I get asked from time to time about usage-based versus universal grammar (UG) perspectives regarding language acquisition. I usually end up summarizing bits of the histories of these two perspectives. Better yet would be to provide a short list of articles that do a more thorough job at walking readers through this history. This is that list:
Goldschneider, J. M., & DeKeyser, R. M. (2001). Explaining the “natural order of L2 morpheme acquisition” in English: A meta‐analysis of multiple determinants. Language learning, 51(1), 1-50.
This article's literature review takes readers back to the first findings by Brown and the children he was researching in the 1970s. He noticed the children developed certain English grammatical endings in similar orders. This sparked interest among researchers regarding how consistent this ordering of acquired morphemes (parts of words) is among all children learning English. The authors of this study then investigated several potential causes of the observed developmental orders.
Tomasello, M. (2000). Do young children have adult syntactic competence?. Cognition, 74(3), 209-253.
This article summarizes research regarding item-based (usage-based perspective) versus rule-governed (UG perspective) observations in the development of grammar.
Christiansen, M. H. (2019). Implicit statistical learning: A tale of two literatures. Topics in Cognitive Science, 11(3), 468-481.
This article ties two separate stands of research together, showing how they have essentially repeated the same work.
Saffran, J. R., & Kirkham, N. Z. (2018). Infant statistical learning. Annual review of psychology, 69, 181-203.
This article summarizes a history of research that initially looked into how very young children were able to start separating speech streams heard from adults into separate words. Soon after, researchers began to explore whether such abilities in such young children were specific to language, or were also used in recognition of visual, sound, and other non-language stimuli. A bit of this research even extended to learning by non-human animals.
I will add more articles when I think of more; for now, less is more:)
Image from Google Images "labeled for reuse"
I often see proficiency-oriented language teachers bring up "time" as part of the motivation for choosing certain teaching practices. Comprehension-based teachers regularly discuss the vast differences in hours of input that a child is exposed to before reaching kindergarten versus limited hours that students in most language classrooms receive. Some teachers argue that there isn't time in a classroom to provide input the way children at home naturally receive, so their solution is to explicitly teach grammar and then push students to practice producing teacher-determined linguistic patterns (typically sounds, words, grammatical structures, and pragmatic actions like asking-answering, greeting, and so on). I think these discussions can benefit from the concept of leverage, which I'm defining here as: Hours spent providing one type of learning experience that are observed to correlate with specific and measurable learning outcomes.
High-leverage Teaching Practices (HLTPs) are explored in the book by Glisan & Donato (2017) through ACTFL. Kearney (2015) looked at one HLTP, leading an open-ended group discussion, in advanced Spanish and Latin classrooms, and Hlas & Hlas (2012) first advocated bringing the HLTP concept over to language education from its original place in math education. Most recently Yue (2019) looked at HLTPs in an introductory-level Chinese language classroom (thanks are owed to Diane Neubauer for bringing this last paper to my attention).
To illustrate the usefulness of talking in terms of leverage, I'll offer a made-up example: One teacher spends the first 20 hours of class time with beginning novice learners of some new language delivering content-based instruction in a full-immersion manner. The teacher only talks at fast native-like speed, and pictures and ongoing context help learners infer the meaning of the teacher's talk. A second teacher delivers similar content-based lessons but speaks very slowly and with pauses in between chunks of words to give learners time to process each sentence that the teacher says. At the end of the 20 hours, all of the learners are given three assessments: one in free-speaking, where learners each draw pictures about whatever they want to talk about in the target language and then they talk about those drawings. The next assessment checks the learners' understanding of fast native-like speech. The third assessment checks their understanding of slow speech. In this imagined scenario, the learners who only heard fast speech (1) score worse than the other group in speaking, (2) are able to pick out more words accurately in the fast listening assessment than the other group, and (3) score equally high in the slow-listening assessment. The learners who only experienced hearing slow and segmented speech perform (1) far better in the free-speaking assessment, (2) pick out far fewer words from the fast listening assessment compared to the other group, and (3) score equally high in the slow listening assessment. So we see the 20 hours spent using the micro-practice of slow versus fast listening resulted in different outcomes for each of the three assessments. Leverage being higher or lower depends on the outcomes measured, but which outcomes are more important depends on the teacher and connected stakeholders (learners, families, colleagues, and so on).
I think it would be productive for teachers to talk about relative measures of leverage ("higher than teacher practice A for X outcome," "Lower than teacher practice B for Y outcome"). A simple measure can be hours, or fractions on an hour, that the teachers spend enacting a particular teaching practice and, consequently, learners spend experiencing that practice. In the imagined examples above, 20 hours spent providing slow speech for beginning novice learners proved to be higher leverage than providing fast speech for the desired outcome of learners of the same proficiency level being able to perform in a free-talking assessment. However, those same 20 hours spent experiencing fast-speech proved to be higher leverage than slow-speech for the outcome of picking words out when listening in the fast speech assessment.
Leverage allows us to talk about short-term developmental goals on the way to reaching advanced proficiency. One example is novice learners in TPRS classrooms writing and telling stories using a minimal set of vocabulary (see Lichtman, 2019, for a review of TPRS classroom research).
A teaching practice might be high leverage for certain learning outcomes for novice learners, but that same teaching practice might be lower leverage if done for intermediate and advanced proficiency learners. The same can be said of a teaching practice that is high leverage for a learning outcome for advanced proficiency learners but lower leverage for novice and beginning learners for that same outcome (obviously, we expect different learning outcomes at different proficiency levels). My hope in drawing out these examples is that teachers can talk past each other less, and get more specific about what we want to see in our learners and what we can do to get there.
Here is a link to my presentation from 2019 SLRF:
This summer (July 2019) I presented the opening day's plenary at the Agen Workshops in Agen, France. Here are my slides:
I owe special thanks to Mery Diez Ortega for inviting me yesterday to present a comprehension-based language teaching workshop in her SLS 303 course at the University of Hawai‘i. I'll offer readers here context for the main idea that I intend to share here by first describing the workshop, and then I'll explain how I believe teachers and researchers can more productively talk about comprehension-based approaches to language teaching. At yesterday's workshop for language teachers, I passed around copies of the handout (file below), then I spent about ten minutes introducing myself, TPRS, and comprehension-based approaches to language instruction (presentation slides in file below). Then I launched into a TPRS Mandarin Chinese "story-asking" whole-class task. I continuously invited learners contributed to help complete a story about Profe Mery traveling the world to find the hot pot she desired (from the student's ideas). Then we read a new story together, chorally from the projector screen. It was a calm class of adults, and I saw a lot of smiles, accurate gestures, and attentive eyes throughout the story-asking segment. The speed of their responses showed me which words and sentences they were understanding more quickly, allowing me to choose when to add new words to our interaction. The class read increasingly loudly as we read chorally from the screen. We had a brief Q&A session after, and I answered their questions, mainly about curriculum design and assessment.
The main goal for this post is to propose a way for teachers and researchers to talk about TPRS as a comprehension-based approach to L2 learning while also talk about about Story Listening and ALG as comprehension-based approaches, when these approaches (and often teacher beliefs about what learning experiences learners need) look very different. TPRS lessons tend to be highly interactive, yet still comprehension-based because only those learners who are ready to speak are expected to speak, leaving the teacher typically the person in the room producing most of the language. Learners respond with gestures, facial reactions, a word from the learners' native language, or a simple word or short phrase from the target language (if they "have" it). By contrast, Story Listening (SL) and Automatic Language Growth (ALG) instructional practices tend to emphasize focused listening and focused reading. SL teachers can vary in how often and what kinds of responses learners can contribute during the story. A mundane comparison I offered in the workshop yesterday was that of a comedian on stage who tells a joke, and in response nobody laughs. Even though audiences are discouraged from speaking during the performance (the person on stage will often treat "talkers" in the audience as "hecklers" or generally problematic), the comedian is continuously looking for audience reactions in the form of laughter and facial reactions, e.g. smiles, looks of surprise, etc. The performer thus has continuous opportunities to modify their talk in real time. Other speaker-audience contexts can include a lecturer lecturing and a storyteller sharing a story around a campfire. These kinds of settings in which response from the audience is highly restrictive can resemble what SL and ALG report about talking to groups of learners. Further still, there are segments of class time where learners are invited to choose a book from a free voluntary reading (FVR) library, and just read on their own time. Learners may also be tasked with watching a video segment, or listening to audio, in class or at home. In these kinds of learning experiences, there is no opportunity for the author to modify their talk or text in response to the learner in real-time.
With these differences in mind, I propose that teachers and researchers talk about comprehension-based approaches to language instruction as following a continuum in "intensity of interaction". I borrow this term, "intensity of interaction," from Bardovi-Harlig (2013), and I'll include the beginning of her paragraph to offer readers here context for the term: "The classic measure of study abroad is length of stay (LOR, length of residence, in early SLA studies). It is easy to measure and learner self-reports are reliable; however, the relevance of length of stay as a meaningful variable has been severely criticized. Dietrich, Klein, and Noyau (1995) concluded that “[d]uration of stay is an uninteresting variable. What matters is intensity, not length of interaction” (p. 277)" (p. 80, underline added here). "Intensity" in this and in a prior study by the same author (Bardovi-Harlig & Bastos, 2011) only measure time spent talking to native speakers, talking with other international students, and time spend watching TV in the target language, in self reports on surveys. Study abroad and classroom contexts share in common the learners' potential exposure to target language input. They also share in common a variety of situations ranging from purely receptive, like watching TV or reading a book, road signs, or menus, to minimally interactive, such as seeing a live performance, to relatively interactive, such as asking a simple question about directions to a destination, and nodding to show understanding while a local resident explains directions. This is, more-or-less, how I explained the "interaction continuum" for comprehension-based approaches in the workshop yesterday (in slides, below).
I will use the term "intensity" here not to refer to time spent in getting exposure to input, as Bardovi-Harlig & Bastos (2011) did, but instead in a more micro-level view in terms of turn-taking and the degree to which participants invite each other to contribute to the discussion. I'll also not use "intensity" here to refer to emotions or attention. A learner can very intensely read a book (with great interest or for a long time), intensely listen to a live storyteller while processing for meaning, and very intensely shout an idea in the context of a whole-group collaborative discussion. I'll distinguish intensity of interaction here in terms "way-ness," ranging from fully one-way to fully two-way interaction, following Long's (1983) discussion of one-way versus two-way interaction. Here, low intensity interaction refers to fully one-way interaction (reading, listening, or viewing pre-written or pre-recorded texts for videos, respectively). High intensity interaction here refers to fully two-way interaction, that is, audience members are continuously encouraged to contribute to the discussion, verbally or non-verbally. Other terms may replace this later, if needed. I can already imagine 'way-ness of interaction' and 'intensity of participant contributions' as alternate labels for this continuum.
Thanks also to Diane Neubauer for first prompting this discussion by pointing out that a lot of research on comprehension-based instructional practices, such as FVR libraries, look far different from the interaction we commonly observe in TPRS classrooms.
Bardovi‐Harlig, K. (2013). Developing L2 pragmatics. Language Learning, 63, 68-86.
Bardovi-Harlig, K., & Bastos, M. T. (2011). Proficiency, length of stay, and intensity of interaction and the acquisition of conventional expressions in L2 pragmatics. Intercultural Pragmatics, 8(3), 347-384.
Files for download:
I created this document to illustrate how one might use free concordancing software to (1) count and rank frequency of lexical items (words) in a collection of texts, and (2) copy and paste into Excel all of the sentences that represent instances of usage of a single lexical item to better identify various context-dependent meanings. I show how Excel can be well suited for annotating, categorizing, and ranking different kinds usage and meaning.
This is a response I wrote for a language teacher group on FaceBook, in response to a question regarding linguistic knowledge versus general knowledge:
VanPatten has said on his show that he sees language to be made from a universal grammar and a general learning architecture. That general learning architecture is not referring to the explicit memorization of rules that we need for learning physics, history, etc. It's largely the associative processes that cognitive scientists study, whereby exposure to many things (e.g. Birds, furniture, plants) develop categories in our minds simply through exposure many times, and through the similarity and frequency of features (wings, legs, eyes, color) across the many examples we see in our experience. If I ask any person "can a bird have six legs", they won't use any memorized rules, nor will they need any strenuous thinking to answer. It is unlikely anyone ever told them how many legs a bird is supposed to have. But it's just as easy to judge as it is to judge an ungrammatical sentence. This is the area that researchers in Cognitive Grammar and Usage-based Linguistics explore, which VanPatten mentions on his show.
In recent years, a lot of great reports about student language success have trickled in from teachers on the iFLT/NTPRS/CI Teaching Facebook feed. Many teachers upload student writing samples, videos of students speaking, and report on great things like students coming back from summer break still able to do most of what they could before they left, months before.
As we talk about more research to be done, one might fear that without a tight and well-planned research design, all of this evidence is useless for research. But this should not be the case. One thing we can do is simply to organize these samples into collections of pilot data.
In my experience submitting presentations to graduate student conferences, I state up front that I'll be presenting "pilot data" as part of an "initial study design". This kind of hedging informs audiences of the incompleteness of the study before they attend my talk. I've usually found that as long as I state that there is much more work to be done, audiences often raise their hands not to poke more holes in my study, but instead to ask guiding and helpful questions.
Research is an ongoing process, and it requires a community of interested people to push it forward. For now, I believe we need lots of evidence of students returning from summer break, still able to read and speak, as evidence of acquisition and learning for long term memory, etc. The more files of this sort we have (video of students reading, reading quiz answers with descriptions of the testing conditions, videos of students spontaneously speaking, maybe describing a picture), the more clearly we can write up IRB applications and consent forms, and cite prior research literature to better design the next studies based on that pilot data.
In sum, don't be afraid to collect data just because you don't yet have a tight research design. Pilot data is the normal first step in research.
Reed Riggs (Author)
I hold a Ph.D. from the University of Hawai‘i. My research looks at entrenchment, frequency effects, and salience along with interactional behaviors from Usage-based Linguistics (UBL) and Conversation Analytic (CA) perspectives.