This research investigates the effects of online discussion on L2 learning of word forms and text comprehension from the perspective of FonF. With the hope to thoroughly examine this learning process, this study applied analysis of variance (ANOVA) combined with analysis of variations in the data, with the elaboration given below.
This study applies both methods to analyze the data because while working together, they offer a comprehensive view on the possible learning effects brought by the interventions. ANOVA serves as a straightforward quantitative method to analyze the data, proving the effectiveness of the FonF interventions through collapsed mean scores. On the other hand, analysis of data variation investigates the diachronic development of the data, thus enabling more intriguing research results to emerge. As analysis of data variation is an approach not adopted as often, how it originated theoretically and how it analyzes data are explained in this section.
4.1.1 Theoretical bases of the analysis of data variations
The analysis of variations in the data is based on Dynamic Systems Theory (DST). DST suggested that all systems are composed of sub-systems, which interact with one another continuously. Such interactions among sub-systems and factors influence the larger systems, which accordingly evolve in a chaotic and unpredictable way (de Bot, Lowie, & Verspoor, 2005).
That DST can be applied in language learning research is founded on the concept that language development itself can be regarded as a dynamic process (Lowie, &
Verspoor, 2015). Factors of language learning, such as learning of pronunciation,
34
grammar, vocabulary, other sub-systems of languages, and learner factors, no matter detectable or not, keep affecting one another. Their relationships continuously change over a long period of time, and thus contribute to the complexity of language development.
Some concepts of DST are applicable to the data analysis in this study as some theoretical bases of FonF learning is consistent with them. For one, the inclusion of word form learning and text comprehension based on “balanced development” of FonF learning (Han, Park, & Combs, 2008) actually represents an attempt to look into the relationship between these two sub-systems of FonF language development. For another, as elaborated previously, FonF intervention might need a rather long period of time to take effects (e.g. Swain & Lapkin, 2001; Alcon, 2007), which supports the claim of DST that evolution of a system (language development in this regard) is a longitudinal and dynamic process.
Based on the justifications above, this study applies both ANOVA and analysis of variations, founded on DST, to analyze the data. While ANOVA provided a quantitative method and a clear insight into the results, analysis of data variations was on the other hand more likely to enable the researcher to better examine the process of the learners learning word forms and improving their text comprehension with the aid of FonF intervention.
4.1.2 Interpreting data variations based on DST
Though most DST studies look at variations in individual cases (Lowie &
Verspoor, 2015), this study chose to analyze variations in diachronic mean scores of the whole group considering its research focus on effects brought by online discussion.
It is assumed in this study that interactions among individuals may facilitate learners’
learning of words and their text comprehension. As the EG participants were divided into different heterogeneous groups, each of which developed their own interpersonal
35
relationships and interaction modes, different learning outcomes would be unavoidably produced in each group. Since differences among different groups are not the research focus, it would be better to analyze the diachronic mean scores of all the participants so that the learning effects of the FonF interventions can be examined in a more comprehensive way. Therefore, instead of looking into language development of individuals as many other DST studies, this study attempts to examine the learning effects brought by FonF interventions by analyzing the mean scores of all the participants.
To do so, this study borrows some of DST’s methods of interpreting variations.
According to DST, system evolution is a dynamic process that constantly undergoes changes. Instead of a stable state, the evolution of a system is full of continuous fluctuations, which result from interactions among sub-systems. As a matter of fact, it is through observing and analyzing these fluctuations that DST researchers could gain some insights into the development of the whole system.
Some fluctuations are especially significant and worth noticing by DST researchers. Through the whole process of system evolution, there might be “attractor states,” indicating periods of time when less variations occur and the development seems relatively stable for the time being (de Bot et al., 2005). In attractor states, sub-systems keep interacting with one another but by lesser degrees. Variations would seem minor until some significant ones are observed. Such states with great fluctuations are called states of “free variations.” According to DST, free variations indicate that some external factors are influencing the system, interrupting the original stable states. In the case of language development, it might be inferred that during states of free variations, some learning brought by exposure to input or interventions are occurring. Learners’ interlanguages are developing (though not necessarily toward the target language). Free variations, with great fluctuations in learners’ language
36
development, must occur before learners’ interlanguages stabilize again.
This study adopted the method of data interpretation elaborated above, examining variations in the WF and comprehension tests scores to see what learning effects might be brought by the FonF interventions and how they influence the participants’ language development in the process.
4.2 Results: Effects of FonF Intervention on Learning of Word Forms