研究：過度努力或將影響語言學習效果

Neuroscientists have long observed that learning a language presents a different set of opportunities and challenges for adults and children.

Adults easily grasp the vocabulary needed to navigate a grocery store or order food in a restaurant, but children have an innate ability to pick up on subtle nuances of language that often elude adults.

For example, within months of living in a foreign country, a young child may speak a second language like a native speaker.

Experts believe that brain structure plays an important role in this “sensitive period” for learning language, which is believed to end around adolescence.

The young brain is equipped with neural circuits that can analyze sounds and build a coherent set of rules for constructing words and sentences out of those sounds.

Once these language structures are established, it’s difficult to build another one for a new language.

In a new study, a team of neuroscientists and psychologists from Massachusetts Institute of Technology (MIT) discovered another factor that contributes to adults’ language difficulties: When learning certain elements of language, adults’ more highly developed cognitive skills actually get in the way.

The researchers discovered that the harder adults tried to learn an artificial language, the worse they were at deciphering the language’s morphology —the structure and deployment of linguistic units such as root words, suffixes, and prefixes.

“We found that effort helps you in most situations, for things like figuring out what the units of language that you need to know are, and basic ordering of elements. But when trying to learn morphology, at least in this artificial language we created, it’s actually worse when you try,” said Amy Flynn a postdoc at MIT’s McGovern Institute for Brain Research.

Finn and colleagues from the University of California at Santa Barbara, Stanford University, and the University of British Columbia describe their findings in journalPLOS ONE.

Linguists have known for decades that children are skilled at absorbing certain tricky elements of language, such as irregular past participles (examples of which, in English, include “gone” and “been”) or complicated verb tenses like the subjunctive.

“Children will ultimately perform better than adults in terms of their command of the grammar and the structural components of language —some of the more idiosyncratic, difficult-to-articulate aspects of language that even most native speakers don’t have conscious awareness of,” Finn says.

In 1990, linguist Elissa Newport hypothesized that adults have trouble learning those nuances because they try to analyze too much information at once. Adults have a much more highly developed prefrontal cortex than children, and they tend to throw all of that brainpower at learning a second language.

This high-powered processing may actually interfere with certain elements of learning language.

“It’s an idea that’s been around for a long time, but there hasn’t been any data that experimentally show that it’s true,” Finn says.

Finn and her colleagues designed an experiment to test whether exerting more effort would help or hinder success.

The Study

First, they created nine nonsense words, each with two syllables. Each word fell into one of three categories (A, B, and C), defined by the order of consonant and vowel sounds.

Study subjects listened to the artificial language for about 10 minutes. One group of subjects was told not to overanalyze what they heard, but not to tune it out either.

To help them not overthink the language, they were given the option of completing a puzzle or coloring while they listened. The other group was told to try to identify the words they were hearing.

Each group heard the same recording, which was a series of three-word sequences —first a word from category A, then one from category B, then category C —with no pauses between words.

Previous studies have shown that adults, babies, and even monkeys can parse this kind of information into word units, a task known as word segmentation.

Subjects from both groups were successful at word segmentation, although the group that tried harder performed a little better. Both groups also performed well in a task called word ordering, which required subjects to choose between a correct word sequence (ABC) and an incorrect sequence (such as ACB) of words they had previously heard.

The final test measured skill in identifying the language’s morphology.

The researchers played a three-word sequence that included a word the subjects had not heard before, but which fit into one of the three categories.

When asked to judge whether this new word was in the correct location, the subjects who had been asked to pay closer attention to the original word stream performed much worse than those who had listened more passively.

The findings support a theory of language acquisition that suggests that some parts of language are learned through procedural memory, while others are learned through declarative memory.

Under this theory, declarative memory, which stores knowledge and facts, would be more useful for learning vocabulary and certain rules of grammar.

Procedural memory, which guides tasks we perform without conscious awareness of how we learned them, would be more useful for learning subtle rules related to language morphology.

“It’s likely to be the procedural memory system that’s really important for learning these difficult morphological aspects of language.

“In fact, when you use the declarative memory system, it doesn’t help you, it harms you,” Finn says.

Still unresolved is the question of whether adults can overcome this language-learning obstacle. Finn says she does not have a good answer yet but she is now testing the effects of “turning off” the adult prefrontal cortex using a technique called transcranial magnetic stimulation.

Other interventions she plans to study include distracting the prefrontal cortex by forcing it to perform other tasks while language is heard, and treating subjects with drugs that impair activity in that brain region.長期以來，神經學家一直認爲語言的學習對成年人和孩子來說有着不同的機遇與挑戰。

成年人很容易地掌握雜貨店或餐館必備詞彙，孩子就有與生俱來能力，辨別語言中細微的差別，而這些差別經常會把成年人難住。

比方說，孩子只要在國外生活幾個月，就可以像外國人一樣說外語。

一些專家認爲大腦結構會影響人們語言學習的敏感期，而敏感期止於青春期。

孩子的大腦含有神經迴路，能夠分析聲音，建立起一套連貫的規則，從而利用這些聲音組詞造句。

這些語言結構建立完成，大腦就很難形成新的結構，學習新一門語言。

在一項研究中，由來自麻省理工學院（MIT）神經學家及心理學家組成的研究團隊發現另一種因素，可以解釋成年人學習語言的難點：在學習語言的某一方面時，成年人高度發達的認知技能實際上阻礙了語言的學習。

研究人員發現，成年人在學習人造語言的過程中越努力，那麼他們在學習語言形態學方面表現就越糟糕。（形態學指的是詞根、後綴及前綴等語言單位的結構及組合。）

“研究發現，多數情況下，努力有所作用，比如，你想要解出你需要知道的是那種語言單位及基礎的成分順序安排。但是，當你想要學習形態學的時候，學習我們自己創造的語言時，越努力，效果越差。”麻省理工學院麥戈文人腦研究所的博士後艾米·弗林（Amy Flynn）表示。

來自加州聖巴巴拉大學、斯坦福大學及英屬哥倫比亞大學的芬恩（Finn）及同事在《公共科學圖書館·綜合》雜誌上發表了他們的研究成果。

幾十年來，語言學家都知道孩子善於消化吸收語言學習中的難點問題，像是不規則過去分詞（英語中的例子，比如“gone”和“been”）或是虛擬語氣中複雜的動詞時態。

“對於語法及句式結構，孩子總會比成年人掌握地更好，比如，即便是說母語的人大多也沒有意識到的一些比較特殊、發音困難的問題等。”芬恩說。

1990年，語言學家愛麗莎·紐波特（Elissa Newport）提出一個假設，成年人理解這類細微差別有困難是因爲他們試圖同時分析過量的信息。成年人的前額皮質比起孩子，更加高度發達，他們往往會絞盡腦汁去學習第二門語言。

事實上，這個高強度的學習過程也許會干擾某些語言學習因素。

“這個想法已存在多時，但是目前爲止，沒有任何實驗數據證明它的正確性。”芬恩說。

芬恩及同事設計了一個實驗，測試付出更多的心力是否有助於或有礙於語言學習。

研究過程

首先，他們創造了九個毫無意義的詞，每個詞兩個音節。再根據輔音及原因的排列，將這些詞分爲三類（A、B及C類）。

研究對象聽這些人造語言大約10分鐘。其中一組被告知不要過度分析他們聽到的內容，也不要置若罔聞。

爲了不讓他們過度分析，他們可以選擇在聽的過程中完成拼圖或給圖上色。另外一組被告知要儘量辨析他們聽到的內容。

每個小組聽的都是同樣的內容，三個單詞爲一個單位，第一個爲A類詞，第二個B類，接着C類，詞與詞中間沒有停頓，

既往研究顯示成年人、孩子，甚至是猴子都能理解這類單詞組信息，亦被稱爲“詞切分”。

兩組實驗對象都能成功地完成詞切分，雖然認真分析的那組表現更好一點。同樣地，在詞彙順序排列的任務中，兩組也都完成得很好。第二個任務要求實驗對象在聽過的正確次序（ABC）與錯誤詞序（ACB）中作出選擇。

最終測試考量他們在辨別語言形態方面的技能。

研究人員播放一連串三詞組和，其中包括他們沒有聽過的詞序，但也符合分類要求。

當爲要求對新詞詞序作出判斷時，認真分析原詞語流的那組實驗對象表現比消極地聽的那組差得多。

研究結果與語言習得的一個理論相符，那就是，語言的某些部分是通過程序記憶學習來的，然而其他的一些是由陳述性記憶習得的。

在該理論體系下，陳述性記憶，用於儲存知識與事實，這種記憶在詞彙與某些語法規則的學習中作用更明顯。

程序性記憶，指導我們在不瞭解學習方式的情況下完成任務，這種記憶在語言形態相關的微妙規則學習中作用更明顯。

“在較有難度的語言形態學方面的學習中，程序性以及系統可能發揮着更大的作用。”

“事實上，使用陳述性記憶體系，並不起作用，反而是幫倒忙。”芬恩說。

令人未解的仍是成年人是否能夠克服語言學習的障礙。芬恩說她也沒有一個合理的答案，但是現在她正使用叫做經顱磁刺激的技術，測試“關閉”成年人前額皮質的作用。

她還計劃研究其他方面的干擾，比如在聽的過程中，通過強制性地完成任務，干擾前額皮質，使用藥物影響大腦特定區域的活動等。