Effects of FLA on L2 performance

As is the case of other individual variables, research into FLA has focused primarily on its effects on learners’ performance and ultimate achievement in the L2 (e.g., Kleinmann, 1977; Young, 1999;

MacIntyre, 1999; Horwitz, 2001). Most of the studies used a correlational design and global measures of L2 achievement such as course grades and standardized proficiency tests. Some studies have also examined the relationship between FLA and specific performance measures, for example cloze tests, translations, self-descriptions, speech content, prose writing (e.g., Steinberg & Horwitz, 1986; Gardner &

MacIntyre, 1993). Another interesting feature of FLA research is that most studies focused on L2 speech production, and only a fewer number of studies investigated how other language skills are affected by FLA (see section 2.2.4.4.3).

Reviews of anxiety literature (e.g., Tóth, 2007; Horwitz, 2010) confirm that the findings have been overwhelmingly consistent when FLA is conceptualized as a situation-specific anxiety construct as described in section 2.2.4.3. Anxiety appears to have a negative impact on both general and specific outcome measures and with regard to the four core language skills (Dörnyei, 2005; Tóth, 2007). In the following I will give a brief account of three lines of inquiry as regards the effect of anxiety on L2 performance, all of which are highly relevant for the project described in the present dissertation: (1) the cognitive effects of anxiety, (2) the interplay between anxiety and other ID factors, and (3) the effects of FLA on performance outcomes in the four language skills.

2.2.4.4.1 Cognitive processes and foreign language anxiety

Understanding the way FLA interferes with the specific cognitive processes involved in L2 learning and communication may help us explain why anxiety has been shown such a reliable correlate of

L2 achievement. In a cognitive psychological view of anxiety effects, cognitive processes are interrupted or distracted by aroused anxiety: worrying about a potential failure draws one’s attentional resources away from the task, and as a consequence, performance is inhibited (Eysenck, 2000). A study by Chamorro-Premuzic and Furnham (2003) examining the relationship between primary personality traits and academic performance concluded that the cognitive “worry” component of anxiety impairs performance on those tasks that require the involvement of working memory. Similar findings were reported earlier by MacIntyre & Gardner (1991) who found that students with higher levels of language anxiety obtained significantly lower scores on a digit-span test and a vocabulary production task. They argued that the memory deficits shown by this performance can be attributed to FLA. As working memory is considered to be responsible for the regulation of attention (see section 2.2.2), this finding is clearly in line with Eysenck’s cognitive interference view of anxiety, a view that is based on the limited attentional capacity model (see section 2.1.4).

Attention, however, does not only mediate the production of L2 output, but the encoding and processing of information that precedes language production. If anxiety indeed interferes with the allocation of attention during task performance, it probably does so throughout the information-processing cycle. MacIntyre and Gardner (1994) set out to test these subtle effects of anxiety on the L2 learning and performance process. They applied Tobias’ (1986) model of the cognitive effects of anxiety as a framework for their experiment. This model describes learning as a three-stage process involving:

(1) an input stage, which concerns the initial representation of language items in memory by the selection and encoding of external stimuli (i. e., making sense of a complex sentence),

(2) a processing stage involving the cognitive operations performed on the items, e.g., organization, storage and retrieval of information (i.e., learning a list of new words), and

(3) an output stage, which refers to language production (i.e., using language to communicate).

It is important to note that these stages are isolated for the purpose of studying their characteristics, in this case the effects of FLA at the different stages of the learning process. However, it would be difficult to specify where exactly one stage ends and the next stage begins. Learning a language is a continuous process, so not only are these stages interdependent, that is the successful completion of one is necessary for the next, but they also occur simultaneously (see section 2.3 on language production). Consequently, the negative effects of anxiety on language production measures may indicate problems at any or all of these stages.

Figure 4

Anxiety and language processing

N o t e . Based on MacIntyr e and Gardner, 1994 and Piechurska-Kuciel, 2008.

According to the model, language anxiety affects each stage of the cognitive processing of language production in such a way that anxiety at one stage causes increased levels of anxiety at subsequent stages. At the input stage learners need to take in information, which may cause anxious students to ask for repetitions or read materials several times. Anxiety present at the input stage further raise anxiety levels experienced at the processing stage, when students have to organize the information they gathered. Language learners usually encounter the highest levels of anxiety at the output stage of

input processing output

Language anxiety

language processing due to the fact that increasing anxiety levels inhibit the necessary cognitive processes by taxing working memory capacities.

MacIntyre and Gardner (1994) developed an 18-item self-report questionnaire with six items measuring each stage of learning identified by Tobias (1986). They used course grades to assess global proficiency and several specific measures to assess performance at the different stages. They correlated the students’ input, processing and output anxiety (IPOA) and their performance on both the global and specific measures of language production. Their findings confirmed Eysenck’s (1979) cognitive interference model of anxiety and showed that anxiety arousal affects all stage-specific tasks negatively.

They concluded that anxiety interferes with cognitive processes by depleting learners’ attentional resources, therefore anxious students do not only have difficulty in demonstrating their knowledge at the output stage, but their knowledge base may also be smaller as a result of interference at the earlier (input and processing) stages.

Eysenck (1979) further hypothesized that anxious students can compensate for their performance defects caused by insufficient cognitive processing by investing more effort into their studies. This idea has received mixed empirical support (Kleinmann, 1977; Price, 1991), but a study by Horwitz’s (1996) is especially noteworthy in this regard as it provides evidence that the negative effects of anxiety can be avoided or reduced. Horwitz showed that non-native language teachers, who can clearly be considered examples of successful language learners, may also suffer from FLA which may even influence their classroom practices and instructional preferences.

The differential effect of individual anxiety levels on language output is highly relevant in this dissertation as it examines the influence of the interaction of learner variables and task complexity on language production. With this regard a recent study by Robinson (2007b) found that cognitive task complexity correlated negatively with the grammatical complexity of L2 speech and that for students with

low output anxiety more complex tasks resulted in more complex speech production. This effect, however, was not detectable for students with high output anxiety.

2.2.4.4.2 The interaction of FLA and other IDs in L2 learning

Research on FLA has shown that it is a highly complex construct which can be related to a number of affective, ability and social factors. It has also been suggested that reciprocal paths may exist between FLA and certain individual learner variables (Gardner & MacIntyre, 1993). For example, emotionality and certain personality traits can make an individual be more prone to higher levels of anxiety arousal, while aroused anxiety may, in turn, impair one’s cognitive abilities resulting in lower level performance.

Poor performance may also reinforce anxiousness (see the classical conditioning and the self-presentation models of social anxieties in section 2.2.4.2), which can result in anxious individuals being caught in a vicious circle. Thus the implications of research into the interplay of anxiety and other ID factors is particularly relevant for language pedagogy as it may shed light on what kind of learners are more (or less) likely to experience anxiety about L2 learning/use and on how task design may help reduce their anxiety level. Table 7 provides a summary of important studies in the field of anxiety-ID interaction.

Table 7. Studies on the interaction between anxiety and other ID factors

Study Focus

Bailey, 1983 Learner beliefs, self-perception, self-comparison with others, competitiveness Young, 1991 Learner beliefs, personal & interpersonal anxieties

Gardner & MacIntyre, 1993 Self-ratings of L2 proficiency, attitudes, motivation, social milieu Sparks & Ganschow, 1995,

2001

Language aptitude, native language skills

Saito & Saminy, 1996 L2 proficiency

Onwugbuzie et al., 1999 Self-perception, age, learning experiences, academic achievement, history of visiting foreign countries

Gregersen & Horwitz, 2002 Perfectionism (personality trait)

Matsuda & Gobel, 2004 Academic achievement, overseas learning experience

Woodrow, 2006 Cultural background

MacIntyre, 2007 Willingness-to-communicate

Tóth, 2007 L2 proficiency, aptitude, motivation, perfectionism, competitiveness, L2 self-concept

Dewaele et al., 2008 Age, gender, education level, number of languages known, age of onset of learning the L2, trait emotional intelligence, social milieu, self-perceived competence, context of acquisition

As this brief summary shows, exploring the interaction of FLA and several individual variables has received much attention from researchers over the last decade. In their extensive and nuanced study Dewaele et al. (2008) outline interesting paths for future research namely, that follow-up studies on certain individual, contextual and situational characteristics would help to further illuminate the complex

interaction of the variables. For example, incorporating the effects of the age of onset of L2 (L3, L4) acquisition, the consistency of FLA in more than one foreign/second language, the context of acquisition, and “sociobiographical” variables on anxiety experienced in different communicative situations may be especially important to further elucidate the concept of FLA.

2.2.4.4.3 FLA and L2 skills

In their review of FLA literature Horwitz (2010) and Tóth (2007) have shown that most studies examine the link between anxiety and L2 spoken output. It is reasonable to assume that speaking may arouse anxiety to a greater extent than writing: (1) speaking in certain specific situations may be perceived as stressful even in the mother tongue, e.g., stage fright, fear of public speaking or telephoning), and (2) speech production is more stressful than writing because of the lack of preparation time (see section 2.3.3 on the differences between oral and written production in the second language). Consequently, output anxiety can be expected to differentiate learners’ performance in speech to a greater extent than in writing.

Studies on the relationship between anxiety and L2 spoken production have overwhelmingly concluded that anxiety had a significant negative impact on both the quantity and the quality of L2 speech (Steinberg

& Horwitz, 1986; Philips, 1992; MacIntyre & Gardner, 1994).

More recently a few studies have also investigated anxiety effects on the other language skills (reading, listening, and writing) (see Table 8).

Table 8. Studies on anxiety and reading, listening, and writing

Study Reading Writing Listening

Hilleson, 1996 + +

Saito et al., 1999 +

Elkhafaifi, 2005 +

Cheng et al., 1999 +

Vogly, 1998 +

In an exploratory study Cheng et al. (1999) found that writing apprehension was a language-skill-specific anxiety independent from other anxieties, and that low self-confidence was an important part of writing-specific language anxieties. Most studies on writing anxiety, however, were conducted in the participants’ L1. No studies to date have compared the spoken and written output of the same population in relation to the effect of input, processing and output anxiety (IPOA) and cognitive task complexity. The study reported in this dissertation attempted to further illuminate the consistency of anxiety linked to the processing stages of language production across modalities.