In this book the editors have brought together an engaging set of chapters focusing on complexity, accuracy and fluency (CAF) in second language acquisition (SLA) research. The different contributors address second language (L2) performance and/or proficiency by means of CAF from a theoretical, methodological or empirical perspective. As a result, the volume gives a comprehensive overview of current L2 research that is based on CAF. It extends the work presented in a special issue of 'Applied Linguistics' focusing on CAF edited by Housen & Kuiken (2009) and is a welcome addition to the field -- researchers using CAF in their own work will benefit from the knowledge presented here. In the following I first briefly review each chapter, then present an evaluation.
In chapter 1, (Complexity, accuracy and fluency: Definitions, measurement and research), Alex Housen, Folkert Kuiken and Ineke Vedder introduce the book by reviewing research into the central concepts, that is, linguistic Complexity (structural and lexical), Accuracy and Fluency, in short CAF. They stress that much earlier work has used CAF as dependent variables to measure L2 performance and proficiency. For example, L2 performance has been evaluated by means of the number of subordinate clauses per total number of clauses as a measure for structural complexity or a type-token based ratio for lexical complexity. Likewise, accuracy is accounted for using a global measure like ‘error free clauses’ or more specifically the ‘target-like use of articles’. Similarly, fluency has been measured by referring to speaking rate (e.g., the average number of syllables per second produced), as well as pausing and repairing behavior. The editors highlight that more recent work has put CAF in the focus of attention. That is, researchers have turned to investigating the constructs of complexity, accuracy and fluency themselves and come up with more precise definitions and better measures. As major challenges for CAF research they raise issues like the definition of CAF as theoretical constructs but also their operationalization as empirical variables, the relationship of CAF with underlying cognitive processes in the L2 and their manifestation in L2 performance and proficiency. Finally, the editors give a short summary of each chapter.
In chapter 2 (Defining and operationalising L2 complexity) Bram Bulté and Alex Housen present a detailed analysis of the construct itself the authors show how highly complex and multidimensional complexity is. They develop a taxonomic model that helps to unravel different aspects of complexity, e.g., the distinction between experienced relative complexity (difficulty) and absolute objective complexity. The latter is defined by a large number of different units (for example, syntactic nodes) and the number and type of interrelations between those units (for example coordinated or subordinated clauses). They then present more in-depth analyses of two constructs underlying linguistic complexity: structural/grammatical complexity on the one hand and lexical complexity on the other hand. In the second part they critically review 40 empirical studies that used complexity as dependent variables to measure L2 performance. The authors come to the conclusion that 'none of the complexity measures employed or recommended in the L2 research is unproblematic, neither in its computation nor in its interpretation.' (p. 40) and they recommend future work on L2 complexity.
In chapter 3 (Complexity, Accuracy and Fluency from the perspective of psycholinguistic Second Language Acquisition research) Richard Towell investigates the relationship of these performance based constructs with underlying representations, processes and mechanisms of L2 acquisition. The chapter first outlines Towell and Hawkins’ (1994) model of SLA that distinguishes (i) (innate) linguistic competence from (ii) learned linguistic knowledge and (iii) the mental representations for the procedures to process linguistic information. The author then analyses each of the CAF constructs using the model. For example, he explains that the development of accuracy and complexity depends on the relationship between linguistic competence and learned linguistic knowledge, while fluency holds stronger bonds with proceduralization of linguistic processes. Finally, Towell shows how the interpretation of L2 data benefits from relating them to the psycholinguistic model. He concludes that it would be desirable that more future work tries to link theoretical SLA research focusing on psycholinguistic models and empirical work investigating L2 performance.
In chapter 4 (Complexity, accuracy and fluency: The role played by formulaic sequences in early interlanguage development), Florence Myles starts by noting that in SLA it is often assumed that a structure is acquired as soon as it is present in L2 performance. However, extensive research on formulaic sequences suggests that, especially at beginning stages of development, L2 learners use complex language without analyzing the underlying units and structures, that is, formulaic sequences. After that, the author explains how the use of formulaic sequences affects CAF. For example, formulaic sequences tend to be more complex, generally accurate and fluent than the rest of a learner's production. Myles then presents corpus data from learners at the beginner and post-beginner levels of French, leading to the conclusion that the complexity, accuracy and fluency of beginners’ L2 performance is often illusory: 'During this early stage, the correspondence between the semantic and phonological representation is rudimentary, and does not involve syntax' (p.89). Future work therefore needs to take care in interpreting beginner performance by means of CAF and it would benefit from taking formulaic sequences into account.
Malin Agren, Jonas Granfeldt and Suzanne Schlyter wrote chapter 5 (The growth of complexity and accuracy in L2 French: Past observations and recent applications of developmental stages). They first review earlier work on developmental stages of L2 development and give a full description of six morphosyntactic stages for L2 French based on the model by Bartning and Schlyter (2004). In the second part of their chapter the authors focus on recent applications of the stages-model. Here they use the model to examine the development of plural marking and agreement in written French and to define L2 proficiency in both written and spoken French by means of CAF and the model. Finally, they introduce a newly developed software that can be used to automatically establish the developmental stage of an L2 learner's performance. The authors conclude that future work could examine in more detail the relations between CAF and the developmental stages that were presented here.
Following these more theoretical accounts of CAF as constructs, chapters 6 to 12 present empirical work using CAF as dependent variables to evaluate L2 performance.
Chapter 6 (The effect of task complexity on functional adequacy, fluency and lexical diversity in speaking performances of native and non-native speakers) by Nivja de Jong, Margarita Steinel, Rob Schoonen, Arjen Florijn and Jan Hulstijn compare in a large-scale study task performances of 208 L2 learners of Dutch with the performance of 59 native speakers. Participants performed eight different tasks that were manipulated by means of formality of context, their descriptive or persuasive nature, task complexity and interaction (monologic vs. dialogic). Results indicate that native speakers used more diverse vocabulary and were functionally more adequate in the complex tasks. Different results for different components of fluency were found. Non-natives speakers performed more poorly in complex tasks in terms of all fluency measures and functional adequacy but, like native speakers, their lexical diversity was higher on complex tasks. Generally, effects of task complexity were more pronounced in L2 speakers than L1 speakers. The authors conclude that apart from CAF measures, future research would benefit from evaluating functional adequacy and from distinguishing different aspects of fluency of task-based performance too.
Folkert Kuiken and Ineke Vedder present in chapter 7 (Syntactic complexity, lexical variation and accuracy as a function of task complexity and proficiency level in L2 writing and speaking) data of three studies that examined task-based performance of Dutch students learning French or Italian. The authors use both global and specific measures of syntactic complexity, lexical variation and accuracy. Results reveal that more complex tasks generally produced more accurate performance whereas findings with regard to syntactic complexity and lexical adequacy were mixed based on target language and language proficiency. Mode (written vs. spoken) did not create large differences in task performance as measured by complexity and accuracy while global and specific measures revealed interesting complementary information. The authors discuss their findings in relation to Robinson's (2005) Cognition Hypothesis and conclude that 'although a relationship between specific task features and specific performance effects exists, the claim of the Cognition Hypothesis that task complexity promotes linguistic complexity in general, is not confirmed by these findings' (p.166).
Chapter 8 (The effects of cognitive task complexity on L2 oral production) by Maya Levkina and Roger Gilabert is also framed around Robinson's (2005) Cognition Hypothesis. They focus on (combined) effects of two task-complexity factors (pre-task planning time and an increased number of elements) on oral performance by Spanish and Russian L2 learners of English. Data were analyzed by means of standardized measures for syntactic and lexical complexity, accuracy and fluency. The results showed no significant effects on accuracy or syntactic complexity -- which the authors attribute to the insensitivity of their measures. Data on lexical complexity and fluency partially corroborate earlier research showing that longer pre-task planning and simpler tasks result in more fluent speech while shorter pre-task planning and more complex tasks increase lexical complexity. These effects seem to strengthen each other; that is, differences in task performance were strongest between performance on the simple task with long planning time on the one hand and the complex task with short planning time on the other hand. The authors conclude that knowledge about task effects on performance is important for L2 pedagogy because it allows informed decisions e.g., about planning time or task sequencing.
In chapter 9 (Complexity, accuracy, fluency and lexis in task-based performance: A synthesis of the Ealing research) Peter Skehan and Pauline Foster summarize and synthesize the findings of seven earlier investigations into task-based performance. Participants of all studies were lower intermediate L2 speakers of English and one study included a native speaker baseline too. The authors are interested in the overarching conclusions of their earlier work regarding how task features and conditions (e.g., task structure, planning time) affect L2 performance by means of CAF. Their first conclusion is that future work needs to be very careful when choosing measures because each instrument reveals different and often complementary information. Secondly, the overall review allows them to draw firmer conclusions about the effect of planning time; that is, it increases accuracy and complexity but has differential effects on underlying aspects of fluency -- which again calls for complementary measures. Finally, they state that there are many more factors of task design that affect performance such that plenty of future research is needed in order to further develop theoretical frameworks like the Trade-off Hypothesis (Skehan 2009).
Chapter 10 (Measuring and perceiving changes in oral complexity, accuracy and fluency: Examining instructed learners’ short-term gains) presents work by Alan Tonkyn investigating whether CAF is able to detect gains in oral performance of participants taking a 10 weeks pre-sessional English for Academic Purposes (EAP) class. A total of 30 different CAF measures targeting grammatical and lexical complexity, accuracy and fluency were used to objectively measure gains from class initial to class final interviews of students. In addition, speech performance was subjectively rated by experienced raters using the IELTS speaking rating scale. Results reveal that precise and global CAF measures (e.g., number of words, general frequency of errors, length of pause-free run) can be used to show short-term gains of intermediate L2 learners of English. These align with subjective ratings but appear more sensitive than the global rating scales. Furthermore, the study shows how CAF measures at times can 'correct' halo-effects of single raters.
In chapter 11 (The development of complexity, accuracy and fluency in the written production of L2 French) Cecilia Gunnarsson uncovers the differential development of writing performance of five Swedish L2 learners of French. The author used writing logs and video-taped think-aloud protocols to investigate the development of her participants' writing over a time of 30 months. The chapter focuses on a comparison of two written performances and reveals large individual differences in the development from time 1 to time 2. While some learners showed gains in accuracy at the expense of fluency others prioritized fluency at the cost of accuracy. Complexity measures did not generate any obvious patterns. The author discusses her findings in terms of Skehan's (2009) Trade-off Hypothesis and Levelt's (1989) distinction of conceptualization and formulation.
Stefania Ferrari describes in chapter 12 (A longitudinal study of complexity, accuracy and fluency variation in second language development) the development of four adolescent L2 learners and two native speakers of Italian. Participants performed various monologic and dialogic tasks over three years. Results show how gains in e.g., complexity create trade-offs in other components, be it accuracy or fluency. Generally, monologic tasks created greater complexity but lower fluency than dialogic performance. From the detailed comparison of the L2 and L1 speakers the author concludes that 'the ability to vary one's language according to the demands of different communicative activities [a skill L1 speakers have] develops very slowly and does not seem to be fully acquired even by highly proficient L2 learners' (p.294).
In the final chapter (Epilogue), the editors recap the content of the volume and conclude: 'Researchers […] seem to agree on the usefulness and validity of complexity, accuracy and fluency as research constructs. However, this is where the consensus ends and the controversy begins' (p.300) As there is still a lack of agreement on, e.g., the definition of the three constructs, the editors remind us to be careful when using CAF and especially when interpreting findings based on CAF. Future research will yield more insights.
'Dimensions of L2 performance and proficiency -- Complexity, Accuracy and Fluency in SLA' is a welcome addition to SLA research, especially since the last volume on CAF (Wolfe-Quintero, Inagaki and Kim, 1998) is no longer available. Research gathered here constitutes a comprehensive review of recent work into dimensions of L2 performance and proficiency. It is in particular rich in addressing relevant issues from many different perspectives, combining theoretical accounts on CAF as topic of investigation (e.g., the search for a definition of complexity in Bulté & Housen), looking at L2 learner performance of beginner levels (Myles) or at various stages of development (Agren et al.), synthesizing reviews of earlier studies that used CAF as a dependent variable (e.g., Kuiken & Vedder and the Ealing Studies of Skehan & Foster), original investigations into task effects expressed by CAF (e.g., de Jong et al., Levkina & Gilabert), as well as interesting developmental accounts for short-term group gains (e.g., Tonkyn) or the longitudinal development of multiple cases (Ferrari).
A further strength is the inclusion of many different linguistic contexts with a variety of source and target languages, investigating oral and written production, focusing on monologic and dialogic performance of native speakers and L2 learners at various levels of proficiency when performing a large variety of different tasks. For researchers interested in using CAF it will be a valuable source on theoretical and methodological issues to consider in future work.
Especially the editors’ introduction summarizes the main areas of debate as well as current definitions of the constructs will be of interest to researchers and may be of use for introducing CAF-based SLA research at the MA level. Each individual chapter is valuable, at times only for specific audiences, e.g., Towell's psycholinguistic account of CAF or Gunnarsson's review of individual development of written L2 production. The fact that well-known scholars and young researchers both present their perspectives on CAF shows again the volume’s wide scope.
There are a few critical points to mention. First, the audience is a research community: SLA practitioners may struggle with the theoretically-oriented content. Second, even though the title suggests that the book addresses performance and proficiency, there is more on the former and the latter is not that well covered. Finally, a third point may be considered a strength: a reader hoping to find answers about CAF could be disappointed since the studies open many more new questions than they answer. As such, the volume is an interesting research oriented collection of innovative work that critically reviews the constructs of complexity, accuracy and complexity in second language research.
Bartning, I., & Suzanne Schlyter, S. (2004). Itinéraires acquisitionnels et stades de développement en français L2. Journal of French Language Studies, 14, 281-299. doi:10.1017/S0959269504001802.
Housen, A., & Kuiken, F. (2009) Special Issue: Complexity, accuracy, fluency (CAF) in second language acquisition. Applied Linguistics, 30(4).
Levelt, W. J. (1989). Speaking: From Intention to Articulation. Cambridge, MA: MIT Press.
Robinson, P. (2005). Cognitive complexity and task sequencing: Studies in a componential framework for second language task design. International Review of Applied Linguistics, 43(1), 1-32.
Skehan, P. (2009). Modeling second language performance: Integrating complexity, accuracy, fluency and lexis. Applied Linguistics, 30(4), 510-532.
Towell, R., & Hawkins, R. (1994) Approaches to Second Language Acquisition. Clevedon: Multilingual Matters.
Wolfe-Quintero, K., Inagaki, S. and Kim, H. (1998). Second Language Development in Writing: Measures of Fluency, Accuracy, and Complexity. Honolulu, University of Hawai’i: Second Language Teaching and Curriculum Center.