calculated for the weekend (the numbers for a typical day at the weekend being
159 timed by two). Finally, the sum of the two numbers for each language is recalculated as a percentage of the child’s waking hours per week.
Recall from Chapter 3 that the majority of bilingualism studies operationalise ‘length of exposure’ as the child’s age at testing minus their age at onset of acquisition, a variable Unsworth (2011a) terms “traditional length of exposure” (TLoE). According to this conceptualisation of length of exposure, a child aged exactly four years on the day of testing and who had received exposure to three languages from birth, would have had exactly four years of exposure to each language on the day of testing. Such a conceptualisation does not take into account that, given the nature of the multilingual experience, a trilingual child will most likely have varying amounts of exposure to each of her languages on any given day and over time, making the above four years of exposure incomparable with the four years of exposure that an age-matched monolingual child would have had. It was to overcome this fallacy that Unsworth (2011a) first introduced the notion of ‘CLoE’, a measure that provides a more accurate reflection of a multilingual child’s length of exposure, based on extensive data regarding the child’s language exposure in the past. The UBiLEC spreadsheet provides an estimation of both TLoE and CLoE, the former simply being calculated on grounds of the date of testing, the child’s birth date and the age at which she first received significant exposure to a given language.
The data entered into the UBiLEC spreadsheet for the calculation of CLoE includes, for each year in the child’s life, information on the number of days per week that the child attended daycare/school and the number of hours spent at out-of-school care per week, followed by the percentage of the interaction at daycare/school/out-of-school care that took place in each language, and the percentage of interaction between the child and each person in the home environment that took place in each language. Capturing these data for each year in the child’s life allows the measure of CLoE to be sensitive towards changing crèche contexts and changes in language use patterns in the home (often as a result of changes in the child’s proficiency in given languages, in turn a result of changes in the crèche context). The UBiLEC suggestions (based on international norms) for typical number of hours spent at daycare and typical number of hours spent napping at daycare per day, in each year of the child’s life, were altered to more accurately reflect the lives of the participants in the present study. These children typically spend 10 rather than eight hours at crèche per day, and still
160 nap at least 1.5 hours at crèche daily between the ages of four and five, as opposed to not napping at all.
Unsworth (2011a:3,5) concedes that both the measures of CAoE and CLoE are based on parental report data and therefore serve only as estimations of the child’s actual exposure, these estimations being only as reliable as the reports are accurate. Whereas there is some debate regarding the accuracy of parental recall, Unsworth (2011a:3), like Paradis, Emmerzael and Duncan (as cited in Unsworth, 2011a), argues that as far as linguistic milestones are concerned, parental recall is typically regarded as a “valid tool”. Studies indicating the reliability and validity of this method of capturing data on children’s linguistic behaviour include, among others, Marchman et al. (2004) and Rodriguez et al. (2009). Unsworth (2011a:5) furthermore notes that the accuracy of the parental reports on changes in language use in the home over time may, especially in cases where parents made conscious, strategic language choices, be increased by the fact that parents are asked to report on their own behaviour and not on that of their child.
The final exposure variable that the UBiLEC spreadsheet calculates is quality of exposure. As mentioned above, ‘quality of exposure’ is operationalised in the UBiLEC as the proficiency level of each input provider in the language of input, proficiency being measured on a scale of zero to five (cf. Section 5.4.1.1 above for a description of each level). This conceptualisation is based on the assumption that more proficient speakers provide better quality language exposure, where ‘quality’ may be interpreted as greater grammatical accuracy, richer vocabulary, etc. (Unsworth, 2011a:4).