作者：郑丽娜
出版社：上海交通大学出版社
格式: AZW3, DOCX, EPUB, MOBI, PDF, TXT
中国学习者对英语规则形式和不规则形式的加工研究(英文版)试读：
前言
在人类的历史长河中，科学家们不断致力于对语言在大脑中运作机制的探索。近年来，计算机科学的发展和先进实验工具的出现大大促进了这个课题的研究。在此背景下，出现了大量有关语言加工的研究，其中对规则形式和不规则形式的加工更是成为过去30年间心理语言学、认知科学、神经科学、脑科学和语言习得研究的一个热点，其探讨的中心问题是：规则形式与不规则形式是由单一机制还是由两个机制存储和提取的？人类对语言的运用主要依靠心理词库和心理语法。前者提供了交际所需的词汇，而后者提供了将词汇组成更大语言单位所需的规则。不规则形式与心理词库相似，规则形式与心理语法相似。因此，通过考察语言中的规则形式和不规则形式可以探索人类语言在大脑中的运作机制。
迄今，规则形式与不规则形式的加工研究多集中在对英语动词过去时态形式的考察上。研究领域包括儿童母语习得、成人母语加工、母语语言障碍，近些年又延伸到二语习得领域。该领域研究从一开始就形成了单机制和双机制两大理论阵营，前者认为规则形式与不规则形式都由联想记忆系统单一机制存储与提取，其主要代表是联通主义模型，后者认为规则形式由规则系统实时加工，而不规则形式由记忆系统存储与提取，其主要代表是词汇规则模型和陈述/程序模型。
本书将规则形式与不规则形式的加工扩展到中国英语学习者，具体探讨：①词频、音系相似度和规则度三个语言因素对规则形式与不规则形式加工的影响; ②英语水平、学习英语的起始年龄和学习英语的年限三个学习者因素对规则形式与不规则形式加工的影响; ③中国英语学习者对英语规则形式与不规则形式的加工是通过单一机制还是双机制完成。为探讨以上问题，本研究设计并实施了三个实验任务：英语动词过去时态形式的产出任务、词汇判断任务和遮蔽启动任务，并对实验数据进行了统计分析。结果显示，词频和音系相似度在中国学习者加工英语规则形式和不规则形式时影响显著，而规则度的作用仅限于对规则形式与不规则形式的区分，但对进一步区分三类不规则形式的作用不显著。英语水平、学习英语的起始年龄和学习英语的年限对中国学习者加工英语规则形式和不规则形式的影响不显著。研究还发现，规则与不规则真实词都表现出了词频效应，只有不规则假词表现出音系相似性效应，规则假词没有表现出音系相似性效应。另外，中国学习者对英语不规则真实词和高频规则真实词的存储和提取都是通过记忆系统完成的，只有低频规则真实词是通过规则分解加工的。而且，在产出任务中当不规则过去时态形式的提取失败时，中国英语学习者就会使用规则形式作为缺省处理，出现过度规则化现象。以上结果表明，虽然中国学习者对英语规则形式和不规则形式的加工更多地依赖记忆，但依然通过记忆系统和规则系统两个机制加工规则形式和不规则形式，这与双机制理论是吻合的，与Pinker（1999）“人类使用记忆系统和规则系统两个机制加工语言具有普遍性”的主张也是一致的。最后，本书对本研究的结果与以往研究进行比较，并对研究结果进行了讨论，还分析了双机制理论的优越性和目前理论模型存在的问题，指出本研究的不足和今后研究的方向。
本书是根据笔者2012年在上海交通大学完成的博士学位论文修改而成的，补充了近五年来与本课题相关的最新研究成果。适合语言学、心理学、认知科学、计算机科学、脑科学等专业的研究生以及从事第二语言教学的教育工作者，尤其是从事语言加工和二语习得的研究者参考阅读。
在本书即将付梓之际，我要感谢我的博士导师陈永捷教授和我的硕士导师俞理明教授长期以来对我的指导、关心和帮助，是他们把我引入了二语习得与加工研究领域，并培养了我治学精神和研究之道。还要感谢张北镇博士在实验设计过程中给予的帮助和指导，杨小虎博士在数据处理过程中给予的帮助，以及常辉教授对本书的校对和修改建议。本书的出版得到了上海市教育委员会科研创新项目：“中国学习者英语加工与认知方式研究”(编号：14YS109)的资助，在此表示感谢。
当然，作为一项探索性研究，本书可能还存在一些不完善的地方，恳切期望国内外同行专家和广大读者惠予批评指正。郑丽娜2017年5月Chapter 1Introduction1.1Research Background
In human history, scientists have been dedicating themselves to exploring how human languages operate in our brains. The development of computer science and the appearance of sophisticated psychological and neurological experimental instruments in recent years have enabled linguists and psychologists to address this issue. As a result, the last several decades have witnessed an unprecedented number of empirical studies of the human language processing, representation, acquisition, use and disorders, allowing this centuries-old mystery to be extensively investigated and better explained (Pinker, 1991: 530). One area which has received increasing attention in the past thirty years for psycholinguistics, cognitive science, neuroscience, brain science and language acquisition is how regular and irregular forms are processed in the human brain. Its central topic is whether there is a dissociation of processing between regular and irregular forms.
The issue of the processing of regular and irregular forms is important for the nature of language representation and the operating mechanism of human languages in that two main capacities characterizing the use of the human language, namely mental grammar and mental lexicon, can be reflected by the use of regular and irregular forms (Ullman, 2001a, 2001b, 2004). Thus, regular and irregular forms serve as an arena for the test of the dual structure of the human language faculty (Clahsen et al., 2010).
The mental grammar is a system of productive, combinatorial grammatical rules that assemble morphemes and simple words into sequential and hierarchical complex words, phrases or sentences. The psychological mechanism to handle the mental grammar is the symbolic computation. The rules in the mental grammar constrain how lexical forms combine into complex representations, and contribute to interpreting the meanings of complex representations even if they have never been heard or seen before. This grammatical capacity to derive meaning from well-formed complex structures underlies the incredible productivity and creativity of human languages. And regular forms are generated by rules that add suffixes to the stem, thus combinatorial, productive and predictable. Their generation is just like the computation of phrases and sentences in mental grammar. Regulars in a language are open-ended and new members can be easily added.
The mental lexicon, on the other hand, is a repository of stored information and contains thousands of arbitrary sound-meaning pairings underlying the morphemes like words of a language, or idiomatic phrases whose meanings cannot be derived transparently from their components. The psychological mechanism to handle the mental lexicon is memory. And irregular forms are often generated in an unpredictable way with grammatical features incorporated into their lexical entries. For a given irregular stem, for example, language learners cannot predict what form its past tense might take. Thus, irregulars have to be learnt on a case-by-case basis and memorized individually, just like individual words in the mental lexicon. Irregulars of a language are generally unproductive, and a closed list in that very few new members can be added.
The research on the processing of regular and irregular forms, therefore, may shed light on the interaction between the mental lexicon and the mental grammar, psychological status of grammatical rules and the implementation of the components of language in the brain.
According to Pinker and Ullman (2002: 456), the debate on the processing of regular and irregular forms began with a report of the connectionist model by Rumelhart and McClelland (1986) and a critique by Pinker and Prince (1988). Since then a number of different approaches and models have been put forward. At first, traditional grammar assumes a clear dichotomy between regular forms and irregular forms by proposing the regulars are formed by concatenating a suffix to the stem while the irregulars are unpredictable and must be individually memorized. Later, the single-mechanism approach (e.g., Chomsky & Halle, 1968; Halle & Mohanan, 1985; McClelland & Patterson, 2002; Rumelhart & McClelland, 1986) postulates that both regular and irregular forms are processed in a single route of either a rules system or an associative memory, whereas the dual-mechanism approach (e.g., Pinker, 1991, 1998, 1999, 2006; Pinker & Ullman, 2002; Ullman, 2001a, 2001b, 2004; Ullman et al, 1997) proposes that regular forms are processed by a rules system or a procedural memory while irregular forms are processed by an associative memory or a declarative memory. More recently, the dual-route model (e.g., Baayen, Dijkstra, & Schreuder, 1997), which is a hybrid model combining computation and storage of inflected words, assumes that both decomposition and whole-word access are activated simultaneously, and the faster route wins the competition. Finally, there are also some other proposals that combine rule-based processing with the use of linguistic probabilities (e.g., Yang, 2002; Baayen, 2003; Bod, Hay & Jannedy, 2003; Gor, 2003, 2010).
Additionally, a variety of paradigms and tasks have been employed to probe the issue of the processing of regular and irregular forms, including acceptability ratings, familiarity ratings, speeded production tasks, visual or cross-modal lexical decisiontasks, priming tasks, computer simulation, eye-tracking, event-related potentials (ERPs), functional magnetic resonance imaging (fMRI), positron emission tomography (PET) as well as Magnetoencephalography (MEG). V arious research fields have been explored, including child first language acquisition, adult first language processing, first language disorders and more recently second language (L2) acquisition, even third language acquisition. In the research, both production and perception data have been collected and analyzed. Until now, however, the debate is far from over, and the processing of regular and irregular forms is still one of the most contentious issues in cognitivescience (Pinker & Ullman, 2002).1.2Regular and Irregular Forms in English
English inflections are relatively impoverished and the distinction between regular and irregular inflections only extensively exists for the past tense forms of verbs and plural forms of nouns. However, the generation procedures of regular and irregular forms, especially the past tense forms, offer an unusually sharp contrast between a highly regular procedure and a highly irregular and idiosyncratic set of exceptions.
Regular English past tense verbal forms are formed by the rule of adding the past tense suffix -ed to an unchanged stem verb, with three different phonologically determined forms[t], [d]and[id]depending on the final segment of the stem. This is an apparently paradigmatic example of a rule-based process. By contrast, irregular English past tense verbal forms are generally idiosyncratic and phonologically unpredictable, although they show a variety of irregularity within a continuum. Some undergo none overt changes (e.g.,rid-rid, hit-hit). That is, their past tense forms are identical to the stem verbs. Interestingly, all the English verbs whose past tense forms undergo none overt changes end with t or d. Some are generated only by changing the final consonant into[t]or[d] (e.g.,bend-bent, make-made). Also some undergo a stem vowel change but end in t or d (e.g.,sleep-slept, feel-felt). About half of the irregular English verbs end in t or d, because they originally took some version of the regular -ed suffix. Many more are formed by a stem vowel change (e.g.,sing-sang, wear-wore). Finally, there are some suppletive past tense verbal forms with highest degree of irregularity (e.g.,is-was, go-went), which involve a completely arbitrary relation between the stem and its past tense form.
Many past tense forms of irregular verbs, however, do not show total arbitrariness in that the phonological content of their stem is largely preserved, and some patterns can be easily recognized with similarities among clusters of irregular verbs in their stems and past tense forms, such as feed-fed, speed-sped, bleed-bled, breed-bred; keep-kept, weep-wept, sleep-slept, sweep-swept. Therefore, many irregulars are not just a set of arbitrary, individually memorized by rote memory and idiosyncratic items, but quasi-regular and cannot be simply attributed to a lexicon of stored exceptions. But on the other hand, nearly all the family members of verbs with phonological similarity have exceptions. For example, the past tense forms of bring and sing are not brung and sung, although the past tense forms of cling, fling, sling, sting, wring and swing are clung, flung, slung, stung, wrung and swung. Also, the past tense forms of drink, sink and shrink are drank, sank and shrank, but the past tense forms of think and slink are thought and slunk. Even trickier, blink and wink are regular verbs and their past tense forms are blinked and winked.
In a very similar way, regular English noun plurals are formed by adding the suffix -s to a countable noun, with three allophones[s], [z]and[iz], whereas there are several classes of irregular nouns with different levels of regularity. One class has a zero plural, for example, sheep and deer. That is, their singular and plural forms are identical. Another class involves the voicing of the final consonant, such as knife-knives, wolf-wolves. The third class is formed by adding the Old English noun plural suffix -en, such as child-children, ox-oxen. The rest native irregular English noun plurals involve a stem vowel change, like mouse-mice, louse-lice, foot-feet, tooth-teeth. Finally, the plurals of the nouns borrowed from Latin and Greek are formed generally by changing the word endings, for example phenomenon-phenomena, criterion-criteria, stimulus-stimuli. Again, many irregular nouns show the phonological similarity effect and form some patterns. For instance, the pairs of foot-feet, tooth-teeth and goose-geese as well as knife-knives, life-lives, wife-wives undergo the same stem vowel change to form the plurals.
Another area of regular and irregular nouns concerns regularity within noun-noun compounds. According to Pinker (1998: 231) and Pinker and Ullman (2002: 459), the first noun is generally a modifier and the second noun is the head, and the new compounds inherit the properties of the rightmost morphemes, which is sometimes called the right-hand head rule. According to Gordon (1985) and Pinker (1999: 207), the first noun of English noun-noun compounds must be singular if it is regular and countable, but it could be plural if it is irregular and countable. For instance, rats eater and standards setter are not acceptable, while mice eater and criteria setter are far more acceptable.
Suffixation of regular inflections is highly productive in the sense that generally there are far more regular words than irregular words in languages, and that the suffixation applies predictably to thousands of words and is generalized to almost all the new words in the language as the default treatment. For example, there are thousands of existing regular English verbs and hundreds of new ones are added all the time, but the number of irregular verbs is only about 180 (Pinker, 1998: 222). Additionally, when a new word like clone is coined, it will be treated as a regular verb and the past tense suffix -ed is added to it to form its past tense form. Likewise, when a foreign word comes into English, it will generally receive the default treatment, even if it has a similar pattern to the existing irregular English words. For example, the plural forms of chief and gulf, which were borrowed from French, are not chieves or gulves, but chiefs and gulfs; the past tense form of deride, which was borrowed from Latin, is not derode, but derided. Sometimes, there is a transitional period of regularization for the irregular foreign words. For instance, when the irregular Latin noun focus was borrowed into English, it kept its original plural form foci. Later it underwent regularization in English as focuses, and now both foci and focuses are acceptable. It can be predicted that the regular form focuses will win over the irregular form foci in the future. Furthermore, the default treatment is also applied to nearly all the novel verbs. For instance, Berko (1958) found that even preschool children could create the past tense forms for the invented words like rick by adding the regular suffix -ed. Moreover, these patterns are even occasionally generalized by human speakers. For example, Bybee and Slobin (1982) found that children occasionally produced novel past tense forms like brang for bring based on sing and spring. According to Xu and Pinker (1995), although these errors are rare, all children make them. And Bybee and Moder (1983) also found that most college students offered splang or splung when they were given the novel verb spling and asked to guess its past tense form.
Irregular forms have a distinctive feature of high frequency. That is, most of the irregulars are among the most common words and used at high frequencies. According to Francis and Kucera (1982), the 13 most frequently used verbs in English- be, have, do, say, make, go, take, come, see, get, know, give, find -are all irregular. And according to Marcus et al. (1995), about 14% out of the 1, 000 most frequent verbs are irregular. Based on Bybee (1985), Old English had about twice as many irregular verbs as Modern English like cleave-clove, abide-abode. These have been regularized over the centuries because of their low frequencies, unless they fall into certain patterns whose family members may strengthen memory traces for each other. One explanation is that if irregular forms are not used frequently, children would not hear these verbs often enough and they might disappear from the language (Marcus et al.,1992). Another explanation from a diachronic perspective for this is that people cannot remember them firmly, and consequently they will be gradually treated as a regular form, the so-called default treatment, and become regular forms (Pinker, 1999).
Another distinctive feature of the irregulars is that all of them are monosyllables or prefixed monosyllables (at least in English) like oversleep and become, and follow the canonical sound pattern for simple English words. Therefore, there are no phonologically unwieldy English irregulars, although some English regulars like hatched, months are less pronounceable.
Up to the present, most of the studies on the processing of regular and irregular forms have focused on English past tense verbal forms, for regular and irregular English verbs are equated for meaning (referring to past), complexity (one word), syntax (tensed) and could be matched on the syllable structure and frequency. As Pinker (1991: 531) stated, past tense inflection is an isolable subsystem in which grammatical mechanisms can be studied in detail, without complex interactions with the rest of the language, and past tense marking -ed is also insensitive to lexical semantics in that both regular and irregular past verbs express past events or states and that the regular-irregular distinction does not correlate with any feature of verb meaning.1.3Research Questions and Significance
In the present study, the debate on the processing of regular and irregular forms will be reexamined in a foreign language context, and some important factors will be investigated as well. To be specific, Chinese learners of English as a foreign language (EFL) will be investigated to examine their processing of the past tense forms of regular and irregular English verbs. The research questions are as follows:
●Research question 1
What is the role of frequency, phonological similarity and regularity in Chinese EFL learners' processing of the past tense forms of regular and irregular English verbs?
●Research question 2
What is the role of Chinese learners' English proficiency, their age of the first exposure to English and their duration of exposure to English in their processing of the past tense forms of regular and irregular English verbs?
●Research question 3
Which theories, the single-mechanism models or the dual-mechanism models, can better explain Chinese EFL learners' processing of the past tense forms of regular and irregular English verbs?
The first research question focuses on the factors of the past tense forms of regular and irregular English verbs and they are within-subjects factors. The second research question focuses on the factors concerning foreign language learners and they are between-subjects factors. And the third research question aims to examine which theory can provide a better account of the Chinese EFL learners' processing mechanism of the past tense forms of regular and irregular English verbs.
The significance of the current study is as follows. First and foremost, the research on foreign language learners' processing of regular and irregular forms could shed new light on the nature of the mental representation of the human languages. This could advance our understandings of the complex mechanisms of the processing of regular and irregular forms and in turn explore the operating mechanism of the human brain. This is not only a central topic in cognitive science and neuroscience, but also the ultimate goal of linguistic inquiry.
Secondly, it provides further evidence for and new insight into the existing models and approaches from the perspective of a foreign language, for the current models are all put forward based on the processing of native languages. Pinker (1999) claimed that the distinction between memory and rules systems in general, and the symbolic rule-based default in particular, is a linguistic universal, which makes it possible to probe the issue of language processing mechanism in second or foreign language contexts. If this is the case, then one should find evidence for these two systems and a symbolic default operating in different languages and by different language users.
And finally, very few studies have systematically investigated the claims of the current models and approaches in the foreign language context. It is important to do so, because foreign languages are quite different from native languages, even the second languages. An investigation of the processing of the past tense forms of regular and irregular English verbs in a foreign language context is an indispensable supplement for the research on the processing of regular and irregular forms. It enables us to identify the ways in which
试读结束[说明：试读内容隐藏了图片]