In this presentation, we describe an interdisciplinary research project in educational science from a philosophical point of view. The project aims to study the mechanisms behind an educational intervention (FonoMix) that is expected to positively affect phonological awareness in children during their first school year (at age 7 in Sweden). In this project, we propose to integrate neurophysiological measures – in the form of event-related potentials (ERP), behavioral measures – in the form of standardized tests, with ethnographic methods – in the form of classroom observations and interviews, in order to develop a theory of the mechanisms that connect FonoMix to its effect on phonological awareness.

Phonological awareness (e.g., Torgesen, Wagner, and Rashotte 1994) is a theoretical term in linguistics and educational research used to describe a set of skills related to the capacity of identifying, distinguishing, and manipulating phonemes (examples of such skills are rhyming and alliteration) and is a robust predictor of reading development (e.g., Gathercole, Willis, and Baddeley 1991).FonoMix is a training program that utilizes a phonological, multi-sensory methodology for teaching reading by connecting visual, auditory, and kinaesthetic senses and by concretizing the relationship between phonemes and graphemes. Evidence suggests that FonoMix has a positive effect on phonological awareness (Fälth, Svensson, and Ström 2020). In our project, we attempt at explaining how, i.e., through which mechanisms, the program generates its effect. Our methodology consists of three parallel lines of investigation. The first consists of a study designed as a randomized controlled trial (RCT) based on a set of standardized tests, measuring the children’s phonological awareness and word decoding skills. The second consists of electrophysiological recordings of participating children to investigate the intervention’s effects on the neural processing of rhyming. Finally, the third line of investigation consists of an ethnographic study of the intervention as educational practice, using participant observation, interviews, and informal conversations with teachers and children. The novelty of our design is that we will integrate these results to develop a mechanistic theory of the intervention. In this presentation, we discuss the potential mechanistic rewards of our integrative project and focus on questions concerning how our integrative methodology can allow the development of a mechanistic theory of FonoMix, and what mechanisms we ultimately need. We argue that answering these questions requires a main preliminary assumption. This assumption rests on a shift from “mechanism of” to “mechanism for”. This means that appraising the relevance and completeness of a mechanistic model requires knowing how the model is going to be used. In our case, the goal of the mechanistic account is to support a recommendation for teachers. This entails the capacity of extrapolating the measured effect of the intervention over a set of classroom contexts with specific characteristics. From this point of view, we are able to discuss the rewards of our integrative approach. First, our project potentially improves on past effect size studies of FonoMix by accounting for how elements of classroom practice and event-related potential effects might modulate and/or mediate the effects measured using behavioral tests. This is an issue that has previously been widely discussed in the philosophy of evidence-based practice and policy (e.g., Grüne-Yanoff 2015; Marchionni and Reijula 2018). Specifying how certain factors modulate and/or mediate an intervention effect, allows extrapolating to further contexts. However,2 (3)we point out that this extrapolation requires knowledge of what class of confounders is specific for the relevant classroom context. This might be solved by some background theory, but in many cases the class of possible confounders is unknown. This clearly demonstrates the reward of using classroom ethnography. By putting the intervention into a concrete context, we might be able to observe which type of confounders emerge when the program is applied in a classroom. Hence, in this case, the mechanistic rewards of our project seem to be afforded by the integration of levels (the inclusion of a classroom context) and the integration of methods (using theory-building—in this case, ethnography, rather than theory-testing methods).The second related reward consists of the specification of the intervention variable. Effect-size studies conceive the intervention as an independent and ontologically coherent event. Classroom ethnography specifies the intervention variable as an array of practices mediated by relations and artifacts. This is important for supporting a recommendation for practitioners since different practitioners might apply the intervention differently. The final issue we discuss is the reward of adding electrophysiological measures. According to a widespread view, neuroscience provides mechanistic explanations(e.g., Zednik 2014). Therefore, there would seem to be a clear mechanistic reward in integrating ERP measures into our theory. However, we might ask what mechanism– that is, what network of entities and activities – an ERP effect reflects. It does not appear to track an entity (or set of entities), as it does not track a particular brain structure, but rather a pattern of electrophysiological activities. Therefore, the reward of including ERP recordings to the study is that of providing a thick description of the relationship between some of the entities involved in the effect of the intervention. In contrast with the first reward, we argue that this requires an integration of theories.

References

Fälth, Linda, Elin Svensson, and Annika Ström. 2020. “Intensive Phonological Training with Articulation—An Intervention Study to Boost Pupils’ Word Decoding in Grade 1.” Journal of Cognitive Education and Psychology, October. https://doi.org/10.1891/JCEP-D-20-00015.

Gathercole, Susan E., Catherine Willis, and Alan D. Baddeley. 1991.“Differentiating Phonological Memory and Awareness of Rhyme: Readingand Vocabulary Development in Children.” British Journal of Psychology82 (3): 387–406. https://doi.org/10.1111/j.2044-8295.1991.tb02407.x.

Grüne-Yanoff, Till. 2015. “WHY BEHAVIOURAL POLICY NEEDS MECHANISTIC EVIDENCE.” Economics and Philosophy, November, 1–21. https://doi.org/10.1017/S0266267115000425.

Marchionni, Caterina, and Samuli Reijula. 2018. “What Is Mechanistic Evidence, and Why Do We Need It for Evidence-Based Policy?” Studies in History and Philosophy of Science Part A, August. https://doi.org/10.1016/j.shpsa.2018.08.003.

McIntyre, Laureen, Susan Protz, and Lynn McQuarrie. 2008. “Exploring the Potential of LiPS Instruction for Beginning Readers.” Developmental Disabilities Bulletin 36: 18–48.

Torgesen, Joseph K., Richard K. Wagner, and Carol A. Rashotte. 1994.“Longitudinal Studies of Phonological Processing and Reading.” Journal of Learning Disabilities 27 (5): 276–86.https://doi.org/10.1177/002221949402700503.

Zednik, Carlos. 2014. “Are Systems Neuroscience Explanations Mechanistic?” In Preprint Volume for Philosophy Science Association 24th Biennial Meeting. Chicago, IL. http://philsci-archive.pitt.edu/10859/.3 (3)