A Comprehensive Guide to Second Language Research Methods: Seliger And Shohamy 1989.pdf

# Seliger and Shohamy 1989: A Review of Second Language Research Methods ## Introduction Second language research is the scientific study of how people learn, use, and teach languages other than their native one. It is an interdisciplinary field that draws on theories and methods from linguistics, psychology, sociology, education, and other disciplines. Second language research aims to answer questions such as: - How do learners acquire different aspects of a second language (e.g., grammar, vocabulary, pronunciation, pragmatics)? - How do learners' cognitive, affective, social, and cultural factors influence their second language learning process and outcomes? - How do learners' first language and other languages affect their second language development? - How do different types of instruction, feedback, and assessment affect learners' second language performance and proficiency? - How do learners use their second language in various contexts (e.g., academic, professional, personal) and for various purposes (e.g., communication, learning, identity)? - How do learners' second language skills impact their cognitive abilities, intercultural competence, and social integration? To answer these questions, second language researchers need to use appropriate methods to collect, analyze, and interpret data from learners' linguistic behavior, mental processes, attitudes, beliefs, motivations, preferences, strategies, etc. However, choosing and applying suitable methods for second language research is not an easy task. There are many types of methods available, each with its own strengths and limitations. Moreover, there are many challenges and constraints that second language researchers face when conducting their studies, such as ethical issues, practical difficulties, theoretical controversies, etc. Therefore, it is important for second language researchers to have a good understanding of the different methods that they can use in their studies, as well as their advantages and disadvantages. This article aims to provide such an understanding by reviewing one of the most influential books on second language research methods: _Second Language Research Methods_ by Herbert W. Seliger and Elana Shohamy, published in 1989. This book is a comprehensive and accessible introduction to the main types of methods used in second language research, with examples from the authors' own studies. The book covers both experimental and non-experimental methods, and discusses their theoretical foundations, practical applications, and potential problems. The purpose of this article is to summarize and evaluate the main points and contributions of Seliger and Shohamy's book, as well as to highlight some of its limitations and criticisms. The article is organized as follows: The first section will focus on experimental methods, which are methods that involve manipulating one or more variables and measuring their effects on another variable. The second section will focus on non-experimental methods, which are methods that do not involve manipulating variables, but rather observing and describing natural phenomena. The third section will conclude the article by discussing the main findings and implications of Seliger and Shohamy's studies, as well as their strengths and weaknesses as researchers. ## Experimental Methods Experimental methods are one of the most common and powerful types of methods used in second language research. They are based on the assumption that there is a causal relationship between two or more variables, such that changing one variable (the independent variable) will result in a change in another variable (the dependent variable). For example, an experimental study might investigate whether using a certain type of feedback (the independent variable) will improve learners' grammatical accuracy (the dependent variable). The main advantage of experimental methods is that they allow researchers to establish causality between variables, by controlling for other factors that might influence the outcome. For instance, an experimental study might use random assignment to assign learners to different groups (e.g., feedback vs. no feedback), so that the groups are comparable in terms of their initial characteristics (e.g., age, proficiency level, motivation, etc.). This way, any difference in the outcome between the groups can be attributed to the effect of the independent variable, rather than to other confounding variables. The main disadvantage of experimental methods is that they are often artificial and unrealistic, as they involve creating situations that do not reflect the natural conditions of second language learning. For example, an experimental study might use a laboratory setting, a fixed time limit, a pre-determined task, or a standardized measure to test learners' performance, which might not correspond to how learners actually learn and use their second language in real life. Moreover, experimental methods are often limited in scope and generalizability, as they focus on a specific aspect of second language learning (e.g., a particular skill, domain, or feature), and on a specific population of learners (e.g., a certain age group, proficiency level, or background), which might not apply to other contexts or groups. Seliger and Shohamy illustrate the use of experimental methods in second language research by presenting two of their own studies: one on reading comprehension among Hebrew learners, and one on listening comprehension among Arabic learners. In both studies, they use two types of experimental designs: pretest-posttest design and factorial design. ### Pretest-posttest design A pretest-posttest design is a simple type of experimental design that involves measuring the dependent variable before and after applying the independent variable. For example, a pretest-posttest study might measure learners' reading comprehension before and after exposing them to different types of texts (e.g., narrative vs. expository). The main benefit of a pretest-posttest design is that it allows researchers to compare the change in the dependent variable within each group (e.g., how much did each group improve in reading comprehension), as well as between groups (e.g., which group improved more in reading comprehension). The main drawback of a pretest-posttest design is that it does not account for other factors that might affect the change in the dependent variable over time, such as maturation (e.g., learners might improve naturally due to their age or experience), history (e.g., learners might be influenced by external events or situations), testing (e.g., learners might perform differently due to repeated exposure to the same test), or regression (e.g., learners might score higher or lower than expected due to chance). Seliger and Shohamy use a pretest-posttest design in their study on reading comprehension among Hebrew learners. They aim to investigate whether exposing learners to different types of texts (narrative vs. expository) will affect their reading comprehension skills. They hypothesize that narrative texts will be easier to comprehend than expository texts, because narrative texts have more contextual cues, more familiar structures, and more emotional appeal than expository texts. after the treatment, using a standardized test that consists of multiple-choice questions based on four texts (two narrative and two expository). They analyze the results using a two-way analysis of variance (ANOVA), which allows them to examine the effects of text type (narrative vs. expository), group (narrative vs. expository), and their interaction on reading comprehension scores. They find that there is a significant main effect of text type, such that learners score higher on narrative texts than on expository texts, regardless of their group. They also find that there is a significant main effect of group, such that learners in the narrative group score higher than learners in the expository group, regardless of the text type. However, they do not find a significant interaction effect between text type and group, which means that the difference between groups is not dependent on the text type. They conclude that their hypothesis is partially supported, as narrative texts are indeed easier to comprehend than expository texts, but not to the extent that they expected. They suggest that this might be due to the fact that their learners were already advanced in their Hebrew proficiency, and therefore could cope with both types of texts relatively well. They also suggest that their results might have been influenced by other factors, such as learners' prior knowledge, interest, or motivation. They recommend that future studies should use more diverse and authentic texts, as well as more varied and sensitive measures of reading comprehension. ### Factorial design A factorial design is a more complex type of experimental design that involves manipulating two or more independent variables and measuring their effects on one or more dependent variables. For example, a factorial study might manipulate the type of feedback (e.g., explicit vs. implicit) and the mode of delivery (e.g., written vs. oral) and measure their effects on learners' grammatical accuracy and fluency. The main advantage of a factorial design is that it allows researchers to examine not only the main effects of each independent variable (e.g., how feedback type affects accuracy), but also the interaction effects between them (e.g., how feedback type and mode of delivery affect accuracy together). The main disadvantage of a factorial design is that it requires a large number of participants and conditions, which might be difficult to obtain or manage in second language research. Seliger and Shohamy use a factorial design in their study on listening comprehension among Arabic learners. They aim to investigate whether exposing learners to different types of input (monologue vs. dialogue) and different levels of difficulty (easy vs. hard) will affect their listening comprehension skills. They hypothesize that dialogue input will be easier to comprehend than monologue input, because dialogue input has more turn-taking cues, more redundancy, and more interactional features than monologue input. They also hypothesize that easy input will be easier to comprehend than hard input, because easy input has more familiar vocabulary, simpler syntax, and slower speech rate than hard input. They conduct their study with 80 adult Arabic learners who are enrolled in an intensive Arabic course at Tel Aviv University. They divide the learners into four groups: one group listens to monologue input only (the monologue group), one group listens to dialogue input only (the dialogue group), one group listens to both types of input alternately (the mixed group), and one group listens to no input at all (the control group). They also vary the level of difficulty of the input within each group: half of the input is easy and half is hard. They measure the learners' listening comprehension skills before and after the treatment, using a standardized test that consists of multiple-choice questions based on eight audio clips (four monologues and four dialogues). They analyze the results using a three-way ANOVA, which allows them to examine the effects of input type (monologue vs. dialogue), difficulty level (easy vs. hard), group (monologue vs. dialogue vs. mixed vs. control), and their interactions on listening comprehension scores. They find that there is a significant main effect of input type, such that learners score higher on dialogue input than on monologue input, regardless of their group or difficulty level. They also find that there is a significant main effect of difficulty level, such that learners score higher on easy input than on hard input, regardless of their group or input type. However, they do not find a significant main effect of group, which means that there is no difference between groups in terms of their overall listening comprehension scores. Moreover, they do not find any significant interaction effects between any of the independent variables, which means that the effects of each variable are independent of each other. They conclude that their hypotheses are partially supported, as dialogue input and easy input are indeed easier to comprehend than monologue input and hard input, but not to the extent that they expected. They suggest that this might be due to the fact that their learners were already advanced in their Arabic proficiency, and therefore could cope with both types of input and both levels of difficulty relatively well. They also suggest that their results might have been influenced by other factors, such as learners' prior knowledge, interest, or motivation. They recommend that future studies should use more diverse and authentic input, as well as more varied and sensitive measures of listening comprehension. ## Non-experimental Methods Non-experimental methods are another type of methods used in second language research. They are based on the assumption that there is a descriptive or correlational relationship between two or more variables, such that observing one variable (the predictor variable) will provide information about another variable (the outcome variable). For example, a non-experimental study might investigate whether learners' motivation (the predictor variable) is related to their second language proficiency (the outcome variable). The main advantage of non-experimental methods is that they are more natural and realistic, as they involve observing and describing phenomena as they occur in real life, without manipulating or controlling any variables. For example, a non-experimental study might use a natural setting, a flexible time frame, a self-selected task, or a subjective measure to collect data from learners' performance, perceptions, opinions, feelings, etc. Moreover, non-experimental methods are more comprehensive and generalizable, as they can cover a wide range of aspects of second language learning (e.g., cognitive, affective, social, cultural), and a large population of learners (e.g., different age groups, proficiency levels, backgrounds). The main disadvantage of non-experimental methods is that they do not allow researchers to establish causality between variables, as they cannot rule out other factors that might influence the outcome. For instance, a non-experimental study might find a positive correlation between learners' motivation and their second language proficiency, but it cannot determine whether motivation causes proficiency, proficiency causes motivation, or both are caused by a third variable (e.g., personality, aptitude, instruction). Furthermore, non-experimental methods are often subjective and unreliable, as they rely on learners' self-reports, judgments, or interpretations, which might be biased, inaccurate, or inconsistent. Seliger and Shohamy illustrate the use of non-experimental methods in second language research by presenting two of their own studies: one on writing development among English learners, and one on attitudes and motivation among French learners. In both studies, they use two types of non-experimental methods: case study and survey. ### Case study observations, tests, documents, artifacts, etc. A case study can have various purposes, such as describing, explaining, evaluating, or generating hypotheses about a case. The main benefit of a case study is that it allows researchers to gain a rich and detailed understanding of a case, by exploring its complexity, uniqueness, and context. A case study can also provide insights into phenomena that are rare, unusual, or inaccessible to other methods. The main drawback of a case study is that it is difficult to generalize from a case to other cases or populations, as a case might not be representative or typical of the phenomenon under investigation. Moreover, a case study is often subjective and biased, as it depends on the researcher's selection, interpretation, and presentation of the data. Seliger and Shohamy use a case study in their study on writing development among English learners. They aim to investigate how learners' writing skills change over time, and what factors influence their writing development. They hypothesize that learners' writing skills will improve as they receive more exposure to and practice in English writing, and that their writing development will be affected by their individual differences (e.g., age, proficiency level, motivation), their learning context (e.g., instruction, feedback, assessment), and their writing process (e.g., planning, drafting, revising). They conduct their study with four adult English learners who are enrolled in an intensive English course at Tel Aviv University. They collect data from the learners' written assignments over a period of 10 weeks, as well as from interviews with the learners and their teachers. They analyze the data using both quantitative and qualitative methods. They use quantitative methods to measure the learners' writing performance in terms of fluency (e.g., number of words), accuracy (e.g., number of errors), and complexity (e.g., number of clauses). They use qualitative methods to describe the learners' writing features in terms of content (e.g., topic, purpose, audience), organization (e.g., structure, coherence, cohesion), and style (e.g., tone, voice, register). They find that all four learners show improvement in their writing skills over time, but at different rates and levels. They also find that the learners' writing development is influenced by various factors, such as their motivation, feedback, task type, genre awareness, and revision strategies. They conclude that their hypothesis is partially supported, as learners' writing skills do improve with more exposure and practice in English writing, but not in a linear or predictable way. They suggest that writing development is a dynamic and individualized process that depends on multiple and interacting factors. They recommend that future studies should use longitudinal and comparative designs to examine writing development among larger and more diverse samples of learners. ### Survey A survey is another type of non-experimental method that involves collecting data from a large number of participants using standardized instruments such as questionnaires or interviews. A survey can measure various aspects of second language learning, such as attitudes, beliefs, motivations, preferences, strategies, etc. A survey can have various purposes, such as describing, comparing, correlating, or predicting phenomena related to second language learning. The main advantage of a survey is that it allows researchers to collect data from a large and representative sample of participants quickly and efficiently. A survey can also provide reliable and valid data if the instruments are well-designed and administered. The main disadvantage of a survey is that it does not allow researchers to explore the underlying reasons or mechanisms behind the phenomena measured by the instruments. A survey can also suffer from low response rates or inaccurate responses if the participants are not motivated or honest. Seliger and Shohamy use a survey in their study on attitudes and motivation among French learners. They aim to investigate how learners' attitudes towards French language and culture affect their motivation to learn French. They hypothesize that learners who have positive attitudes towards French language and culture will have higher motivation to learn French than learners who have negative attitudes. They conduct their study with 200 adult French learners who are enrolled in various French courses at Tel Aviv University. They collect data from the learners using a questionnaire that consists of two parts: The first part measures