Malay Malaysian English
4.5 Data analysis
4.5.2 Mean length of utterance (MLU) This section describes the method of analysis for the morphological aspect of Rina’s language development Mean length
of utterance or MLU is the average morphemes per utterance in a child’s speech production. MLU is calculated by counting the morphemes in each utterance, sums over them and divides by the number of utterances, as in example 6 (Lust, 2006, p. 126):
6. Computing MLU dat bunny 2
dat bunny get juice on it 6 sloppy bunny 2
bunny hops 3 Total: 13/4 = 3.25 MLU
Brown (1973) considered MLU as, “an excellent simple index of grammatical development” (p.53) rather than age because the addition of linguistic knowledge usually leads to increase in children’s length of utterances. When he compared the children in his study by age, Brown observed a high degree of variability. This leads him to the conclusion that matching the children by MLU are preferable to get them to be at the similar complexity level than by matching them with age. Brown then developed five stages of linguistic development based on MLU values of 1.75, 2.25, 2.75, 3.5 and 4.0. MLU 1.75 is parallel to Stage I and MLU 4.0 parallel to Stage V. This stage will be elaborated further in Chapter 5 when I compare Rina’s English MLU with Brown’s MLU stages. Since Brown’s groundbreaking work (1973), MLU has been used widely in children language acquisition studies as a gross measurement of grammatical development. In the context of bilingual children, MLU can be used to indicate progress in both languages as well as to show the relative dominance between the two developing languages (Dopke, 1998; Itani-Adams, 2013). In this study, MLU is used to show Rina’s basic progress in English and Malay during the period of investigation.
In this study, MLU was calculated by counting words instead of morphemes from both Malay and English (MLU word). Words rather than morphemes were calculated because of the issue in determining whether a morpheme was used productively by the child (Dopke, 1998; Hickey, 1991). In a comparative study investigating the correlation between MLU word (henceforth MLUw) and MLU morpheme (henceforth MLUm) in a typically developing English-speaking children by Parker and Brorson (2005), they found that both the MLUw and MLUm are perfectly correlated, which shows that MLUw can be used reliably as MLUm. Similar to Dopke (1998), Parker and Brorson also point out that researchers do not have to make the decision whether the child is using the morpheme productively when using MLUw. Conveniently, MLUw can also be used to compare across languages as it will alleviate the concerns about MLU inflation in highly inflected languages (Arlman- Rupp, Van Niekerk-de Haan, & Van de Sandt-Koenderman, 1976; Hickey, 1991; Parker & Brorson, 2005). In this study, given Rina’s language ability during the period of investigation, the researcher believes that using MLUw will give a better estimate of her general language progress in Malay and English than MLUm. To calculate MLU in Rina’s Malay utterances, Malay reduplicated words in her utterances such as
mainan-mainan ‘toys’ and kucing-kucing ‘cats’ were counted as one word. There were also some inflections found in Rina’s Malay words such as prefix ter- (one of the functions of ter- prefix in Malay language is that once it is combined with an adjective, it carries a superlative meaning) in tercantik ‘the most beautiful’ and suffix -nya (-nya is a pronoun form, equal to ‘it’ in English) such as cantiknya ‘it’s beautiful’, banyaknya ‘it’s a lot’, and comelnya ‘it’s adorable’. These words were counted as one word instead of two unit of morphemes. Similarly, English words in Rina’s utterances which has inflections, such as suffix -s and -ing (e.g. cats, dogs, playing, wearing) were calculated as one word. Following Itani-Adams (2013), the utterances that belonged to the following categories were excluded from MLU calculation:
7. a. onomatopoeias
b. counting numbers, (e.g., 1, 2, 3, 4, ...)
c. backchannelling and fillers (e.g., hhm, mhm) d. exclamation (e.g., ah, wow)
e. repeated word and/or particles within an utterance
MLU values were calculated separately according to the utterance type; English, Malay and mixed utterances. It should be noted that since the MLU count is differentiated based on the utterance type, it was not possible to calculate the MLU based on 100 utterances. Therefore, similar to Dopke (1998), Lanza (2004) and Qi (2011), the MLU computed in this research is fewer than 100 utterances. A sample of each of Rina’s utterance type (Malay, English and mixed), together with glosses and some contextual notes are given in Table 4.5. English words are italicised in the mixed utterances.
Table 4.5
Sample of Rina’s utterance type from age 2;10 to 3;10
Time of utterance Utterances Gloss and contextual notes
2;10 Malay Rina makan Rina eats
2;11 English look car Look at the car
3;0 Mixed napa ada monkey? Why is there a monkey?
3;1 Malay meh Rina pegang Come let Rina hold it
3;2 English look love heart (She saw a heart-shaped picture)
3;3 Mixed Rina tak main game Rina is not playing any game
Time of utterance Utterances Gloss and contextual notes 3;5 English present for Cinderella A present for Cinderella (doll) 3;6 Mixed no I want cakap English No I want to speak in English
3;7 Malay ayah beli kereta banyak Father bought many cars
3;8 English I want the pink one (asking for a pink toy)
3;9 Mixed because orang jahat nak datang Because an evil person will come 3;10 Mixed last day we go beli princess cake Yesterday we bought a princess cake
There have been recurring issues raised by researchers concerning the use of MLU as an adequate tool to measure children’s language abilities. Dopke (1998) and De Houwer (2009) argued that as a measure of comparison across languages, MLU might be problematic as MLU is not comparable if the languages involved are of different morphological types. For instance, if a child is acquiring Turkey and Cantonese. if a child is acquiring Turkey (an agglutinative language) with Cantonese (an isolating language), then the child’s MLU is not comparable (Yip & Matthews, 2007). This is resolved by the usage of MLUw, which I have previously discussed. However, in children language acquisition, there are yet to be alternative instruments developed to evaluate the language development in general. MLU in this study is used with the combination of qualitative analyses and generally serves to approximate the child’s linguistic development at certain points in time.
4.5.3 Lexical development. MLU, as discussed before, measure the child’s