I’d like to thank Dan Brown, who’s doing an MA in applied linguitics at Leicester university for help with this. Any references cited below can be found in Martínez-Fernández (2008).
As I mentioned in my last post, few scholars these days question that incidental learning from context (extensive reading) accounts for a substantial proportion of vocabulary acquisition. There is equally wide agreement that retention of information depends on the nature of the information processing involved. Many studies in the 1980s examined various types of rehearsal and their effect on vocabulary learning, all looking for support for the hypothesis that the more deeply words are processed, the more successfully they are retained. Running along side these rehearsal studies were studies that examined the question of whether the addition of some kind of lexical intervention in extensive reading programmes can further promote lexical development, without negating the principle aim of extensive reading which is to concentrate on “getting the message” and avoid studying the text. Studies investigating the effectiveness of different lexical intervention tasks during reading have led to conflicting results, and this is also true of the Involvement Load Hypothesis proposed by Laufer and Hulstijn (2001), which has been fairly rigorously tested and has received both supporting and challenging evidence. The hypothesis has important pedagogical implications, since it allows us to manipulate task features and predict what tasks will be more effective.
Background to “Incidental Learning”
One of the most important distinctions made in SLA is that between conscious and unconscious learning. Some scholars, trying to get a better, more precise fix on this conscious and unconscious dichtomomy, refer to “explicit and implicit learning”, or “intentional and incidental” learning. Perhaps the most famous attempt to build a theory on this distinction is Krashen’s Motitor Model, where the term “acquisition” is used for unconscious or implicit learning and the term “learning” is used for conscious or explicit learning. Krashen’s theory was, in fact, a response to the view that second languages were best taught by doing a contrastive analysis of the L1 and the target L2 and then teaching the differences by explicit instruction. Krashen said that the most important ingredient of SLA was “comprehensible input” and that grammar and vocabulary teaching had a very minor part to play in the process.
Following Krashen (chronologically, I mean), many SLA researchers took a processing approach which sees L2 learning as the process by which linguistic skills become automatic. In this view, initial learning requires controlled processes, which require explicit attention to the code, and time; with practice the linguistic skill requires less attention and becomes routinized, thus freeing up the controlled processes for application to new linguistic skills. But there was no agreed model. While McLaughlin (1990) likens the SLA process to driving a car with a clutch (where the proficient driver no longer needs to think about how to use the clutch), Bialystok (1994, 2001) likens the L2 learner to a library user. Bialystok sees the L2 learner’s knowledge as a mental library – the contents of the learner’s linguistic knowledge. The knowledge can be structured to different degrees, and this represents different degrees of control over the knowledge. The learner’s ability to retrieve a book represents his access to the linguistic knowledge he has. The learner’s knowledge can be more or less analysed; according to Bialystok the information is the same, but the more it is analysed, the more the learner is aware of the structure of the information. Bialystok therefore seems to be arguing the opposite of McLaughlin: more conscious control is necessary to speed access and processing.
The question of implicit and explicit knowledge, conscious and unconscious knowledge, acquisition and learning, is one that, in different ways, vexes many of those working on a theory of SLA. The problem is how to conceptualise this difference in such a way that the explanation it offers of the SLA process avoids the circularity of Krashen’s theory. In later work, McLaughlin suggests that since we can’t clearly define these terms in an empirically testable way, we should get rid of them, and use the terms “controlled processing” and “automatic processing” instead. Alas, these new terms suffer from much the same vagueness. Schmidt, rather than accept McLaughlin’s advice to abandon the search for a definition of “consciousness”, attempts to do away with its “terminological vagueness” by examining three senses of the term: consciousness as awareness, consciousness as intention, and consciousness as knowledge. Consciousness and awareness are often equated, but Schmidt distinguishes between three levels: Perception, Noticing and Understanding. As you probably know, the second level, Noticing, is the key to Schmidt’s eventual hypothesis (see here for more on all this).
So that’s the background to Laufer and Hulstijn’s (2001) attempt to work with “incidental vocabulary learning”. I’m not sure how helpful the background is, and I’m aware that I haven’t really worked through the question of how much the interest in incidental learning stems from the fact that it’s much easier to pin down the construct of incidental learning than intentional learning.
Laufer and Hulstijn (2001) chose a simple definition of incidental learning for their hypothesis (a variation of one used by many investigating vocabulary acquisition), namely: incidental learning occurs when new information is processed without the intention to commit the new information to memory. The construct is further operationalised by adding that learning remains incidental provided that students are under the assumption that their knowledge of the new vocabulary will not be specifically tested at a later date. This second provisio is intended to deal with the problem that if students know they will be subsequently tested on the new words involved in the study they will try to memorise them, thus invoking intentional learning.
The Involvement Load Hypothesis (ILH) states that the potential for learning and retaining unfamiliar words is dependent upon the degree of involvement in processing these words (Laufer and Hulstijn, 2001b:544). The degree of involvement can be measured by measuring 3 components: need, search and evaluation.
The need component is the motivational component; it relates to the need that a task imposes upon a learner. Need has 3 levels: absent, moderate and strong. Need is absent when a learner simply does not need to know a word; moderate when the need to know it is imposed externally (e.g. when a teacher asks a student to look up the meaning of an unfamiliar word or phrase) term; and strong when it is generated internally; for example, when a student decides to look up the meaning of a new word in a dictionary.
Search and evaluation are the two cognitive components of the ILH. Search relates to making an attempt to find the meaning of an unknown L2 word; for instance, searching for the meaning of an unknown L2 word on Google. The search component can be either present or absent. It is present when it is necessary to look up the meaning or translation of a word, for example by the use of a dictionary; and is absent when the meaning or translation is provided.
The third component, evaluation, entails the comparison of a new word with other words in order to assess its suitability for a given context. Evaluation can be absent, when there is no requirement to compare new terms with new contexts, moderate when comparisons have to be made with words for which the context is provided (for example, a gap filling activity), or can be strong when the evaluation requires a decision as to how additional words will fit with new words in self-made, original sentences.
According to the ILH, tasks with a higher involvement load are considered to be more effective for word learning and retention than tasks with lower involvement loads. For comparison purposes, each task is assigned a specific number which relates to an involvement load index. Total absence of a factor is assigned 0, a moderate presence is assigned 1 and strong presence is assigned a score of 2.
(0) Absent: Learner doesn’t need to understand or produce word.
(1) Moderate: Learner is required to learn the word by external source (teacher).
(2) Strong: Learner makes decision to learn or produce the word.
(0) Absent: Meaning or translation of word is provided.
(1) Present: Learner must look up meaning / translation of a word.
(0) Absent: Words are not compared with other words.
(1) Moderate: Words are compared to other words in provided contexts.
(2) Strong: Words are compared to other words in self-created contexts.
For example, if a teacher provides students with some new words and their definitions and asks students to create original sentences with them, the task would be assigned the following involvement load score:
Need: Moderate, (1): the assignment is imposed by the teacher.
Search: Absent (0): the definitions are provided.
Evaluation: (2) High: the students need to write their own original sentences.
Total Score: 3
In any study, the strength of each component is considered to be equal, allowing for the systematic scoring of tasks and for the control of variables.
The first direct test of the hypothesis was conducted by Laufer and Hulstijn (2001b). The study investigated the effects of Task Induced Involvement in a parallel experiment involving participants in two separate countries. The aim of the investigation was to test whether the success rate of vocabulary acquired was contingent upon task involvement load.
Three tasks of varying involvement loads were selected and each task was administered to different groups of students. The task with the lowest involvement load was a reading comprehension task with marginal gloss. In this task, ten target words whose understanding was relevant to the task were chosen. These words were then glossed into the margin. Participants were required to simply read the text and answer the comprehension questions.
In task two, participants were assigned the same text and same comprehension questions as the students in group one, except this time, the ten target words were deleted from the text. These ten words were then printed onto a separate sheet with their L1 translations, alongside five other random words that were not part of the text. In order to complete this task, participants had to read the text and fill in the blanks with the correct vocabulary words from the separate sheet.
In task three, participants were only provided with the target words and their L1 translations and were asked to create an original composition in the form of a formal letter. In each of the tasks, the involvement load variables of need and search were held constant. According to the index, need was held at moderate (1) because need was induced by the task, search was absent (0) as the vocabulary items were supplied. Holding these components constant meant that evaluation could be isolated as the only variable in the task. It varied as follows: group one, absent (0); group 2 moderate (1) because the context for the words was provided; group 3, high (2) because the words had to be used in original contexts.
Once the tasks were complete, receptive knowledge of the words was tested after one week and again after two weeks. For the test, participants were required to write L1 equivalents of the new vocabulary.
The results of the test found that retention of the new vocabulary directly correlated with involvement load. Participants who had completed tasks with the lowest involvement load scored lowest and those who had completed tasks with the highest involvement load scored highest. This provides evidence in support of the ILH.
The other study that empirically tested the ILH is a study by Keating (2008), who studied a group of beginning learners of Spanish. The participants in this investigation were first year university students studying Spanish as part of a compulsory general education requirement. The study was a replication of Laufer and Hulstijn (2001b), except for one aspect which I’ll ignore here. The results mirrored the Laufer Hulstijn (2001b) study: the lowest word retention rates came from tasks with the lowest involvement load and the highest word retention rates came from tasks with the highest involvement loads. The results clearly show that tasks with a higher involvement load enable learners to retain vocabulary more effectively.
So, there you are. This hypothesis suggests that classroom reading tasks, using a relatively-short text and various well-established activities pertaining to the text, can be manipulated in such a way as to induce a high involvement load, which in turn results in dramatic improvements in vocabulary acquisition. But nothing’s that simple, right? Right! In the next post, I’ll discuss the complicated bits.
Jan H. Hulstijn1 and Batia Laufer (2001) Some Empirical Evidence for the Involvement Load Hypothesis in Vocabulary Acquisition. Language Learning, Volume 51, Issue 3, pages 539–558.
Martínez-Fernández, A. 2008. Revisiting the Involvement Load Hypothesis: Awareness, Type of Task and Type of Item. In Selected Proceedings of the 2007 Second Language Research Forum, ed. Melissa Bowles, Rebecca Foote, Silvia Perpiñán, and Rakesh Bhatt, 210-228. Somerville, MA: Cascadilla Proceedings Project.