I conclude this chapter by addressing some of the limitations of the study and possibilities for future research. Furthermore, I propose some considerations about the usefulness of such an approach for policy-making purposes.
One of the most important things to point out about the model presented in this chapter is that it is strictly focused on understanding one aspect of language acquisition, that is, the consequences of linguistic distance and lan-guage learning in terms of linguistic disenfranchisement. Obviously, there are many other factors that come into play in the process of language acqui-sition, such as daily exposition to a given language in one’s daily life. An interesting development of the model presented in this chapter could be rep-resented by the inclusion of these other factors, with a view to studying their interaction. For example, one might be interested in seeing whether and to what extent exposition to a certain language through the media can compen-sate for the distance between languages.
For practical reasons, I have only taken a selection of EU official languages into consideration to calculate disenfranchisement indexes. However, the lin-guistic landscape of the EU has significantly changed over the past few years as a consequence of migration from the outside, and it is likely to keep on changing in the coming years. Migration can change the relative proportions of language speakers by either reinforcing one particular language group (say, Spanish speakers from South America) or creating allophone commu-nities, i.e. groups of people speaking a non-EU language (for example, Chi-nese migrants). This picture becomes all the more intricate if we add to the equation all the regional and minority languages spoken throughout the EU.
Besides, many people are able to speak more than one language in addition to their own, at different levels of proficiency. All in all, it can be challeng-ing to sketch the lchalleng-inguistic landscape of the EU, let alone to predict its evo-lution. However, if EU institutions are to take linguistic disenfranchisement into consideration when drafting language policies, keeping track of changes in the linguistic landscape becomes crucial. With this objective, agent-based modelling, as well as computational methods in general, are a good resource to have in the policy-maker’s toolbox. For a more complete view of the EU linguistic landscape, an interesting extension to this model would be the in-clusion of the remaining EU official languages and other non-EU languages.
However, a non-trivial difficulty that would need to be tackled would be the modelling of linguistic distance encompassing also non-Indo-European
languages among all of the language considered. If possible, it would be ad-visable to rely on more sophisticated metrics of language distance, in that, as said, the linguistic distance index has its flaws.
Finally, this study was started when the UK was still part of the Euro-pean Union. As of today, it is still not clear what the future of English within the EU institutions will be. The relative weight of English in terms of native speakers was drastically reduced after Brexit, and this could have repercus-sions not only on the communication practices among EU employees, but also on education policies across various member states. Therefore, it would be interesting to repeat this study in a few years, leaving some time for the linguistic repercussions of Brexit to manifest.24
Before concluding, I also want to point out that I limited myself to mak-ing objective and technical observations about the results generated by the model, avoiding policy recommendations. This choice derives from the fact that policy making is an eminently political process. Starting a discussion on whether a specific regime should be preferred or whether a fair and equal treatment of citizens should be prioritized is well beyond the scope of this chapter and left for future research.
24For an interesting discussion on this matter, as well as on the dominant position of En-glish in the EU, see https://www.publico.pt/2016/07/01/mundo/opiniao/as-linguas-na-europa-o-que-mudara-com-o-brexit-1736870.
Chapter 5
Concluding Remarks: Complexity and Language Matters
5.1 Overview
In this dissertation I tried to explore in depth the many ways in which com-plexity manifests itself. Although my research work focused mainly on lan-guage matters, I referred to numerous examples from the social and natural sciences, in order to provide as comprehensive an overview of complexity as possible and to gently usher the readers into the more specific applica-tions on language-related issues. Obviously, this required not only a review of complexity theory, but also a reinterpretation of language policy from a complexity theory perspective. This is what Chapter1 was devoted to. In Chapters 2 through 4 I presented some applications of agent-based mod-elling, one of the major analytical tools of complexity theory, to three tra-ditional language policy issues, namely, multilingual communication, lan-guage dynamics, and linguistic disenfranchisement. In so doing, my goal was to explore and explain the potential contribution of complexity theory to language policy making. Although they are clearly three separate appli-cations that share the same methodology, they all contribute to showing the usefulness of agent-based modelling in language policy. Chapter2focuses in particular on how micro-level qualitative observations can be extended and generalized through agent-based modelling in order to observe their impact on the macro-level. Building on this, Chapter 3 adds social theories to the picture, showing how theory-based observations can inform agent behaviour even in the absence of direct observation. Finally, Chapter4develops a very practical application for policy making and complements the discussion of the preceding chapters by moving the attention from theory to practice. By highlighting different virtues of agent-based modelling, Chapters2through
4allow me to make a strong point in favour of the adoption of agent-based modelling in policy making.
Admittedly, the approach adopted to develop the cases presented in this dissertation diverges slightly from more orthodox approaches, that would prescribe that various modelling options are considered before selecting the most appropriate. However, as I said in many occasions, the same issues discussed in this dissertation have been already studied by means of other methodologies. My objective was not to find out which methodologies is best for each case. Rather, I aimed at showing that different research meth-ods should not be seen as alternative but as complementary and that agent-based modelling, a methodology still to a certain extent disregarded by soci-olinguists and language policy makers, can complement significantly other methodologies and result in improved research and policy making.
In this final chapter I wrap up the main findings of this dissertation, ac-knowledge critical aspects and potential limitations, and discuss opportuni-ties for future research. I also mention briefly some other promising tech-niques of computational social science that can contribute to language policy research and other research methods that can be integrated with agent-based modelling in order to create ever more realistic and accurate simulations.