Ch.7 Language Acquisition(語言習得)
Language Acquisition(語言習得) / Language Learning(語言學習)implicit, subconscious / explicit, conscious
informal situations / formal situations
uses grammatical 'feel' / uses grammatical rules
depends on attitude / depends on aptitude
stable order of acquisition / simple to complex order of learning
Theories in First Language Acquisition
1)The Behaviorist Learning Theory (popular in the 1950s and 60s)：
1. B.F. Skinner
2. Viewpoint: Language Learning is a kind of behavior similar to other human behavior.Language is learnt in much the same way as anything else is learnt.
stimulus→ organism → response
↓ ↓ ↓
languageinput the learner imitation
e.g. ‘This is a pencil.→‘This is a pencil.’
3.Four steps for a child to acquire his/her L1:
imitation→reinforcement→repetition→habituation
模仿强化重複成形
4.Criticisms of behaviorist learning theory
① overemphasize the external factors
② ignore the internal factor, i.e. the role of learner himself
③ overemphasize the role of imitation
2) The InnatenessHypothesis：This theory, also known as the nativist approachis representedby Chomsky, Mcneill and Lenneberg. Chomsky attacked behavioristic theory of language learning and reasserted the mentalist views of L1acquisition. Chomsky stressed the active contribution of the child and minimized the importance of imitation and reinforcement. Nativists strongly held that language acquisition is innately determined, that human beings are born with a build-in device of some kind that predisposes us to language acquisition, resulting in the construction of an internalized system of language. The child is born with the innate knowledge of language. This innate knowledge, according to Chomsky, is embodied in a “little black box” of sorts which Chomsky called language acquisition device or LAD. He assumes that the LAD probably consists of three elements---linguistic universals, a hypothesis making device, and an evaluation procedure. The so-called LAD has a number of linguistic universals, or universal grammar (UG) in store. It also has a hypothesis-making device, which is an unconscious process and enables the child to make hypotheses about the structure of language in general, and about the structure of language learning in particular. The hypotheses that the child subconsciously sets up are tested in its use of language, and continuously matched with the new linguistic input that the child obtains by listening to what is said in his immediate environment. This causes the child’s hypotheses about the structure of language to be changed and adapted regularly, through the evaluation procedure, and through a process of systematic changes towards the adult rule system.
3) Connectionism (連結論): A theory which views language as a complex system of units which become interconnected in the mind as they are encountered together. The more often units are heard or seen together, the more likely it is that the presence of one will lead to the activation of the other.
This approach to study language learning focuses on human brains and on construction of brain-like computer models to explain the process.
Connectionism is an alternative to both nativism and behaviorism and a subtle combination of the two. Grammar is seen by connectionists as the brain’s solution to the problem of exchanging a complicated set of meaning between people via a limited speech channel.
There are several variants of connectionism, going by names such as “emergentism突現主義,”“the Parallel Distributed Processing (PDP) Model 平行分散處理模式”and “neural networks類神經網路.”
4) Psycholinguistic Models
據說，一個德國傳說中的英雄(Baron Münchhausen)能透過，拉他自己的靴帶而把自己從沼澤拉出來。这个典故带来了bootstrapping（自启）这个词。這個典故帶來了bootstrapping（自啟）這個詞。在计算机科学里，这个词（通常被简化为booting）涉及到任何由一个简单的系统激活一个更复杂的系统的程在計算機科學裡，這個詞（通常被簡化為booting）涉及到任何由一個簡單的系統激活一個更複雜的系統的程序。在语言习得里，一个婴儿第一次解读成人的话语以及获得它们的意思和形式的过程也称为bootstrapping。在語言習得裏，一個嬰兒第一次解讀成人的話語以及獲得它們的意思和形式的過程也稱為bootstrapping。
In psycholinguistics, bootstrapping refers to the question of how language acquisition "gets started." A child gradually acquires a great deal of interlocking knowledge about the structure and vocabulary of his or her language. It has sometimes been proposed that some specific type of linguistic knowledge can be acquired early, and that this enables the child to analyze words or sentences well enough to acquire further knowledge from them. Metaphorically, this early knowledge would serve as bootstraps by which the child pulls himself or herself up.
Syntactic Bootstrapping：Using syntactic cues to learn verbs and
other relational terms.Syntactic bootstrapping is the idea that children use syntactic knowledge they have developed to help learn what words mean -- semantics builds on top of syntax.
一些研究人员发现，小孩在他们弄清楚某些词语的概念前就经常使用它们了。一些研究人員發現，小孩在他們弄清楚某些詞語的概念前就經常使用它們了。例如，当一个小孩模仿一些大人说Mom giggles（妈妈咯咯笑），他可能不知道giggle的意思。例如，當一個小孩模仿一些大人說Mom giggles（媽媽咯咯笑），他可能不知道giggle的意思。但是，因为在giggles后面没有任何名词短语，他可能知道咯咯笑是某种妈妈能够自己一个人做的事。但是，因為在giggles後面沒有任何名詞短語，他可能知道咯咯笑是某種媽媽能夠自己一個人做的事。如果他也学说Arnold smacks Gloria（Arnold掌掴Gloria），他能够解出，smacking包括了对某物或另外某个人的作用。如果他也學說Arnold smacks Gloria（Arnold掌摑Gloria），他能夠解出，smacking包括了對某物或另外某個人的作用。因此，小孩可能会用他们已经形成的句式的知识来帮助了解词语的意思。因此，小孩可能會用他們已經形成的句式的知識來幫助了解詞語的意思。这被Leila Gleitman定义为句法的自启，意思是语义构建在句法之上。這被Leila Gleitman定義為句法的自啟，意思是語義構建在句法之上。
Semantic Bootstrapping
Pinker (1984, 1989, 1994), following Grimshaw (1981) andMacnamara (1982) hassuggested “semantic bootstrapping” to account for children’s beginning acquisitionof syntax. Pinker postulates that children are born with linkings between somesemantic categories (e.g. agent of action) and syntactic categories (e.g. subject ofactive sentence). Learners then use real world/contextual information to identifythe semantic category and then “bootstrap” their way into the syntax.
Steven Pinker在不久之后提出了相反的观点——语义的自启：当小孩明白了一些词语的意思，他们能够通过观察成人所使用的词语在句子中的位置来发现这些词语的句法作用，也就是，句法构建在语义之SemS
Steven Pinker在不久之後提出了相反的觀點——語義的自啟：當小孩明白了一些詞語的意思，他們能夠通過觀察成人所使用的詞語在句子中的位置來發現這些詞語的句法作用，也就是，句法構建在語義之上。
Phonological Bootstrapping哪一个先出现，句法还是语义？is the hypothesis that a purely
phonological analysis of the speech signal may allow infants to
start acquiring the lexicon and syntax of their native language (Morgan & Demuth, 1996a).
哪一個先出現，句法還是語義？虽然现在还没有裁决，但是很明显的是，这个争论和小婴儿接触语言的第一阶段是没有关系的。雖然現在還沒有裁決，但是很明顯的是，這個爭論和小嬰兒接觸語言的第一階段是沒有關係的。一个新生的婴儿所注意到的既不是词语也不是句子结构，而是一串从成人口中发出的声音——一串可能带有停顿、元音加长、重音和某种语调模式的声音。一個新生的嬰兒所注意到的既不是詞語也不是句子結構，而是一串從成人口中發出的聲音——一串可能帶有停頓、母音加長、重音和某種語調模式的聲音。婴儿对于这样的音韵特征是相当敏感的。嬰兒對於這樣的音韻特徵是相當敏感的。如果一个婴儿熟悉了一个成人对milk（牛奶）发音的这些音韵特征，那么下一次大人说Want milk（想要牛奶）的时候，他可能会推断出这串声音由两部分组成——Want如果一個嬰兒熟悉了一個成人對milk（牛奶）發音的這些音韻特徵，那麼下一次大人說Want milk（想要牛奶）的時候，他可能會推斷出這串聲音由兩部分組成——Want和milk。这个推理标志着把一串连续不断的话语解剖为单个组成部分的开始，这个重大的转折点被称为音韵自启。這個推理標誌著把一串連續不斷的話語解剖為單個組成部分的開始，這個重大的轉折點被稱為音韻自啟。接着，婴儿可能会猜测到，一些组成部分与事物有关，其他则与动作或事件有关。接著，嬰兒可能會猜測到，一些組成部分與事物有關，其他則與動作或事件有關。这样，他们逐渐有能力分辨名词、动词和形容词以及在一个句子中分辨出话题和述题。這樣，他們逐漸有能力分辨名詞、動詞和形容詞以及在一個句子中分辨出話題和述題。如果那个婴儿有一只和他拥有同等机会接触到成年人的语言表现的宠物作伴，那么那只宠物，不管是猫还是狗，都不可能会“自启”，也就是，自发地划分一串话语声音并且从中辨认出不同的词语。如果那個嬰兒有一隻和他擁有同等機會接觸到成年人的語言表現的寵物作伴，那麼那隻寵物，不管是貓還是狗，都不可能會“自啟”，也就是，自發地劃分一串話語聲音並且從中辨認出不同的詞語。
5) The Sociolinguistic Models: These theories assume that language acquisition is a product of children’s early social interactions with the important people in their life. They point out that children communicate and interact socially with other people before they are able to produce language forms. They believe that language develops as a natural consequence of these interactions. That is, children’s attempts to communicate and socialize prompt their parents and other caregivers to provide the language appropriate for these exchanges. As children develop language, their communicative and social skills increase, allowing more mature and sophisticated interactions. These more mature interactions prompt more complex language forms from the parents, and the cycle continues until children’s language system and corresponding skills reach adult levels.
※What is parameter setting (參數設定)?
(Principles and Parameters Theory)原則與參數理論
Principles are generalizations which need not be learned, which all languages share, such as pronoun binding principle.
Parameters are specific points of variation between languages, such asNull-subject parameter.
Principles encode the invariant properties of languages, that is, the universal properties that make languages similar.
Parameters encode the properties that vary from one language to another; they can be thought of as switches that must be turned on or off.
- Chomsky 認為人生來就具有"語言習得機制 (Language Acquisition Device), 即LAD是人習得語言的先天因素，是語言習得的初始階段（initial stage 或State-0),其中數量有限的原則體系，規定具體語言習得的範圍。有了這種機制，人出生後通過與語言非常有限的接觸（即語言暴露），有關原則就可，以參數化成具體的語言規則，或由初始階段成長成具體語言，即使一種State-0→L的過程。這就是PPH假設(Principles-and-Parameters Hypothesis)。
In grammatical theory, the notion that children are born with grammatical parameters that are preset to certain values. Language acquisition is seen as a matter of resetting these parameters to the values of one’s native language.
In trying to explain the universality and rapidity of language acquisition, the notion of parameter setting offers one explanation to why we know how to generate grammatically correct language in numerous different cases without having
had to memorize them or even heard them before. Parameters correspond to certain general features of languages and have a wide effect on the grammatical climate. A classical example is the pro-drop or null subject parameter that determines whether it is correct to drop the subject in a sentence:
Example:
English: I love you (subject-verb-object)
Spanish: Te Quiero (object-(conjugated)verb)
(you love)
In Spanish, the verb is conjugated so that you know who loves just by looking at the ending (the -o in present), and this makes it perfectly sensible to omit the subject. So the null-subject parameter is set to on in Spanish, and off in English. (It would not be grammatically correct to say love you or you love.) Because grammatical rules in a
language interact so tightly, the null-subject parameter implies a subset(子集原則) of other grammatical features or rules:
Example:
English: Who do you think phoned?
Spanish: Quien piensas QUE llamaba?
(Who (do you) think THAT phoned?)
Adding "that" to the sentence is incorrect in English because the non-null-subject parameter does not allow dropping the subject, whereas it makes the sentence correct in Spanish. (I am not going to explain why, but this subset rule interacts with the null-subject parameter.)
There are also other parameters like the word order parameter that corresponds to how free the word-order is allowed to be: E.g. in German the verb usually comes last,
the Scandinavian languages have a less strict word order than English, and in the Aborigine language Warlpiri, word order within a clause is almost totally free (meaning there are no or few rules for the order of subject, verbs and objects).
The most important part of the parameter setting theory is to understand how the very nature of our languageacquisition offers a lot of short-cuts. When learning languages, we don’t need to know every grammatical rule because we assume that a language takes a certain subset of grammatical rules once we have the parameters sorted out: When I learned Spanish, it seemed natural to me to add the that to the sentence in the example (above), even though it is ungrammatical in my own language (which is not a pro-drop language). Thinking about the vast number of possible combinations of words and tenses, it seems impossible to acquire language by memorizing input alone. The parameter setting theory lends an explanation to this.
- 母體Population
- 是一個集合, 其中的元素是我們有興趣的物件或人。
- 樣本Sample
- 是母體的部份集合
- 通常樣本所包含的元素個數要較母體中的元素個數少得多。
母體→樣本→抽樣→推論
- 母體參數Population Parameter
- 「母體」中的元素個數很多, 而通常個體與個體之間是不一樣的. 例如, 養豬場有上萬頭豬, 每頭豬的體重都不相同, 面對這許多不同的體重, 也看不出個所以然來。
- 我們傾向利用某些總結歸納的數字來大概地描述這許多不同的體重，而這些總結歸納的數字即為母體參數。
- 母體參數是一常數(往往是未知數)
- 母體平均數population mean, 養豬場中所有豬的平均體重
- 母體變異數population variance, 用以反映豬隻體重差異程度
※欲推測2000年總統大選各組候選人的支持率？
母體：台灣地區的有選舉權的公民（大母體）
樣本：從台灣地區的有選舉權的公民抽取1000名進行研究分析（大樣本）
母體參數值（比例）：未知（開票後，連戰 23.10%，陳水扁39.30%、宋楚瑜36.84%)
樣本統計量（比例)：TVBS調查連戰 25%，陳水扁26%、宋楚瑜24%)
※Definition of Metalinguistic Awareness後設語言覺知
Metalinguistic awareness is a term used to describe a construct, theory or model to explain the interaction between language and written text, primarily in bilingual learners and literacy development (Bialystok, 2007.) There term was first used by Cazden (1974) to describe and explain the transfer of linguistic knowledge and skills across languages. Metalinguistic awareness (MA)is defined as an awareness or bringing into explicit consciousness of linguistic form and structure in order to consider how they relate to and produce the underlying the meaning of utterances. MA is also termed metalinguistic ability. The construct describes the ability to make language forms objective and explicit and to attend to them in and forthemselves. MA is the ability to view and analyze language as a thing,language as a processand language as a system.MA in bilingual learners is the ability to objectively function outside one language system and to objectify languages as rules, structures and functions. Code-switching and translation are examples of bilinguals of MA.
※Cazden, C. R. (1974). Play with language and metalinguistic awareness: One dimension of language experience. The Urban Review, 7, 28-29.
(＊）後設語言功能 (metalinguistic function)
後設語言一詞由維也納學派的邏輯學者率先使用，Rudolf Carnap（1891-1970）把語言區分為，日常說的語言和用來討論其他語言的語言（Jakobson & Halle, 1980）。Barthes （1992）在《符號學的基本元素》（Elements of Semiology ） 中 主 張 ， 後 設 語 言 (metalinguistics）一詞可以解釋為，在文化中，語言系統對其他符號系統的全面關係。因此，後設語言是經由「意指」作用的過程所掌握，其內容則是由一個「意指」系統所構成。例如，當符號研究者向作曲家詢問音樂作品的悲傷、強烈、愉快等指涉意涵時，在這種情形下，就必須依賴後設語言對於音樂的指涉予以分析。後設語言具有分析或解釋「符碼」的功能，例如，被丟棄的香菸盒可能會被視為垃圾，但如果把它黏在報紙上並鑲上畫框，懸掛在藝廊的牆上，它就成為藝品，這個畫框就具有後設語言的功能；因為它顯示了「應依據藝術的意涵來解碼這幅圖畫」，並引導我們去尋找美學上的關係和比例。在傳達的過程中，所有的訊息都直接或間接地具有後涉語言的功能，它們以特殊方式顯示出所屬的「符碼」體系，誠如 Jakobson所言：「現代邏輯對語言區別為兩個層面：一為對象語言，陳述一個對象；一為後設語言，陳述的是語言本身。」（ Jakobson, 1992:189）。因此，後設語言功能運用在語文裡，指的就是文章的評論，在視覺語言裡則是色彩、構圖等的分析；而在音樂的領域，它的功能不是作為情感的表現，而是對音樂「符碼」的解釋和分析；例如，節拍器，可以對樂曲的節奏作適當的控制，並解釋和分析音樂符碼使用的正確性。
※語言覺知 (language awareness)
語言覺知亦稱為後設語言覺知 (metalinguistic awareness)，乃指使用語言思考語言，把語言當作物品來表徵、思考，以及對語言結構、規則做有意識的覺察 (Owens, 1996)。在本語言理解處理歷程架構圖中，則指辨認輸入的語言訊息中的詞彙、成語；分析句子的組成成份，判斷語言訊息中的語法、語意、語用之適切性，也因此可在聽到別人所問的問題時，抑或是在聽到不清楚或不完整的訊息時，會使用內在表徵系統中所儲存或表徵的的語言知識來解釋該句話之意。
閱讀技能有各種重要的因素。「後設語言意識」（Metalinguistic awareness）就被視為特別重要的一環。後設語言意識包含四個部份。第一個部份是「句法意識」（syntactic awareness)，這牽涉到文法認知。第二部份是「語言應用意識」（pragmatic awareness)，這牽涉到如何適當、貼切的使用語言，第三部份是「文字意識」（word awareness），這部份牽涉到字彙的發展與恰當運用文字的認知。第四部份則是「音素意識」（phonemic awareness），這部份牽涉到孩子意識到口語是由各種聲音，亦即「音素」，所組成。
在這四者當中，研究發現，音素意識是閱讀的最佳指標。有些研究學者甚至認為音素意識比智商更能預測孩子的閱讀能力。研究發現，較早能把握好單字中不同音素的孩子通常比無法掌握音素的孩子閱讀能力好。有了這樣的認識，研究學者建議父母，讓孩子參與與聲音相關的遊戲活動。童詩對嬰幼兒就很有幫助。對幼童來說，強調一個字的起頭音與尾音，還有重複有趣的字以及相仿的發音，對於發展音素意識是很重要的活動。除了這些活動，唱歌對小孩子來說是很棒的活動，對再大一點的孩子，附有文字遊戲的生動圖書也很好。童詩、繞口令對大一點的孩子則是攸關聲音、押韻、韻律的好遊戲。
其他與閱讀有關的重要因素包括，讓孩子周遭充滿文字，讓孩子發展語言能力，增加字彙，還有充實孩子的基本知識。說到印刷文字，豐富的圖書，招牌、通知單，打油詩，報章雜誌，在在顯示出，閱讀在日常生活中不可或缺。就文字來說，找一些對語文運用有啟發性的讀物，有韻律感、押韻、文字不斷反覆出現的作品。透過這些材料，孩子會發展出對文字、用詞，以及文法的喜愛，而這些都會對發展他們的閱讀能力有所助益。而從知識的角度看，孩子對這個世界和對世界上形形色色的人知道得越多，他們就越有辦法閱讀觸及這個世界和世間人物的讀物。
Theories of Second Language Acquisition (第二語習得理論)
(1)The habit-formation theory (習慣形成理論)
The habit-formation theory comes from the behaviourist psychology and was very popular in the 1950s and 60s. According to behaviourists, learning a second language means the formation of a new set of linguistics habits. Imitation and practice play an important role in the process of habit-formation. According to the habit-formation theory, the old habit — mother tongue of the learner — will either facilitate or get in the way (有障礙) of the second language learning. Negative transfer means the learner transfers the ways of expression in the mother tongue to the target language. This will cause errors.
(2)The hypothesis of linguistic universals (語言普遍性理論)
It is believed that there exist certain linguistic properties which are true to all the natural languages in the world. The hypothesis of linguistic universals is born of the study of linguistic universals. Core grammar and peripheral grammar: Chomsky divides the grammar of a natural language into core grammar and peripheral grammar. According to him, human beings are born with a language acquisition device which consists of a set of general principles. The core grammars of nature language agree with the inborn set of general principles while the peripheral grammar can not be governed by the language acquisition device. The core grammar agrees with the inborn general principles and is much easier to learn.
(3)The acculturation theory (文化適應理論)
The meaning of the theory: By acculturation they mean that individuals of one culture have to go through the process of modification in attitudes, knowledge, and behaviour in order to function well in another culture. It involves social and psychological adaptations.
The relation between acculturation and second language acquisition: The degree of acculturation will control the degree of second language acquisition. Factors which determine the degree of acculturation success: The social and psychological distance plays a decisive role in acculturation success. The following negative psychological factors will increase the psychological distance: language shock, culture shock, low motivation and high ego boundaries.
(4)The discourse theory (話語交際理論)
The discourse theory was put forward by Hatch in the late 1970s. it was developed from Halliday’s theory of first language acquisition. The theory believes there is little difference between the first language acquisition process and the process of second language acquisition — only through communication discourses.
(5) The Input Hypothesis (輸入理論)
The input theory was put forward by Krashen in the late 1970s. The theory consists of the following five hypotheses:
1)The acquisition-learning hypothesis (習得與學習假說)
The theory claims that adult learners of a second language have two ways of developing their competence — acquisition and learning. The basic distinction between language acquisition and language learning is whether the learner pays a conscious attention to the rules of the target language. Acquisition refers to the subconscious process in which learners develop their language proficiency. Learning refers to the conscious process in which learners acquire the knowledge of rules of the target language.
2)The monitor hypothesis (自我監控假說)
Different functions — According to Krashen, acquisition is responsible for the fluency of the utterances produced by speakers while learning is responsible for the accuracy of the speeches or passages.
Three conditions — In order to perform this monitor function, language learners have to satisfy at least three conditions: sufficient time to monitor his production, to have his focus on form, and to have clear knowledge of the target language.
3)The natural order hypothesis(自然次序假說)
Same order — This hypothesis claims the that foreign language learners acquire the rules of the target language in the same order no matter where, when and how they are learning the language.
Speed — In Krashen’s point view, language teaching cannot change the natural order of language acquisition. It can only facilitate the speed of acquisition.
4)The input hypothesis (輸入假說)
Language input and language acquisition — According to Krashen, the only way for people to acquire a language is by understanding messages or receiving comprehensive input. They move from their current level to the next level by understanding input.
5)The affective filter hypothesis (情意濾器假說)
The affective filter hypothesis attempts to explain the variation in speed of language acquisition among individuals of the same group. The three affective factors determine the speed of success — motivation, self-confidence, and low anxiety. Influence of the three factors — learners with high motivation, self-confidence, and low anxiety will do much better than those that are unmotivated, lacking in self-confidence and concerned too much with failure. That is to say, learners with a low affective filter will get more input than learners with a high affective filter.
(6) The cognitive theory (認知理論)
*Cognitive psychologists regard learning as a cognitive process because they think it involves internal presentations which offer regulation and guidance for performance.
*Notions — In the cognitive theory, automaticity (自動加工) and restructuring(重新構建) are the most important notions because learners have to select appropriate vocabulary, grammatical rules and conventions governing language use.
*Modes — Cognitive psychologists think the process of language communication is a kind of information processing. When processing information, people use two ways which are called automatic and controlled modes. Teaching and practice will help the learner to acquire the automatic processing capacity.
*Different stages — The cognitive theory holds that language learning at the beginning stage involves more of the process of automaticity(自動加工) while at the advanced stage it involves more of the process of restructuring(重新構建).
1．自動加工
McLaughlin提出了語言學習是一個控制的、自動的加工過程的觀點。他認為，人的記憶是各種結點(node)的集合，每個結點都是一類信息的集。通過學習，這些結點相互聯繫而組成了一張記憶網。在長時記憶中，大多數的結點是靜止的，每當有相應的外部語言輸入時，記憶中的最近結點會被自動激活，而後擴散到相應的結點，這就是信息的自動加工。
2.重新構建
McLaughlin同時也認為，信息處理能力可以通過重新構建得到擴展。重構涉及知識在學習者大腦中的表徵及學習者對學習策略的運用。表徵的變化是指表徵從建立在例子上(exemplar-based)變為建立在規則上(rule-based)。而對學習策略的靈活運用也促進了重構。學生在習得複雜的技能時，往往設計一些新框架來儲存新知識，並將已有的舊知識也安排到這個框架中來。學習的過程是主動建構的過程，是對既有的知識體係不斷進行再創造、再加工以獲得新的意義、新的理解的過程。 McLaughlin的研究進一步表明，語言學習的佼佼者在重構語言規則時表現出更大的靈活性，他們更能避免某些語言錯誤。同時，練習對重構有重要的作用。
(7) Swain’s Output Hypothesis:
自80年代以來，二語習得研究領域開始對語言輸出在學習中的作用進行了探討。Swain (1985, 1995)提出的可理解輸出假設明確地闡明了二語學習者的語言輸出有助于促進他們流利和準確地使用語言。Swain (1985)指出僅僅依靠可理解輸入還不能使二語學習者準確而又流利地使用語言﹔成功的二語學習者既需要接觸大量的可理解輸入又需要產出可理解輸出。Swain (1995) 對這一假設進行了更為明確的論述，認為可理解輸出可以從三個方面促進二語習得,即它的三個功能。1）注意/觸發功能：語言輸出促使二語學習者意識到自己語言體系中的部分語言問題，進而可以觸發對現有語言知識的鞏固或獲得新的語言知識的認知過程。2）假設驗證功能：二語學習者可以把語言輸出視為是驗證自己在學習過程中形成的有關語言形式和語言結構新假設的途徑。3）元語言功能：二語學習者使用語言在語言交際中對語言形式進行協商。在這三個功能中，注意/觸發功能尤其重要。學習者通過輸出注意到自己語言體系中的問題，然後對語言形式進行有意識地分析﹔再產出修正後的輸出，提高語言輸出的準確性。
Three functions of output in second language learning:
1)the noticing/triggering function (注意/觸發功能)：The claim here is that while attempting to produce the target language (vocallyor silently), learners may notice that they donot know how to say (or write) precisely the meaning they wish to convey.
The noticing/triggering function occurs when learners are attempting toproduce the second language and they consciously notice that there are holesin their knowledge. That is, they do not know how to say what they want tosay.
In other words, under some circumstances, the activity ofproducing the target language may prompt secondlanguage learners to recognize consciously some of theirlinguistic problems: it may bring their attention tosomething they need to discover about their secondlanguage (possibly directing their attention to relevantinput).
2) the hypothesis-testing function (檢驗假設功能)：The claim here is that output may sometimes be, from the learner’s perspective, a “trialrun”(對新事物的）初步試驗，試行reflecting their hypothesisof how to say (or write) their intent.
Thisinvolves the learner trying out something and then maintaining or modifyingit on the basis of perceived success and feedback.
In other words, under some circumstances, the activity ofproducing the target language may prompt secondlanguage learners to recognize consciously some of theirlinguistic problems: it may bring their attention to something they need to discover about their secondlanguage (possibly directing their attention to relevant input).
3)the metalinguistic reflective function (後設語言反思功能)：The claim hereis that using language to reflect onlanguage produced by others or theself, mediates (促成) second languagelearning.
This idea originates withVygotsky’s sociocultural theoryof mind.Sociocultural theory is about people operating with mediatingtools (中介工具).Speaking is one such tool.