of naturalistic exposure to the L2. Aptitude is therefore a theoretical construct, operationalized in the form of a test, which aims to predict phenomena that characterize second language acquisition (SLA) (such as incidental learning, metalinguistic awareness, fossilization, and others), and the extent to which successful SLA occurs as a result. Although little was known about these SLA phenomena during the period when aptitude tests were first developed (the 1930s to the 1950s), recent attempts to conceptualize and measure aptitude are addressing the extent to which tests of aptitude predict them—including, for example, the extent of successful incidental L2 learning (Robinson, 2005a), metalinguistic awareness of the L2 (Roehr, 2018), and the influence of each on levels of ultimate L2 attainment (Granena & Long, 2013). Some of these issues are described below, following a discussion of early developed conceptualizations and measurement of aptitude for language learning.
The first tests of language‐learning aptitude were developed in the early part of the 20th century at about the same time as tests of general intelligence were developed. These tests were developed largely in response to institutional and educational concerns which continue to influence proposals for operationalizing and using aptitude tests today. One of these concerns was to select those people best able to learn languages. For example, in many countries military and diplomatic personnel have to learn other languages. So to select those best able to do this, institutions such as the Defense Language Institute in the USA developed their own aptitude tests (Petersen & Al‐Haik, 1976). Another reason for developing aptitude tests is to diagnose relative strengths and weaknesses in the abilities thought to contribute to language learning, so as to differentiate instruction and exposure to the L2 for individual learners in ways that optimally match their strengths in abilities, and compensate for areas of weakness (Robinson, 2007).
Perhaps the best known test of language learning aptitude, the Modern Language Aptitude Test (MLAT), was developed in the 1950s by John Carroll and Sidney Sapon (Carroll & Sapon, 1959). It is still used widely in SLA research into aptitude today, in its original and in translated versions. Using this test, in the 1960s Carroll (1962) showed that students who were successful at learning Spanish and other L2s in schools in the USA were also, largely, those who scored high on the MLAT. The positive correlation between scores on this measure of aptitude, and scores on achievement tests in these programs, was reported as between 0.4 and 0.65. John Carroll argued that the MLAT predicted the rate or speed of foreign‐language learning. He did not claim that those who scored low on aptitude tests could never reach high levels of ultimate attainment in the L2, but only that they would take longer to do this than those who scored high on his test.
Since the MLAT measure of aptitude is very similar to other, currently available tests that have been developed (e.g., Pimsleur, 1966; Petersen & Al‐Haik, 1976), and has been the measure most widely used, to date, in SLA research, it will be described here in some detail, as well as problems associated with it. The MLAT is a paper and pencil test, composed of five parts, and three of these are described below.
1 “Paired Associates” requires learners to memorize 24 foreign‐language words, which are presented with their English translations. This measures what Carroll called “rote memory.”
2 “Words in Sentences.” There are 45 items of the following type. Given a sentence such as Mary is cutting the APPLE, which of the following underlined words performs the same grammatical function: My brother John is hitting his dog with a big stick. And the answer is “dog,” the direct object. This measures what Carroll called “grammatical sensitivity.”
3 “Phonetic Script” measures whether the person taking the test can match a nonsense word that they hear to its written form on a page. This measures what Carroll called “phonetic sensitivity.”
Undoubtedly, being able to memorize foreign vocabulary, and understand grammar, and correctly identify sounds in a language you are studying are all very useful. If you are good at these you should do particularly well in L2 classrooms, because memorizing word meanings, understanding grammar, and correctly identifying sounds are often assessed in measures of achievement—particularly those adopted in the audiolingual programs of instruction that were prevalent in the 1960s when the MLAT was developed. However, there are two major problems with the MLAT measure of aptitude:
1 Learning a language involves different abilities at different stages of development. The MLAT and other current aptitude tests don't measure these.
2 Learning a language takes place in many different situations and classroom contexts. The MLAT and other current aptitude tests are insensitive to these.
These problems of developmental and situational insensitivity are problems of construct validity. Aptitude tests like the MLAT don't actually measure the full range of abilities that contribute to language learning over time, and across settings in which L2 exposure can occur. They measure only a small number of the abilities that are drawn on during SLA. For example, what happens when we begin to learn a language? Obviously, we must develop a basic vocabulary. We do this by listening to, understanding the meaning of, and remembering new L2 words. The paired associates subtest of the MLAT measures this ability. We must also begin to understand the order in which words occur in sentences of the L2, and to understand other aspects of grammar. The words in sentences subtest measures this ability too. And finally we must correctly identify new L2 sounds we hear, and the phonetic sensitivity subtest measures this as well.
But what happens after these beginning stages of language learning? We must then use words, in the correct order, while speaking to, and managing interaction with others in the L2. There are many abilities for speaking and interacting in an L2, but the MLAT doesn't measure any of them. It has no measure of the ability to produce a language in speech or writing. But these abilities are important to intermediate success and progress in L2 learning. And at more advanced stages of L2 learning, often outside the classroom, we must learn how to behave appropriately in the L2, and learn the appropriate formal and informal phrases and expressions needed at work, and when socializing. The MLAT doesn't measure the abilities for learning these. Based on these considerations Robinson (2005b) has argued that aptitude for language learning develops as the abilities drawn on in the initial stages of language learning reconfigure and are supplemented by abilities drawn on in later stages of learning. No current test of aptitude operationalizes this developmental perspective, though Robinson (2005b) has nominated clusters of abilities that should be important to aptitude for learning at beginning, intermediate, and advanced stages of L2 learning.
The second problem of situational insensitivity referred to above concerns the match between the abilities measured by an aptitude test, and the processing conditions under which learners are exposed to L2 input. For example, how do we learn new words or new grammar, or how to pronounce words in the L2? One way, of course, is to listen to a teacher explaining and demonstrating each of these, and then to remember and practice yourself. This is explicit learning: it is something you do intentionally and it requires effort and concentration. In contrast, children do not learn their first language (L1) in this way. Very young children have no explicit memory for the past. They have no long‐term memory, and very limited short‐term memory and attention spans. Consequently, children learn the patterns of the L1 implicitly and they are very sensitive to the frequencies with which forms co‐occur (Saffran, Aslin, & Newport, 1996). They learn these co‐occurring forms and patterns of language from exposure alone, and not as a result of explicit instruction.
There is no doubt that adults learn a lot of the L2 explicitly, by concentrating only on what they have to learn, and then later remembering it. But they also have the ability to learn the L2 incidentally, and this is closely related to child L1 implicit learning. Incidental learning is unintentional learning. You go to a birthday party, and next day you can remember names and faces of people you met there that you didn't know before. You read a book in the L2 and afterward you know the meaning of some new words, but you didn't intend
