Titone and Tiv (Reference Titone and Tiv2022) present a well-crafted, engaging, and persuasive case for a Systems Framework of Bilingualism to guide our understanding of the complex sources of sociolinguistic context that influence people's language use, development, and cognition.
It is only natural that from their beginnings both neuroscience and cognitive psychology prioritized looking inwards for their answers. In so doing, they ignored at their peril the advice of ecological psychologists to ‘ask not what's inside your brain – ask what your brain is inside of’. After pursuing some early research showing developmental dyslexia involves far less cut-and-dried dissociations than the acquired disorders surface vs. deep dyslexia, like Titone and Tiv, my epiphany came from reading Rethinking Innateness (Elman, Bates, Johnson, Karmiloff-Smith, Parisi & Plunkett, Reference Elman, Bates, Johnson, Karmiloff-Smith, Parisi and Plunkett1996) – modules are made not born.
I've got to admit, it's getting better. As I explained in Essentials of a Theory of Language Cognition (Ellis, Reference Ellis2019), Post-Cartesian Cognitive Science sees cognition as not just “in the head”: it extends well beyond the skull and the skin – it is embodied, environmentally embedded, enacted, encultured (4E Cognition). Cognition is socially distributed (Distributed Cognition and the Extended Mind hypothesis). Hutchins’ Cognitive Ecology showed us how the traditional focus upon individual cognition (bounded in social space, in physical space, in time, in brain, and in mind) resulted in an attribution problem: the commitment to the notion that all intelligence is inside an inside/outside boundary forced us to cram everything inside that is required to produce the observed behaviors.
The solution is to go wild, studying language cognition, usage, and emergence in their normal social habitats as “the webs of mutual dependence among the elements of a cognitive ecosystem” (Hutchins, Reference Hutchins2010). As Titone and Tiv acknowledge, the Douglas Fir Group (Atkinson, Byrnes, Doran, Duff, Ellis, Hall, Johnson, Lantolf, Larsen–Freeman, Negueruela, Norton, Ortega, Schumann, Swain & Tarone, Reference Atkinson, Byrnes, Doran, Duff, Ellis, Hall, Johnson, Lantolf, Larsen–Freeman, Negueruela, Norton, Ortega, Schumann, Swain and Tarone2016) encouraged such unbounded perspectives on SLA, as did earlier steps in the socio-cognitive dance following the social turn in SLA (Atkinson, Reference Atkinson2011; Ellis, Reference Ellis, Cadierno and Eskildsen2015; Hulstijn, Young, Ortega, Bigelow, DeKeyser, Ellis, Lantolf, Mackey & Talmy, Reference Hulstijn, Young, Ortega, Bigelow, DeKeyser, Ellis, Lantolf, Mackey and Talmy2014). I appreciated the aesthetics whereby Titone and Tiv (Reference Titone and Tiv2022, Figure 1) emphasize the importance of time by adding moon phases to Douglas Fir Group (2016, Figure 1). I admire the range of sociolinguistic influences they consider and the range of data they present supporting each layer. In understanding language, it is indeed essential to consider the cognitive and the social, as Tomasello has championed: socially-extended cognition, where our mental states are partly constituted by the states of other thinkers, has origins in our enculturation (Tomasello, Reference Tomasello1999) and in our uniquely human skills of intentionality: joint intentions, joint attention, collaboration, imitation, prosocial motives, and social norms (Tomasello, Reference Tomasello2008). Language is the quintessence of distributed cognition.
Recognizing the multiple agents and their dynamics in a systems framework calls for whole new ways of thinking, new methods, and modelling techniques. The seeds sown in Rethinking Innateness now flourish in Emergentist Approaches (Ellis, Reference Ellis1998; MacWhinney, Reference MacWhinney1999; MacWhinney & O'Grady, Reference MacWhinney and O'Grady2015). The Five Graces Group (Beckner, Ellis, Blythe, Holland, Bybee, Ke, Christiansen, Larsen-Freeman, Croft, Schoenemann & Five Graces Group, Reference Beckner, Blythe, Bybee, Christiansen, Croft, Ellis, Holland, Ke, Larsen-Freeman and Schoenemann2009) outlined the approach as Language as a Complex Adaptive System. Complex systems research (e.g., Newman, Reference Newman2011) emphasizes such factors as scale-free properties, power-law distributions, robustness, small-world phenomena, agent-based modelling, and networks. Network Science (Barabási & Pósfai, Reference Barabási and Pósfai2016) has revolutionized our understanding of social systems, of brain connectivity (Sporns, Reference Sporns2011) and the human connectomeFootnote 1, and I believe it has important consequences for the robustness of language acquisition, showing how Plato's problem might be helped to be solved by the latent structures of language usage wherein Zipfian power-law frequency distributions of constructions present to the learner a lion's-share of basic level exemplars that are close to the prototypical meaning and central and well-connected in the semantic network (Ellis, O'Donnell & Römer, Reference Ellis, O'Donnell, Römer, MacWhinney and O'Grady2015; Ellis, Römer & O'Donnell, Reference Ellis, Römer and O'Donnell2016). For these reasons, I encourage not just systems thinking, but complex adaptive systems thinking.
A good analogy can often help in our thinking (Hofstadter & Sander, Reference Hofstadter and Sander2013). When considering bilingualism, language and cognition, one of my favorite analogical domains is transportation, cars and society. When buying a used car, it's always useful to look under the hood / bonnet (AmE / BrE), to consider how old it is, how many miles are on the clock, number of prior owners, history of accidents or trauma, whether it has been regularly serviced, garaged, cherished etc. A dead battery can ruin your day, as can a traffic jam at rush hour, but in different ways. With regard to the relationship that Titone and Tiv demonstrate between their measures of language entropy and aspects of language control and/or resting state functional connectivity in bilinguals, I am reminded of EPA fuel economy statistics which give, for each make of car, separate miles-per-gallon estimates for city and highway driving. High entropy, stop-and-go urban driving uses more gas, energy, and oxygenFootnote 2; it requires more control (consider how tired and attentionally / emotionally drained you would be after an hour of driving in a new city like Paris or London vs. an hour on the interstate); and it involves many more connections (consider the number or detail of directions a maps app requires for a ten-mile city vs interstate trip). The number and nature of edges / links / connections and their strength / carrying-capacity / bandwidth is key in social, neural, transport (…, language, and semantic) networks, as are the connectedness and centrality of the nodes they connect.
Of course, analogical thinking is only a start in generating hypotheses. The real work is in the language science that follows – in vivo, in vitro, in silico; in the lab, in the scanner, in corpora and computer models, and in the wild. I'm excited to see where these complex adaptive systems approaches will take our research.
Titone and Tiv (Reference Titone and Tiv2022) present a well-crafted, engaging, and persuasive case for a Systems Framework of Bilingualism to guide our understanding of the complex sources of sociolinguistic context that influence people's language use, development, and cognition.
It is only natural that from their beginnings both neuroscience and cognitive psychology prioritized looking inwards for their answers. In so doing, they ignored at their peril the advice of ecological psychologists to ‘ask not what's inside your brain – ask what your brain is inside of’. After pursuing some early research showing developmental dyslexia involves far less cut-and-dried dissociations than the acquired disorders surface vs. deep dyslexia, like Titone and Tiv, my epiphany came from reading Rethinking Innateness (Elman, Bates, Johnson, Karmiloff-Smith, Parisi & Plunkett, Reference Elman, Bates, Johnson, Karmiloff-Smith, Parisi and Plunkett1996) – modules are made not born.
I've got to admit, it's getting better. As I explained in Essentials of a Theory of Language Cognition (Ellis, Reference Ellis2019), Post-Cartesian Cognitive Science sees cognition as not just “in the head”: it extends well beyond the skull and the skin – it is embodied, environmentally embedded, enacted, encultured (4E Cognition). Cognition is socially distributed (Distributed Cognition and the Extended Mind hypothesis). Hutchins’ Cognitive Ecology showed us how the traditional focus upon individual cognition (bounded in social space, in physical space, in time, in brain, and in mind) resulted in an attribution problem: the commitment to the notion that all intelligence is inside an inside/outside boundary forced us to cram everything inside that is required to produce the observed behaviors.
The solution is to go wild, studying language cognition, usage, and emergence in their normal social habitats as “the webs of mutual dependence among the elements of a cognitive ecosystem” (Hutchins, Reference Hutchins2010). As Titone and Tiv acknowledge, the Douglas Fir Group (Atkinson, Byrnes, Doran, Duff, Ellis, Hall, Johnson, Lantolf, Larsen–Freeman, Negueruela, Norton, Ortega, Schumann, Swain & Tarone, Reference Atkinson, Byrnes, Doran, Duff, Ellis, Hall, Johnson, Lantolf, Larsen–Freeman, Negueruela, Norton, Ortega, Schumann, Swain and Tarone2016) encouraged such unbounded perspectives on SLA, as did earlier steps in the socio-cognitive dance following the social turn in SLA (Atkinson, Reference Atkinson2011; Ellis, Reference Ellis, Cadierno and Eskildsen2015; Hulstijn, Young, Ortega, Bigelow, DeKeyser, Ellis, Lantolf, Mackey & Talmy, Reference Hulstijn, Young, Ortega, Bigelow, DeKeyser, Ellis, Lantolf, Mackey and Talmy2014). I appreciated the aesthetics whereby Titone and Tiv (Reference Titone and Tiv2022, Figure 1) emphasize the importance of time by adding moon phases to Douglas Fir Group (2016, Figure 1). I admire the range of sociolinguistic influences they consider and the range of data they present supporting each layer. In understanding language, it is indeed essential to consider the cognitive and the social, as Tomasello has championed: socially-extended cognition, where our mental states are partly constituted by the states of other thinkers, has origins in our enculturation (Tomasello, Reference Tomasello1999) and in our uniquely human skills of intentionality: joint intentions, joint attention, collaboration, imitation, prosocial motives, and social norms (Tomasello, Reference Tomasello2008). Language is the quintessence of distributed cognition.
Recognizing the multiple agents and their dynamics in a systems framework calls for whole new ways of thinking, new methods, and modelling techniques. The seeds sown in Rethinking Innateness now flourish in Emergentist Approaches (Ellis, Reference Ellis1998; MacWhinney, Reference MacWhinney1999; MacWhinney & O'Grady, Reference MacWhinney and O'Grady2015). The Five Graces Group (Beckner, Ellis, Blythe, Holland, Bybee, Ke, Christiansen, Larsen-Freeman, Croft, Schoenemann & Five Graces Group, Reference Beckner, Blythe, Bybee, Christiansen, Croft, Ellis, Holland, Ke, Larsen-Freeman and Schoenemann2009) outlined the approach as Language as a Complex Adaptive System. Complex systems research (e.g., Newman, Reference Newman2011) emphasizes such factors as scale-free properties, power-law distributions, robustness, small-world phenomena, agent-based modelling, and networks. Network Science (Barabási & Pósfai, Reference Barabási and Pósfai2016) has revolutionized our understanding of social systems, of brain connectivity (Sporns, Reference Sporns2011) and the human connectomeFootnote 1, and I believe it has important consequences for the robustness of language acquisition, showing how Plato's problem might be helped to be solved by the latent structures of language usage wherein Zipfian power-law frequency distributions of constructions present to the learner a lion's-share of basic level exemplars that are close to the prototypical meaning and central and well-connected in the semantic network (Ellis, O'Donnell & Römer, Reference Ellis, O'Donnell, Römer, MacWhinney and O'Grady2015; Ellis, Römer & O'Donnell, Reference Ellis, Römer and O'Donnell2016). For these reasons, I encourage not just systems thinking, but complex adaptive systems thinking.
A good analogy can often help in our thinking (Hofstadter & Sander, Reference Hofstadter and Sander2013). When considering bilingualism, language and cognition, one of my favorite analogical domains is transportation, cars and society. When buying a used car, it's always useful to look under the hood / bonnet (AmE / BrE), to consider how old it is, how many miles are on the clock, number of prior owners, history of accidents or trauma, whether it has been regularly serviced, garaged, cherished etc. A dead battery can ruin your day, as can a traffic jam at rush hour, but in different ways. With regard to the relationship that Titone and Tiv demonstrate between their measures of language entropy and aspects of language control and/or resting state functional connectivity in bilinguals, I am reminded of EPA fuel economy statistics which give, for each make of car, separate miles-per-gallon estimates for city and highway driving. High entropy, stop-and-go urban driving uses more gas, energy, and oxygenFootnote 2; it requires more control (consider how tired and attentionally / emotionally drained you would be after an hour of driving in a new city like Paris or London vs. an hour on the interstate); and it involves many more connections (consider the number or detail of directions a maps app requires for a ten-mile city vs interstate trip). The number and nature of edges / links / connections and their strength / carrying-capacity / bandwidth is key in social, neural, transport (…, language, and semantic) networks, as are the connectedness and centrality of the nodes they connect.
Of course, analogical thinking is only a start in generating hypotheses. The real work is in the language science that follows – in vivo, in vitro, in silico; in the lab, in the scanner, in corpora and computer models, and in the wild. I'm excited to see where these complex adaptive systems approaches will take our research.