Implicitness is a term that crops up ever more frequently in research on language attitudes, language regard and the social meaning of language variation. Sometimes it refers to the psychological constructs studied in these fields, at other times to a type of method used to measure them. The term is used alongside other terminology like ‘covert’, ‘private’ or ‘deep’ attitudes, sometimes equivalently, sometimes in contrast. What exactly is understood by all these terms is almost never clearly defined, either on the conceptual level or in relation to the terms’ methodological and theoretical implications. In cases where the use of the term “implicit” is delineated, researchers differ in their interpretations or build on assumptions for which the evidence is rather limited. In this special issue of Linguistics Vanguard, we bring together a number of studies that question traditional interpretations of implicitness in experimental language attitudes research in order to provide new perspectives on the interpretation and relevance of the concept for the study of language variation and change. Based on these contributions, we argue that sociolinguistic research on the evaluation of language needs to abandon a number of traditional assumptions regarding different types of language attitudes and their relation to various measuring techniques. Instead, we recommend that language attitude researchers approach the field with an open mind, building on both old and new empirical techniques, and also interact more closely with other fields that study attitudes and evaluations, like social psychology (cf. Pantos, Rosseel et al., and Adams in this issue). Our ultimate goal is to promote convergence between distinct brands of attitude research on the delineation and definition of different attitude types, and to lay the groundwork for developing a shared toolbox with which to elicit attitudes in response to specific theoretical questions.
In this paper, we first delve into the issues related to the concept of implicitness in sociolinguistic attitude research from three perspectives: how can/should it be defined, what is its theoretical significance and how can we measure it? After that, we briefly present the contributions to this special issue.
From a conceptual point of view, several definitions and interpretations of implicitness have been put forward. In linguistics the focus tends to be on awareness or the level of consciousness: a distinction is made between attitudes that speakers are aware of and those that lie beneath the level of consciousness (e.g. Labov 1972; Kristiansen 2009; Garrett 2010; Grondelaers and Kristiansen 2013; Preston 2013, 2015). In the field of social cognition, by contrast, the concept of implicitness has been investigated much more intensively and researchers have proposed multidimensional definitions that recognize more facets to the concept of implicitness than just awareness, facets which are not typically considered in linguistic research. Implicitness in this field is usually understood in terms of automaticity, a variable which comprises multiple features (unintentionality, resource-independence, uncontrollability as well as unconsciousness) that need not all be present, but can qualify the way in which the outcome of an attitude measure is implicit (De Houwer et al. 2009; De Houwer and Moors 2010; Gawronski and De Houwer 2014). Such definitions of implicitness seem to allow for a conceptualization in terms of gradience, or a continuum between implicitness and explicitness.
When it comes to the theoretical importance of implicitness, it has been claimed that implicit, covert, private or deep evaluations (to the extent that these terms represent the same psychological construct) offer privileged access to the perceptual correlates of linguistic change (see especially Kristiansen 2009 and Grondelaers and Kristiansen 2013 for the most ambitious version of the claim; but see also Preston 2013 and Kristiansen and Pharao in this issue). Although researchers do not always explicitly acknowledge this assumption and the available evidence is not unequivocal, many studies draw on it (see especially Kristiansen and Grondelaers 2013). Without such evidence, and in the absence of a consensus definition of implicitness, it is hard to evaluate the role of implicit attitudes as a driving force in language variation and change. For some speech communities, as documented for instance by the LANCHART Project in Denmark, unconsciously offered attitudes have been found to correlate with standard language innovation. More concretely, when young Danes are unaware of the fact that they are evaluating varieties of Danish, they attach dynamic prestige to a Modern Copenhagen accent that rivals Standard Danish in certain domains of society. Consciously offered evaluations, by contrast, corroborate a traditional standard language ideology (SLI), associating conservative prestige with Standard Danish (cf. Kristiansen and Pharao in this issue). However, in other communities such as Flanders, the available evidence does not completely confirm this complementary distribution between explicit and implicit attitudes. In a matched guise experiment, Grondelaers and Speelman (2013) found evidence that Flemish listeners attribute dynamic prestige to a widely used colloquial variety, which is sometimes claimed to put pressure on the position of Standard Belgian Dutch. Given that participants were unaware of the linguistic attitude object – and the exclusion of subjects who had guessed the experimental purpose – Grondelaers and Speelman (2013) claimed not only that positive attitudes towards the colloquial variety were implicit, but also that such evaluations could only be extracted on this implicit level. In a series of Implicit Association Tests (IAT), however, Rosseel (2017) found that it was especially the conservative prestige of the official standard rather than the dynamic prestige associated with the colloquial variety that was robustly attested when language attitudes were measured under implicit conditions. More importantly, a number of other studies have also shown that positive dynamism evaluations of the colloquial Belgian variety can be documented on the explicit level. In a free response experiment which extracted keywords in reaction to language variety labels from subjects who were perfectly aware of that task, Grondelaers et al. (under review) found obvious dynamism perceptions for the publicly stigmatized variety (see also Rosseel 2017 for explicit dynamism evaluations of colloquial Belgian Dutch). This raises pertinent questions like: do we always need implicit attitudes if they are typically more difficult to extract? And what exactly is the theoretical significance of implicit attitudes and the level of awareness on which they are elicited for the study of language variation and change? A concern which raises its inconvenient head in this respect is the fact that participant awareness almost categorically splits the field: while Kristiansen and Pharao (this issue) and Grondelaers and Van Gent (this issue) do not extract their dynamism evaluations under conditions of automaticity, they design their matched guise experiments to guarantee that participants are completely unaware of the linguistic attitude object; conversely, sociolinguists working in the social cognition tradition – like Rosseel et al. (this issue) – extract more automatic and indisputably deeper attitudes from participants who are nevertheless fully aware of the fact they are participating in research on stigmatized language variation. A question which inevitably arises in this respect is whether the different attitudes obtained in Grondelaers and Speelman (2013) and Rosseel (2017) are ontologically different, or whether they are an artefact of different views on implicitness. This is a crucial problem that has to be solved if we want the sociolinguistic community to reconsider the status of subjective factors (evaluation, attitudes, ideology,…) as a driving force behind language change: Labov’s (2001) decreasing interest in such factors (see Kristiansen 2010 for an overview) is plausibly related to the methodological disagreement in the attitude measuring field.
This brings us to our third point: challenging the linguistic conception of implicitness has important methodological consequences. If we ask ourselves the question of what exactly we mean by implicitness, and if we agree that it is a multifaceted concept, we should also ask ourselves which aspect of implicitness we are measuring with specific methods and tools. This goes for traditional sociolinguistic methods like matched guise experiments, but the question is especially relevant in the context of the recent upsurge in social psychological measures to study implicit associations. Almost 60 years after adopting the matched guise technique from social psychologists (Lambert et al. 1960), linguists are once again finding inspiration in this branch of psychology and are gradually starting to use methods like the Implicit Association Test (e.g. Babel 2010; Redinger 2010; Campbell-Kibler 2012; Pantos and Perkins 2012; Lee 2015; Loudermilk 2015; Llamas et al. 2016; Álvarez-mosquera 2017; Rosseel et al. 2018), affective priming (Speelman et al. 2013) or the Relational Responding Task (Rosseel et al. in this issue). Yet, these researchers do not always question what it is exactly that these tools measure, and how the measurements compare to the ones obtained from more traditional tools. To briefly take up the latter question, if we adopt a view of implicitness as a gradient construct, as suggested above, perhaps we should also consider the possibility that some methods capture attitudes that are situated somewhere in the middle between implicit and explicit. An example of such a method is the free response task, mentioned above, introduced in Grondelaers and Speelman (2015), and elaborated in Grondelaers et al. (under review), in which participants are asked to jot down the first three words they associate with specific languages varieties. The time pressure under which the elaborated guise of this task is performed makes for more automatic circumstances than when participants fill out a questionnaire at their own pace. What is more, in an ongoing follow-up study, the stimulus to be evaluated is no longer the variety label, but a speech sample representing each variety. As a result, the participants’ awareness of the task as a language-related evaluation tool is likely lower, taking the associations harvested with the technique even closer to the implicit level. Of all the tools showcased in this special issue, Adams’ measurements of implicit attitudes towards six accent varieties of English most closely approximate the implicit pole of the explicit-implicit continuum, as responses in her task were extracted under time pressure from participants who were unaware of the fact they were evaluating the accents (see below).
All the studies in this special issue focus on one or more of the three perspectives on implicitness just introduced. Andrew Pantos opens the issue with a helpful review of the social psychological source literature on the correlations between implicitness, automaticity and (sub)consciousness. Its starting-point is the recognition that most language attitude researchers are sociolinguists who borrow methods and terms from social psychology, but use them in a different way. The ensuing confusion and unintended implications can be avoided, Pantos claims, by using terms such as “implicit”, “automatic”, and “conscious” consistently and with specificity. This does not mean that social psychology itself is unanimous in its terminology. It is noted that the social cognition literature is inconsistent in its application of the term “implicit” to the measurement procedure, the attitude measured, or both. Different social psychology strands, moreover, highlight alternative aspects of implicitness. The attention research tradition defines implicitness in terms of learning and control: well-learned attitudes are inescapable constructs which can be automatically retrieved from memory, whereas poorly learned attitudes require more cognitive effort to activate, as a result of which they are not automatic. The implicit memory tradition defines implicitness in terms of consciousness: explicit attitudes are those that an individual is conscious of, and can control, while implicit attitudes are constructs an individual is not conscious of and which are outside the purview of individual control. This definition of implicitness engenders divergence between social psychologists and sociolinguists interested in language attitudes. To the latter, attitudes obtained with the matched guise technique are implicit when experimental subjects are ignorant of the attitude object. Social psychologists, by contrast, regard the outcome of a matched guise experiment as explicit since it requires participants to introspect on their evaluations (as a result of which they are neither automatic nor unconscious).
The methodological delineation of implicitness as a concept is of central importance to Tore Kristiansen’s experimental attitude investigations in Denmark. Against the backdrop of the rich body of production and perception data which provided the input to their inspirational work on Danish standard language dynamics, Tore Kristiansen and Nicolai Pharao discuss the issue of implicitness in relation to the pivotal question how implicit perceptual correlates of language change should be extracted. In their Danish research, Kristiansen and Pharao had found evidence for a modern prestige (dynamism) boost for the increasingly vital københavnsk accent, an evaluation they regarded as an implicit type of attitude. Their main point is that the use of indirect methods such as the matched-guise technique can only tap into such implicit change correlates when experiments are designed and conducted with a view to keeping participants unaware of the linguistic attitude object. On this point, Kristiansen and Pharao’s approach diverges from the bulk of the social psychology tradition which defines implicitness in terms of control and automaticity but does not typically keep participants unaware of the attitudes-to-language object (see also Rosseel et al., this issue). This need not be problematic: in investigations with research interests other than language change – such as the arousal of social stereotypes – the insistence on participant ignorance may be of less relevance.
Stefan Grondelaers and Paul van Gent investigate traditional and modern prestige (dynamism) evaluations of the Moroccan accent of Netherlandic Dutch. While the authors stick to the matched-guise technique, they adapt the method in an attempt to get a better grip on standard language dynamics in Dutch: they include a three-level accent strength variable in the stimuli, and measure dynamism-evaluations on independently validated scales. In order to gauge the nature of the evaluations investigated (more or less explicit), they also elicit attitudes with a more direct tool, viz. the perceived occupation experimental speakers were applying for. Findings provide crucial nuance to existing accounts in two ways. While Moroccan Dutchmen are downgraded on traditional “highbrow” prestige, they are deemed much more dynamic than any other group included in the experiment. Following Kristiansen (2009) and Kristiansen and Pharao (this volume), these modern prestige evaluations are theorized to be the driving force of the increasing tolerance for, and vitality of, the Moroccan accent. But in view of the finding that these modern evaluations could also be extracted with the occupational variable, Grondelaers and Van Gent challenge Kristiansen and Pharao’s suggestion that change-inducing dynamism perceptions are implicit evaluations, at least in Dutch.
A second contribution focusing on Dutch, though in this case Belgian Dutch, is the paper by Laura Rosseel, Dirk Speelman and Dirk Geeraerts. In their study, they introduce the Relational Responding Task (RRT) to linguistic attitude research. The RRT is a reaction time based categorization task which was recently developed in social psychology to measure beliefs under implicit conditions (De Houwer et al. 2015). Rosseel and colleagues use the RRT to measure beliefs associated with two varieties of Belgian Dutch: Standard Belgian Dutch and Colloquial Belgian Dutch. Their results show that the standard variety is strongly evaluated in terms of prestige. The colloquial variety, by contrast, is not considered prestigious, but is associated with dynamic evaluations, just like the Moroccan accent in the study by Grondelaers and Van Gent. The authors show that the RRT can be a useful new tool for sociolinguists to measure language attitudes. Yet, they also point out a number of caveats and suggest linguists keep carefully exploring the measure’s potential for sociolinguistics.
Dennis Preston’s is the only paper in this volume which is not concerned with the experimental elicitation of (implicit) attitudes. It relies on discourse analysis to investigate language production features (presupposition and implicatures) which reveal the speaker’s implicit beliefs about language varieties. A key concern of such an analysis is to distinguish between consciously produced assertions on the one hand, and the implicated and presupposed elements which involve nonconscious activity on the other (it is in this respect that Silverstein 1979 suggests that “the best place to look for language ideology may lie in the terms and presuppositions of metapragmatic discourse, not just in its assertions”). Preston then goes on to analyze a conversation about the perception of Oklahoman speech, which makes sense only by virtue of a number of beliefs which are clearly shared by the interlocutors although they are implicated and presupposed rather than explicitly expressed (these beliefs include the stereotype that Southern speech is derogated as “hick” speech, whose perceived unintelligence can be compensated, however, by evidence of high professional skill). It takes careful deployment of pragmatic and discourse analytic tools, Preston argues, to isolate and identify this belief processing, which is not under the conscious control of speakers.
Katherine Hilton and Sunwoo Jeong investigate the competition between linguistic and non-linguistic impression triggers in discourse and, more particularly, the extent to which negative perceptions prompted by syntactic variation (nonagreement in existentials), prosodic variation (impoliteness-inducing assertive rising declaratives), and interactional variation (interruptive overlapping speech) can be overruled by positive impression cues in the interactional context. In a series of 10 perception experiments, identical variants of the three stimulus types were evaluated on scaled measures in increasingly richer contexts, by large samples of participants who were also invited to comment on the linguistic variables. Crucially, syntax- and prosody-induced impressions recorded in informationally poor contexts were eclipsed by other impressions conjured up in the richer contexts, a finding which the authors attribute to the low social salience of the linguistic variables concerned (as revealed in the participant comments). The fact that the negative associations of overlapping speech could not be salvaged by any favorable impression in a richer context, was attributed to the strongly enregistered indexical link between interactional overlap and interruption (which occasioned ten times more metalinguistic comments than the syntactic and the prosodic variable). Although the authors do not commit themselves to any classification of their linguistically triggered attitudes in terms of explicitness or implicitness, it is attractive to link their strongly and weakly enregistered attitudes to the distinction they make in their introduction between strongly and weakly held attitudes. The former, they claim, are well learnt stored in memory, and impervious to contextual influence, whereas the latter are more likely to be constructed in the moment.
Hilton and Jeong end their paper with an inevitable limitation of all experimental attitude research:
(…) people participating in experiments have very little incentive to develop detailed impressions of speakers or incorporate all available information. Outside of experimental settings, the impressions we form about each other have important consequences, determining who we trust, admire, or dislike.
Against the backdrop of this caveat, a crucial feature of Zoe Adams’ contribution is a higher ecological validity on account of its focus on the impact of accent variation on the persuasive power of dental health recommendations. In view of the fact that (regional) accent is a socially sensitive attitude object in Britain, Adams measures both explicit attitudes – which may reflect social expectations – and automatic evaluations, which are not consciously mediated. Automatic evaluations were recorded as time-conditioned true/false responses to trivia statements produced in the six investigated accents; these responses were subsequently compared to the same participants’ prior responses to written (and hence unaccented) versions of the same statements, and the difference between the audio and written responses was regarded as evidence for the persuasive or dissuasive valence of the accents. Explicit attitudes were extracted in two matched-guise experiments in which participants scored the credibility of speech clips produced in the six investigated accents. In their implicit attitudes, participants found Estuary English, the practical standard of British English, more persuasive than the other varieties; explicit attitudes confirmed RP as the official high-status standard, and displayed outspoken in-group solidarity towards the participants’ own accent.
Although the papers in this issue potentially raise more questions than they provide answers, it is these questions that provide a new impetus to explore the nature of language attitudes. Taken together, the papers collected in this issue point to new directions in which we can take the study of the social meaning of language variation as a way of ensuring that language attitudes are (again) taken seriously as determinants of language variation and change.
De Houwer, Jan & Agnes Moors. 2010. Implicit measures: Similarities and differences. In Bertram Gawronski & B. Keith Payne (eds.), Handbook of implicit social cognition: Measurement, theory and applications, 176–196. New York: Guilford Press. Google Scholar
De Houwer, Jan, Sarah Teige-Mocigemba, Adriaan Spruyt & Agnes Moors. 2009. Implicit measures: A normative analysis and review. Psychological Bulletin 135(3). 347–368. CrossrefWeb of ScienceGoogle Scholar
De Houwer, Jan, Niclas Heider, Adriaan Spruyt, Arne Roets & Sean Hughes. 2015.The relational responding task: toward a new implicit measure of beliefs. Frontiers in Psychology 6. (article 319).Web of ScienceGoogle Scholar
Garrett, Peter. 2010. Attitudes to language. Cambridge: Cambridge University Press. Google Scholar
Gawronski, Bertram & Jan De Houwer. 2014. Implicit measures in social and personality psychology. In Harry T. Reis & Charles M. Judd (eds.), Handbook of research methods in social and personality psychology (2nd ed.), 283–310. New York: Cambridge University Press. Google Scholar
Grondelaers, Stefan & Tore Kristiansen. 2013. On the need to access deep evaluations when searching for the motor of standard language change. In Tore Kristiansen & Stefan Grondelaers (eds.), Language (de)standardisations in Late Modern Europe: Experimental studies, 9–52. Oslo: Novus Press. Google Scholar
Grondelaers, Stefan & Dirk Speelman. 2013. Can speaker evaluation return private attitudes towards stigmatised varieties? Evidence from emergent standardisation in Belgian Dutch. In Tore Kristiansen & Stefan Grondelaers (eds.), Language (de)standardisations in Late Modern Europe: Experimental studies, 171–191. Oslo: Novus Press. Google Scholar
Grondelaers, Stefan & Dirk Speelman. 2015. A quantitative analysis of qualitative free response data. Paradox or new paradigm? In Jocelyne Daems, Eline Zenner, Kris Heylen, Dirk Speelman, & Hubert Cuyckens (eds.), Change of paradigms – New paradoxes: Recontextualizing language and linguistics, 361–384. Berlin/Boston: De Gruyter Mouton. Google Scholar
Grondelaers, Stefan, Dirk Speelman, Chloé Lybaert & Paul Van Gent. under review. Getting a (big) data-based grip on ideological change. Evidence from Belgian Dutch. Submitted to Journal of Linguistic Geography. Google Scholar
Kristiansen, Tore. 2010. Attitudes, ideology and awareness. In Ruth Wodak, Barbara Johnston & Paul Kerswill (eds.), The Sage handbook of sociolinguistics, 265–278. Los Angeles/London/New Delhi/Singapore/Washington DC: Sage. Google Scholar
Kristiansen, Tore & Stefan Grondelaers (eds.). 2013. Language (de)standardisation in Late Modern Europe: Experimental studies. Oslo: Novus Press. Google Scholar
Labov, William. 1972. Sociolinguistic patterns. Philadelphia: Pennsylvania University Press. Google Scholar
Labov, William. 2001. Principles of linguistic change. Social factors ( Vol. 2). Oxford: Blackwell. Google Scholar
Lee, Rachelle. 2015. Implicit associations with Welsh in two educational contexts. York Papers in Linguistics 2(14). 81–105. Google Scholar
Llamas, Carmen, Dominic Watt & Andrew E. MacFarlane. 2016. Estimating the relative sociolinguistic salience of segmental variables in a dialect boundary zone. Frontiers in Psychology 7. Article 1163. Web of ScienceGoogle Scholar
Loudermilk, Brandon C. 2015. Implicit attitudes and the perception of sociolinguistic variation. In Alexei Prikhodkine & Dennis Preston (eds.), Responses to language varieties. Variability, processes and outcomes, 137–156. Amsterdam/Philadelphia: John Benjamins. Google Scholar
Pantos, Andrew J. & Andrew W. Perkins. 2012. Measuring implicit and explicit attitudes toward foreign accented speech. Journal of Language and Social Psychology 32(1). 3–20. Web of ScienceGoogle Scholar
Preston, Dennis. 2015. Does language regard vary? In Alexei Prikhodkine & Dennis Preston (eds.), Responses to language varieties. Variability, processes and outcomes, 3–36. Amsterdam/ Philadelphia: John Benjamins. Google Scholar
Redinger, Daniel. 2010. Language attitudes and code-switching behaviour in a multilingual educational context?: The case of Luxembourg. York: The University of York PhD dissertation. Google Scholar
Rosseel, Laura. 2017. New approaches to measuring the social meaning of language variation: Exploring the Personalized Implicit Association Test and the Relational Responding Task. Leuven: KU Leuven PhD dissertation. Google Scholar
Rosseel, Laura, Dirk Speelman & Dirk Geeraerts. 2018. Measuring language attitudes using the Personalized Implicit Association Test: A case study on regional varieties of Dutch in Belgium. Journal of Linguistic Geography 6(1). 20–39. CrossrefGoogle Scholar
Silverstein, Michael. 1979. Language structure and linguistic ideology. In P. Clyne, W. Hanks & C. Hofbauer (eds.), The Elements: A Parasession on Linguistic Units and Levels, pp. 193–247. Chicago: Chicago Linguistic Society.Google Scholar