This short research note is part of the Durkheim and the Internet project.
‘Home language’ is a variable often used in policy-oriented research on language-in-education. It is assumed that differences in ‘home language’ are causally related to differences in learning outcomes in diverse populations. In Belgium, for instance, systematically reoccurring PISA-results indicating lower scores for ‘migrant’ learners are easily attributed to one ‘home language’ factor: the assumption that Dutch is not the ‘home language’ in many immigrant learners’ families. This point is correlated with, and in a self-confirming loop supported by, two other variables: the ‘level of education’ and ‘occupation’ of the parents of the learner.
Aaron Cicourel (1964) told us half a century ago that variables used in statistical research need to be ecologically validated – they need to be grounded in ethnographically observable facts, where ‘ethnography’ refers to a methodology in which the ‘insiders’ perspective’ is being described. Such facts, Gregory Bateson underscored (1972: xxviii) cannot be denied, and they are always evidence of something. This something can be a pattern as well as an idiosyncrasy, and what it is precisely cannot be determined by assumption; it must be investigated empirically. The trouble with variables such as ‘home language’ in the kinds of research I pointed to, is that they are established as unchecked assumptions and turned into powerful explanatory factors, while, in actual fact, they remain poorly argued and fragile assumptions.
Let me point out some crucial weaknesses in this mode of practice.
- Behind ‘home language’, a particular, and elaborate, sociological imagination is hidden, and this imagination is carried along in the usage of the term as variable and explanatory factor. So the general question to be raised about ‘home language’ is: what exactly is meant by this? Which realities is it supposed to stand for? And once we have found an answer to this, how can these realities be used as an explanation for other realities (i.c. educational performance scores by ‘migrant’ learners).
- In current practice, we see the following sociological imagination emerge
- ‘Home language’ refers to the language(s) spoken among the members of the family in direct interaction;
- More precisely, it refers to parent-child interaction; very often, the mother is implicitly seen as crucial in this respect;
- This ‘home language’, thus established, has a transmission effect: children learn and adopt the language(s) of their parents;
- This transmission effect is important, even crucial: the language(s) transmitted in direct interactions within the family act(s) as a resource as well as a constraint for learning. Home is the crucial socialization locus.
- From an ethnographic point of view, all of these points are weak hypotheses. Here are some critical remarks.
- As to 2.1: what is meant by ‘language’? Is it just the spoken language? If so, where is literacy? And why would the spoken variety of a language prevail over its literate registers when we are trying to determine the effects of ‘home language’ on learning outcomes, knowing the important role of schooled literacy in formal learning trajectories? I shall add more complications to this issue below.
- About 2.1 and 2.2. Is parent-child interaction all there is to ‘home language’? Children usually grow up in a ‘home’ environment where popular culture, social media and peer groups are very much part of what ‘home’ is all about. Thus, even if parent-child interaction would be ‘monolingual’ (in reality it never is, see below), the actual ‘home language’ environment experienced by children could be outspokenly ‘multilingual’, with complex modes of spoken and written interaction deployed in a variety of relationships – with parents and family members, non-family friends and peer group members both online and offline, and ‘distant’ popular culture networks, to name just these. Children might spend far more time interacting with, say, members of their after-school soccer team than with their parents.
- About 2.2. Even if we accept parent-child interactions as being of paramount importance in defining the ‘home language’ environment, which types of interactions are we talking about? There are homes where parent-child interactions predominantly revolve around order and discipline (the ‘eat-your-veggies-and-clean-up-your-room’ type, say) and homes where more intimate and elaborate genres are practiced (the ‘mom-is-your-best-friend’ type, say). If we consider parent-child interaction a crucial form of input in language socialization, we need to be precise about what such modes of interaction actually involve, for children will learn very different bits of language depending on the types of interaction effectively practiced.
- About 2.2. The previous remark leads us to a more fundamental one (complicating my point (4) above): ‘language’ is a very poor unit of analysis for determining what different modes of interaction actually do in the ‘home language’ environment. Register is far more relevant as a unit: we organize different modes of interaction by means of very different linguistic and communicative resources. Concretely, when a child grows up in the ‘eat-your-veggies-and-clean-up-your-room’ culture mentioned above, it is likely to learn the discursive resources for commands and instructions, not those for talking about one’s deeper feelings or dreams. In that sense, ‘monolingual’ is always a very superficial descriptor for any real sociolinguistic regime – it’s never about language, and always about specific bits of language(s) operating in normatively defined (and complex) form-function mappings (called ‘languaging’ in current literature).
- About 2.3. That there is a transmission effect cannot be denied – see the previous point. The thing is, however, what exactly is transmitted? Which particular register features ‘spill over’ from parents onto children in the different modes of interactions mentioned earlier? And which ones are activated, acquired and shaped in the different forms of interaction, within the broader reality of ‘home language’ described above? And how about the specific school-related registers? How do they actually relate to the registers deployed in the ‘home language’?
- About 2.3 and 2.4. What really needs to be established is the actual structure of the repertoire of the children. And how does parent-child interaction (and its transmission effects) fit into such repertoire structures? We might learn, from such inquiries, that children might actually reject the ‘home language’ in its narrow definition and that far more powerful transmission effects emerge from, e.g., peer groups or popular culture (and not just by teenage children). Socialization, we should realize and accept, happens in far broader social-systemic environments, and the home (in the imagination outlined above) cannot a priori be assumed to be the most important one. The specific role of the home within such broader socialization environments needs to be established empirically. In an age of intense online-offline dynamics, the old Durkheim-Parsonian views of ‘primary’ socialization units such as the family need to be critically revisited.
- A general remark. I referred to some other variables commonly correlated with ‘home language’: the level of education and the professional occupation of the parents – usually the mother. An unspoken assumption is that optimal learning effects can be derived from (a) a Dutch-dominant ‘home language’ environment, (b) with highly educated parents (c) employed in prestige-carrying occupations, acting as main transmission agents. But according to the logic of this particular bit of sociological imagination, the most powerful transmission effects may come from parents not fitting this picture. An unemployed parent is likely to be far more available for parent-child interaction than a full-time employed one. As for the latter, such powerful transmission effects cannot be just assumed, and the earlier issue of interaction types and specific registers becomes more pressing. In homes with ‘absent parents’, the effects of the broader socialization environment must be taken seriously. The implicit status hierarchy contained in (a)-(c) above just may be a sociological fiction.
From an ethnographic viewpoint – and, by extension, a viewpoint emphasizing ecological validity in research – the unquestioned use of ‘home language’ in the sense outlined here will inevitably result in fundamentally flawed research, the outcomes of which are entirely dependent on a series of assumptions that do not stand the test of empirical control. The problem is situated at the level of the sociological imagination motivating such assumptions; and this imagination, we know, has lost touch with sociological reality. The good news, however, is that there is a significant amount of ethnographic research addressing these issues, from which one can draw a more realistic set of assumptions and against which the ecological validity of current findings can be checked. The potential benefit of doing that has been, one hopes, sufficiently established here.
Bateson, Gregory (1972 ) Steps to an Ecology of Mind. Chicago: University of Chicago Press
Cicourel, Aaron (1964) Method and Measurement in Sociology. New York: Free Press
An intriguing rebuttal came my way shortly after posting this text: the kinds of research I advocate here would yield way too much diversity and, thus, prevent generalization. It’s an old argument, and those who use it display an amazingly superficial knowledge about generalization as a scientific practice. Reading the two classics I cite here, and especially Bateson’s old Steps to an Ecology of Mind (1972), could be helpful. But more disturbing is the implication of this argument: that sociological fiction is fine because it is generalizable – we know that our assumptions are wrong or unfounded, but we will still use them because they satisfy a formal-methodical criterion of ‘generalizability’. Unfortunately, fiction doesn’t become science when it’s generalized. It becomes generalized fiction.