This introduction sketches current strands of UB research and outlines the structure of this edited volume, which pays tribute to Heike Behrens, who has contributed to and shaped usage-based (UB) research efforts from the start. The chapter describes how various UB approaches have emerged and evolved over the last three decades, reshaping how we view, research, and model language and interaction, first and second language development, and multilingualism. It also describes how UB approaches have embraced differing perspectives, disciplines, and methodologies that share a focus on authentic language use and experience, shedding new light on processes of language usage and development.

For well over a half-century, child language researchers have struggled with Chomsky’s characterization of “the poverty of the stimulus” — the assertion that the child does not receive enough linguistic data to arrive at knowledge of grammatical structure. Whatever one’s conception of linguistic competence, it is clear that the learner has to make use of both innate capacities and linguistic experience to master one or more native languages. This chapter focuses on experience — the speech encountered by the child learner. The past decades of child language research have revealed important ways in which “the stimulus” is really not so “impoverished”. Here I point to several examples of “the wealth of the stimulus” in my own research and that of colleagues. The learner gleans information about language from attending to ways in which linguistic forms change in the course of verbal interaction with others. Interlocutors expand and question children’s utterances; some of those modifications are taken up by learners. Interlocutors press for clarification, demand uptake. They rephrase their speech in successive utterances, offering a wealth of grammatical and lexical variety around a continuing topic. The interlocutors are not only caregivers, but also peers and older children who present different vocabularies and communicative demands. Children are actively engaged in linguistic worlds with many types of speech acts. Eventual explanations of language acquisition will have to attend to realistic descriptions of the rich data available to learners.

This article proposes a micro-sequential and multimodal approach to scaffolding in parent-child interactions. We argue for a more detailed understanding of how parents manage to transform their scaffolding into a time-limited intervention, transferring responsibility for the task to the child as soon as possible. In a social-interactional perspective following usage-based accounts of language acquisition (Tomasello, 2003), adults play a supportive role, helping children to develop linguistic and discourse skills through structured and dialogical interactions (cf. the Language Acquisition Support System LASS, Bruner, 1981, and the Discourse Acquisition Support System DASS, Hausendorf & Quasthoff, 2005). Whereas traditional approaches focus on asymmetry — where adults model and correct language — a more dynamic perspective, namely the Interaction Acquisition Support System (IASS, Behrens & Pfänder, 2024), recognises interaction as a collaborative and dynamic process. Based on the notions of contingency, fading and transfer (Van de Pol, Volman, & Beishuizen, 2010), we propose a more nuanced view of these three elements of caregivers’ scaffolding. Based on recordings of French families living in Germany, we present a refined scaffolding approach that provides a detailed understanding of how parents manage to limit their intervention while encouraging their child’s active agency.

How do words mean? Scholars have historically taken different views. Shakespeare (1597) and de Saussure (1916) emphasized signification ; Firth (1957) “you shall know a word by the company it keeps”; and Wittgenstein (1953) “in most cases meaning is use”. These maxims have influenced major contemporary research areas — signification imbues structuralism, semiotics, cognitive linguistics, psycholinguistics, embodied (4E) cognition, and neuroscience; lexical company inspires corpus linguistics, computer science, and artificial intelligence (AI); and usage-based approaches relate everyday social use to language acquisition and pedagogy. This chapter examines the historical and logical lineages of these ideas. Language is a complex adaptive system involving many types of agents — words and meanings, speeches and texts, speakers and social groups, cultures and customs. In language learning, signification, lexical company, and social usage are separable but symbiotic sources to meaning. In words, these sources of information are fused together. And who has done the fusing? You. You contain multitudes. You are a strange loop, an agentive self-referential hierarchical conscious system, who has hosted and dynamically merged these three complementary streams of information over the course of your lifetime, from before your first word, probably some version of mother, to now. This chapter reviews how.

Our goal in this chapter was to test a usage-based computational model of vocabulary learning (CLASSIC). CLASSIC simulates variation in vocabulary growth as a product of interactions between linguistic input quantity and the child’s current knowledge state, using an associative learning mechanism with a fixed capacity limited processing window. We directly compared the model’s results to those from children taking part in a number of empirical tasks and showed that, simply by varying the amount of input received by the model, we can successfully simulate a number of effects that we see in the children’s performance: individual differences in the rate of vocabulary growth between 19 and 30 months of age; correlations between vocabulary size, familiar word processing speed (as measured by Looking-While-Listening tasks) and phonological working memory capacity (as measured by non-word repetition tasks), and the effect of processing speed and non-word repetition performance on subsequent vocabulary growth. The results suggest that exposure to linguistic input, read through a fixed capacity processing constraint, leads learners to store phonological knowledge in chunks of information of varying size, and it is this stored knowledge that, at least partially, determines familiar word processing speed, non-word repetition performance, and vocabulary growth.

It is a standard assumption of construction grammar that grammar constitutes a network, known as the constructicon. Thus far, research on the constructicon has been mainly concerned with the taxonomic organization of grammar. However, a number of recent studies have argued that the constructicon is not only a taxonomy of hierarchically related grammatical patterns but that it also includes horizontal relations between constructions at the same level of abstraction (e.g., Cappelle, 2006; Diessel, 2019a; Sommerer & Smirnova, 2020; Van de Velde, 2014). Building on this view, the current paper argues that horizontal relations are crucial to the analysis of two general phenomena of grammar: construction families and grammatical paradigms. Construction families are groups of similar constructions that share some of their semantic and formal properties, whereas grammatical paradigms are pairs, or sets, of alternating constructions that profile a particular contrast or set of oppositions. Psycholinguistic evidence for horizontal relations comes from research on syntactic priming, linguistic productivity, blocking (or preemption), and neighborhood effects in language use and acquisition.

Usage-based approaches assume that children acquire grammatical constructions by linking form and meaning from ambient speech. Input shape is crucial in this process. One key construction type is verb-argument constructions (VACs), such as the ditransitive construction. Studies on the acquisition of English show that the verbs in VACs exhibit a Zipfian frequency distribution, with prototypical verbs occurring most often, facilitating learning. This study extends the investigation to L1 German by analyzing the development of the ditransitive VAC in a dense corpus of a child’s speech (aged 2–5) and caregiver input. Results confirm a Zipfian distribution of the ditransitive VAC verbs in the input, supporting abstraction of VAC meaning. In his ditransitive VAC production, the child initially uses prototypical verbs expressing transfer but quickly expands verb variety, and coercion occurs from early on. This rapid development is linked to the child’s growing construction network, communicative needs, and familiarity with transfer event scripts.

UB approaches have recently expanded to investigate the active role language learners have in unfolding conversations while using language to do things together (Behrens & Pfänder, 2022, p. 231). In this chapter, I will expand on this notion further from a pragmatic point of view, defined by Behrens (2009) as the relation between language forms and their functions with respect to language users as opposed to the world they are talking about (semantics). We will review how children’s language develops once they have surmounted the hurdles of prelinguistic development, and of learning words and early sentence structures, so that we find them able to string sentences together. At this point, managing topics extended in time relies on pragmatic skills that allow us to choose linguistic forms in harmony with our human cognitive systems — limited memory spans, fluctuating attentional states, and models of the world, the other, and the discourse — so that we can find joint purpose through communication. Here, cognitive-conceptual development and language development go hand in hand.

Drawing on conversation analysis and usage-based models of language, this chapter is part of larger on-going projects exploring the social-local worlds of second language (L2) speakers’ daily lives and investigating and mapping out their L2 learning, in situ and over time. The projects describe and analyze L2 learning as a matter of developing resources in and for social interaction — that is, they tackle L2 learning as a matter of locally instantiated, observable behaviors and long-term development of linguistic resources for accomplishing actions in concert with others in social interaction.

Drawing on a longitudinal corpus of L2 Icelandic data from the everyday life of a Canadian English-speaking learner of Icelandic, I will show how two linguistic patterns (subject-verb congruence in 1st person plural present tense and congruence between a modal verb and adjacent main verbs) are exemplar-driven in learning and how productive construction templates emerge in repair sequences.

Research on studying abroad (SA) suggests that input-rich contexts help L2 learners develop advanced word use embedded in target-like multiword constructions, or formulaicity. Native-like formulaic use is a key indicator of L2 proficiency, but little research has explored SA contexts to confirm its impact on formulaicity development. Using a dynamic usage-based (DUB) approach (Verspoor & Behrens, 2011), this study examines how verb-argument constructions (VACs) develop in written L2 English during SA. Data come from the SALA Diary Subcorpus (Pérez-Vidal, 2014), collected from 26 Catalan/Spanish advanced L2 learners over a semester (12–17 weeks) in an anglophone context. VACs were analyzed using TAALES (Kyle et al., 2018). A generalized additive mixed-effects model showed significant effects of random (participant behavior) and fixed factors (time), supporting the DUB claim of nonlinear, variable individual trajectories. Participants achieved significantly higher association strength scores for VACs. Further analysis confirmed the link between syntactic, lexical, and VAC indices with holistic formulaicity scores, implying that human raters consider all subdimensions of written L2 performance when making holistic judgements.

Taking a dynamic usage-based perspective (Verspoor & Behrens, 2011), this study explores the written English as a foreign language (EFL) development of adolescent learners in their first two years at a pesantren, a commonly found Indonesian Islamic boarding school in which English learning is promoted especially through extensive peer interaction. Usage-based approaches posit that languages are acquired through meaningful exposure and interaction. Research in second language acquisition has demonstrated that peer interaction can facilitate foreign language (FL) development, particularly when learners provide corrective feedback to one another (see Loewen & Sato, 2018). However, it remains unclear what occurs when FL learners primarily interact with each other, with minimal meaningful input from expert language users. Using qualitative analyses and generalized additive mixed modelling (GAMM), we found that (1) students rarely use corrective feedback, (2) prior knowledge of English makes a significant difference in EFL growth, (3) a discernible effect on EFL learning can be observed in year 1, whereas in the subsequent year, minimal to negligible progress is noted, and (4) the texts contained various non-target-like forms, especially L1 forms and constructions.

Parallel to the rise of usage-based, emergentist, and constructivist theories on language acquisition over the last decades, context and usage have also come into focus in the area of speech-language therapy (SLT), where the traditional medical model of a “specialist fixing the patient’s speech and language system” is increasingly called into question. Focusing on oral language disorders in monolingual and bilingual children, in particular developmental language disorder (DLD), I will detail the role of the cognitive underpinnings and of the linguistic and socio-interactional context, both for more theoretical research on the nature of these disorders and for applied research on SLT assessment, intervention, and prevention.

This introduction sketches current strands of UB research and outlines the structure of this edited volume, which pays tribute to Heike Behrens, who has contributed to and shaped usage-based (UB) research efforts from the start. The chapter describes how various UB approaches have emerged and evolved over the last three decades, reshaping how we view, research, and model language and interaction, first and second language development, and multilingualism. It also describes how UB approaches have embraced differing perspectives, disciplines, and methodologies that share a focus on authentic language use and experience, shedding new light on processes of language usage and development.

For well over a half-century, child language researchers have struggled with Chomsky’s characterization of “the poverty of the stimulus” — the assertion that the child does not receive enough linguistic data to arrive at knowledge of grammatical structure. Whatever one’s conception of linguistic competence, it is clear that the learner has to make use of both innate capacities and linguistic experience to master one or more native languages. This chapter focuses on experience — the speech encountered by the child learner. The past decades of child language research have revealed important ways in which “the stimulus” is really not so “impoverished”. Here I point to several examples of “the wealth of the stimulus” in my own research and that of colleagues. The learner gleans information about language from attending to ways in which linguistic forms change in the course of verbal interaction with others. Interlocutors expand and question children’s utterances; some of those modifications are taken up by learners. Interlocutors press for clarification, demand uptake. They rephrase their speech in successive utterances, offering a wealth of grammatical and lexical variety around a continuing topic. The interlocutors are not only caregivers, but also peers and older children who present different vocabularies and communicative demands. Children are actively engaged in linguistic worlds with many types of speech acts. Eventual explanations of language acquisition will have to attend to realistic descriptions of the rich data available to learners.

This article proposes a micro-sequential and multimodal approach to scaffolding in parent-child interactions. We argue for a more detailed understanding of how parents manage to transform their scaffolding into a time-limited intervention, transferring responsibility for the task to the child as soon as possible. In a social-interactional perspective following usage-based accounts of language acquisition (Tomasello, 2003), adults play a supportive role, helping children to develop linguistic and discourse skills through structured and dialogical interactions (cf. the Language Acquisition Support System LASS, Bruner, 1981, and the Discourse Acquisition Support System DASS, Hausendorf & Quasthoff, 2005). Whereas traditional approaches focus on asymmetry — where adults model and correct language — a more dynamic perspective, namely the Interaction Acquisition Support System (IASS, Behrens & Pfänder, 2024), recognises interaction as a collaborative and dynamic process. Based on the notions of contingency, fading and transfer (Van de Pol, Volman, & Beishuizen, 2010), we propose a more nuanced view of these three elements of caregivers’ scaffolding. Based on recordings of French families living in Germany, we present a refined scaffolding approach that provides a detailed understanding of how parents manage to limit their intervention while encouraging their child’s active agency.

How do words mean? Scholars have historically taken different views. Shakespeare (1597) and de Saussure (1916) emphasized signification ; Firth (1957) “you shall know a word by the company it keeps”; and Wittgenstein (1953) “in most cases meaning is use”. These maxims have influenced major contemporary research areas — signification imbues structuralism, semiotics, cognitive linguistics, psycholinguistics, embodied (4E) cognition, and neuroscience; lexical company inspires corpus linguistics, computer science, and artificial intelligence (AI); and usage-based approaches relate everyday social use to language acquisition and pedagogy. This chapter examines the historical and logical lineages of these ideas. Language is a complex adaptive system involving many types of agents — words and meanings, speeches and texts, speakers and social groups, cultures and customs. In language learning, signification, lexical company, and social usage are separable but symbiotic sources to meaning. In words, these sources of information are fused together. And who has done the fusing? You. You contain multitudes. You are a strange loop, an agentive self-referential hierarchical conscious system, who has hosted and dynamically merged these three complementary streams of information over the course of your lifetime, from before your first word, probably some version of mother, to now. This chapter reviews how.

Our goal in this chapter was to test a usage-based computational model of vocabulary learning (CLASSIC). CLASSIC simulates variation in vocabulary growth as a product of interactions between linguistic input quantity and the child’s current knowledge state, using an associative learning mechanism with a fixed capacity limited processing window. We directly compared the model’s results to those from children taking part in a number of empirical tasks and showed that, simply by varying the amount of input received by the model, we can successfully simulate a number of effects that we see in the children’s performance: individual differences in the rate of vocabulary growth between 19 and 30 months of age; correlations between vocabulary size, familiar word processing speed (as measured by Looking-While-Listening tasks) and phonological working memory capacity (as measured by non-word repetition tasks), and the effect of processing speed and non-word repetition performance on subsequent vocabulary growth. The results suggest that exposure to linguistic input, read through a fixed capacity processing constraint, leads learners to store phonological knowledge in chunks of information of varying size, and it is this stored knowledge that, at least partially, determines familiar word processing speed, non-word repetition performance, and vocabulary growth.

It is a standard assumption of construction grammar that grammar constitutes a network, known as the constructicon. Thus far, research on the constructicon has been mainly concerned with the taxonomic organization of grammar. However, a number of recent studies have argued that the constructicon is not only a taxonomy of hierarchically related grammatical patterns but that it also includes horizontal relations between constructions at the same level of abstraction (e.g., Cappelle, 2006; Diessel, 2019a; Sommerer & Smirnova, 2020; Van de Velde, 2014). Building on this view, the current paper argues that horizontal relations are crucial to the analysis of two general phenomena of grammar: construction families and grammatical paradigms. Construction families are groups of similar constructions that share some of their semantic and formal properties, whereas grammatical paradigms are pairs, or sets, of alternating constructions that profile a particular contrast or set of oppositions. Psycholinguistic evidence for horizontal relations comes from research on syntactic priming, linguistic productivity, blocking (or preemption), and neighborhood effects in language use and acquisition.

Usage-based approaches assume that children acquire grammatical constructions by linking form and meaning from ambient speech. Input shape is crucial in this process. One key construction type is verb-argument constructions (VACs), such as the ditransitive construction. Studies on the acquisition of English show that the verbs in VACs exhibit a Zipfian frequency distribution, with prototypical verbs occurring most often, facilitating learning. This study extends the investigation to L1 German by analyzing the development of the ditransitive VAC in a dense corpus of a child’s speech (aged 2–5) and caregiver input. Results confirm a Zipfian distribution of the ditransitive VAC verbs in the input, supporting abstraction of VAC meaning. In his ditransitive VAC production, the child initially uses prototypical verbs expressing transfer but quickly expands verb variety, and coercion occurs from early on. This rapid development is linked to the child’s growing construction network, communicative needs, and familiarity with transfer event scripts.

UB approaches have recently expanded to investigate the active role language learners have in unfolding conversations while using language to do things together (Behrens & Pfänder, 2022, p. 231). In this chapter, I will expand on this notion further from a pragmatic point of view, defined by Behrens (2009) as the relation between language forms and their functions with respect to language users as opposed to the world they are talking about (semantics). We will review how children’s language develops once they have surmounted the hurdles of prelinguistic development, and of learning words and early sentence structures, so that we find them able to string sentences together. At this point, managing topics extended in time relies on pragmatic skills that allow us to choose linguistic forms in harmony with our human cognitive systems — limited memory spans, fluctuating attentional states, and models of the world, the other, and the discourse — so that we can find joint purpose through communication. Here, cognitive-conceptual development and language development go hand in hand.

Drawing on conversation analysis and usage-based models of language, this chapter is part of larger on-going projects exploring the social-local worlds of second language (L2) speakers’ daily lives and investigating and mapping out their L2 learning, in situ and over time. The projects describe and analyze L2 learning as a matter of developing resources in and for social interaction — that is, they tackle L2 learning as a matter of locally instantiated, observable behaviors and long-term development of linguistic resources for accomplishing actions in concert with others in social interaction.

Drawing on a longitudinal corpus of L2 Icelandic data from the everyday life of a Canadian English-speaking learner of Icelandic, I will show how two linguistic patterns (subject-verb congruence in 1st person plural present tense and congruence between a modal verb and adjacent main verbs) are exemplar-driven in learning and how productive construction templates emerge in repair sequences.

Research on studying abroad (SA) suggests that input-rich contexts help L2 learners develop advanced word use embedded in target-like multiword constructions, or formulaicity. Native-like formulaic use is a key indicator of L2 proficiency, but little research has explored SA contexts to confirm its impact on formulaicity development. Using a dynamic usage-based (DUB) approach (Verspoor & Behrens, 2011), this study examines how verb-argument constructions (VACs) develop in written L2 English during SA. Data come from the SALA Diary Subcorpus (Pérez-Vidal, 2014), collected from 26 Catalan/Spanish advanced L2 learners over a semester (12–17 weeks) in an anglophone context. VACs were analyzed using TAALES (Kyle et al., 2018). A generalized additive mixed-effects model showed significant effects of random (participant behavior) and fixed factors (time), supporting the DUB claim of nonlinear, variable individual trajectories. Participants achieved significantly higher association strength scores for VACs. Further analysis confirmed the link between syntactic, lexical, and VAC indices with holistic formulaicity scores, implying that human raters consider all subdimensions of written L2 performance when making holistic judgements.

Taking a dynamic usage-based perspective (Verspoor & Behrens, 2011), this study explores the written English as a foreign language (EFL) development of adolescent learners in their first two years at a pesantren, a commonly found Indonesian Islamic boarding school in which English learning is promoted especially through extensive peer interaction. Usage-based approaches posit that languages are acquired through meaningful exposure and interaction. Research in second language acquisition has demonstrated that peer interaction can facilitate foreign language (FL) development, particularly when learners provide corrective feedback to one another (see Loewen & Sato, 2018). However, it remains unclear what occurs when FL learners primarily interact with each other, with minimal meaningful input from expert language users. Using qualitative analyses and generalized additive mixed modelling (GAMM), we found that (1) students rarely use corrective feedback, (2) prior knowledge of English makes a significant difference in EFL growth, (3) a discernible effect on EFL learning can be observed in year 1, whereas in the subsequent year, minimal to negligible progress is noted, and (4) the texts contained various non-target-like forms, especially L1 forms and constructions.

Parallel to the rise of usage-based, emergentist, and constructivist theories on language acquisition over the last decades, context and usage have also come into focus in the area of speech-language therapy (SLT), where the traditional medical model of a “specialist fixing the patient’s speech and language system” is increasingly called into question. Focusing on oral language disorders in monolingual and bilingual children, in particular developmental language disorder (DLD), I will detail the role of the cognitive underpinnings and of the linguistic and socio-interactional context, both for more theoretical research on the nature of these disorders and for applied research on SLT assessment, intervention, and prevention.