What predicts success in second language acquisition? Or: babies know statistics! Part B.

 Frost, R., Siegelman, N., Narkiss, A., & Afek, L. (2013). What predicts successful literacy acquisition in a second language?. Psychological science,24(7), 1243-1252.‏  http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3713085/

In the previous post I wrote about the hypothesis according to which second language acquisition is like any other learning process.  In any learning process we identify and perceive systematic and probabilistic structures in our environment.  Acquiring  second ( and first) language is mainly a process of acquiring and assimilating the statistical features of our linguistic environment.

Israeli researchers Frost, Siegelman, Narkiss  and  Afek, studied this hypothesis with American students at the overseas school of the Hebrew University in Jerusalem.  The students took part in the research during their first year of learning Hebrew as a second language.

Words in Hebrew are normally composed by intertwining a tri-consonantal root, which carries the core-meaning of the word, with word-pattern morphemes in which there are “open slots” for the root’s consonants to fit into.   The typical marker of reading Hebrew is sensitivity to the location of root consonants.   The writing system of Hebrew consists of letters that mostly represent consonants (thereby root information), while most of the vowels can optionally be superimposed on the consonants as diacritical marks (“points”). These points, however, are omitted from most adult reading material. Reading Hebrew fluently cannot be carried out by simply applying grapheme-phoneme conversion rules, but requires a deep understanding of the language. 

Reading Hebrew was assessed at the beginning and at the end of the school year by three tasks. The first monitored speed of decoding of pointed nonwords and reflected the assimilation of the characteristics of the Hebrew writing system. The second monitored accuracy in naming unpointed words, reflecting the implicit learning of Hebrew phonological word patterns. The third -- cross-modal morphological priming -- directly tapped the main marker of reading in Hebrew: the assimilation of the morphological root-based composition of words. 

The authors hypothesized that there will be individual differences in students'  ability to assimilate the structural features of Hebrew, and that these differences will be linked with the students' general ability to acquire and assimilate the statistical features of their environment.

  In order to study this hypothesis, the authored  employed a visual statistical learning (VSL) task  in which 24 relatively complex visual shapes   were presented in a consecutive stream that lasted about ten minutes. The 24 shapes were organized in eight triplets ("words"), which were presented in the 10-minute stream in random order.  Participants were not told that the stream was constructed of "words". However, following this phase, participants were presented with sets of 2 shape sequences :  one was a "word" that  appeared in the stream and the other was a "non-word" made of a sequence of three shapes that were taken from the stream, but did not appear in that order. Their success rate in distinguishing the "words" from the "non-words" reflected their implicit learning of the structure of the visual shapes within the stream. 

The results showed that participants who scored well in the VSL task, that is, picked up the implicit statistical structure embedded in the continuous stream of visual shapes, on average, scored well on the tasks that monitored the assimilation of the Semitic structure of Hebrew words! This suggests that a general non-linguistic faculty of statistical learning accounts, at least to some extent, for success in second language acquisition when the first and the second languages differ in their basic statistical properties. Such an outcome also implies that a simple and short test involving visual shapes could predict the speed of assimilating a new linguistic environment, even before the first foreign word has been learnt.

What predicts success in second language acquisition? Or: babies know statistics!

Apparently, there are significant individual differences in the  people's ability to learn a second language.  What causes them?  And what predicts fast and successful acquisition of a second language?

There are two approaches that try to explain these individual differences.  Probably the best explanation combines both approaches.

The first approach argues that the acquisition of a second language is dependent on the same linguistic abilities that allowed for the first language acquisition.  This approach is supported by studies showing that phonological awareness, syntactic knowledge, orthographic knowledge and vocabulary in the first language usually predict success in second language acquisition.  When a child immigrates with high cognitive academic language proficiency in this first language (a level which is usually achieved through reading), he usually acquires high cognitive academic proficiency in the second language  faster and better than a child immigrating with only basic communication skills in his first language.  According to this view, linguistic measures are the best predictors of success in second language acquisition.

The second approach argues that second language acquisition is like any other learning process.  In any learning process we identify and perceive systematic and probabilistic structures in our environment.  Acquiring  second ( and first) language is mainly a process of acquiring and assimilating the statistical features of our linguistic environment.

What does this mean?

Statistical learning refers to the cognitive process by which repeated patterns, or regularities, are extracted from the sensory environment. Such learning often happens without an intention to learn and without an awareness of what was learned.

How do babies acquire language, and how do they identify words in the stream of voice sounds they hear when someone talks to them?  Saffran and colleagues showed that 8-month old 2 infants are sensitive to auditory statistical regularities. They exposed infants to a stream of syllables, constructed from 12 syllables (e.g., tu, pi, ro, bi, da, ku, go, la, bu, pa, do, ti) that formed four tri-syllabic “words” (e.g., tupiro, bidaku, golabu, padoti). After a 2- minute exposure to streams such as “bidakupadotibidakugolabutupiro …”, infants were tested in a habituation procedure with two tri-syllables. One was a “word” (e.g., “bidaku”) heard during the 2 min-exposure phase, and the other was a foil (e.g., “tudabu”). The foils were composed of three syllables that were never paired together. Saffran and colleagues found that infants showed more interest in the foil than in the word, as indexed by increased listening time (e (i.e., the duration of showing interest in each tri-syllable). That is, infants are capable of statistical learning after just two minutes of exposure to sound sequences.

Oral  and written words in any language are built by rules that restrict and determine their internal structure (for example, the sound sequence "shchtz" is not possible in English, and "lpstzr" cannot be an English word).  Each language has its own statistical structure, and when people acquire a second language, they implicitly acquire, according to the second approach, a new set of statistical rules.

Thus, according to the second approach, the fundamental cognitive faculty of implicit correlation-learning which underlies any form of learning plays a primary role in second language acquisition.

The degree of similarity between the statistical characteristics of the first and second languages can affect the process of statistical learning of the second language structure.  Like any other cognitive ability, individual differences in sensitivity to correlations in the environment can affect second language acquisition.

Is it really so?  And how do we measure it?  In the next post.