Temporal Discourse Models for Narrative Structure

STUDIES IN LITERATURE AND LANGUAGEVol. 1, No. 6, 2010, pp. 75-81ISSN 1923-1555 [P RINT ] ISSN 1923-1563[O NLINE ] 75Metaphors in Advertising DiscourseLuu Trong Tuan 1Abstract: Metaphors are the mappings of the abstract world into the concrete world through human senses or experiences. In Vietnamese advertisements, brands are metaphorized and brand metaphors can be categorized into ontological and structural metaphors. BRAND IS MOTION is a structural metaphor, and BRAND IS A CONTAINER, BRAND IS A VALUABLE RESOURCE, BRAND IS A COMPANION, and BRAND IS A GLADIATOR are instances of ontological metaphors.Keywords: conceptual metaphors; ontological metaphors; structural metaphors; brand; Vietnamese advertisements1. INTRODUCTIONMetaphors can be used verbally in the headline and/or copy. There are many advantages linked with the use of metaphors in advertising discourse. First, they elicit more cognitive elaboration than literal messages (Toncar and Munch 2001), presumably since individuals need to comprehend the complex message to draw inferences (Mick 1992). Their artful deviations provide intrinsic rewards that come from processing various interpretations of the text (Barthes 1986). Second, resolving such deviations or incongruities leads to favorable attitudes (McQuarrie and Mick 1999). Third, metaphors inject novelty, thus increasing motivation to read and process the ad (Goodstein 1993). Fourth, promotional metaphors, which are usually apt, comprehensive, and memorable, influence consumer beliefs and affect (Ward and Gaidis 1990). Another advantage of metaphors is their centrality to the process of imagination (Goldman 1986; Oliver, Robertson, and Mitchell 1993). According to Zaltman and Coulter (1995): "Without metaphors, we cannot imagine. They are the engines of imagination." Finally, McQuarrie and Phillips (2005) observed that consumers are more receptive to multiple, distinct, and positive inferences about the brand when metaphoric advertising is adopted.Metaphor has been portrayed as “an expression which describes a person or object in a literary way’ (Walter & Woodford 2005) and is the ‘result of some operation performed upon the literal meaning of the utterance…’ (Lakoff & Johnson 1980a: 453). This traditional way of defining metaphor had been used for centuries; and although currently deemed ‘false’ and obsolete (Lakoff 1993: 202), this traditional portrayal of metaphor can still be seen in the dictionary entry.Nonetheless, traditional views on metaphor were ultimately modified when Lakoff & Johnson (1980b) introduced a different approach to understanding and categorizing metaphor with their theory of conceptual 1National University of Ho Chi Minh City Bio Data Luu Trong Tuan is currently an EFL teacher at National University of Ho Chi Minh City. He received his M.TESOL from Victoria University, Australia in 2004. Besides his focus on TESOL, his recent publications such as Language Transfer is Cultural Transfer between Communities, Social Sciences Review , No. 11, 2004, pp. 60-63; and Principles for Scientific Translation, Social Sciences Review , No. 8, 2004, pp. 63-67; and Building Vietnamese Medical Terminology via Language Contact, Australian Journal of Linguistics , Vol. 29, No. 3, September 2009, pp. 315-336 show his interest in language contact and translation areas. *Received 15 July 2010; accepted 29 August 2010metaphor. In this theory, they argue that metaphors pervade our way of conceiving of the world and are encountered extensively in several of our languages, thoughts and actions (p. 3). A reason for such widespread use is potentially that, by definition, conceptual metaphor aids the understanding of the non-physical by contrasting and categorizing abstract concepts with physical reality (Kövecses 2002: 6); this would be hard, if not impossible, to accomplish without the use of metaphor (p. 7).Instances of abstract concepts employed to relate to physical realities include time, for example, that has been defined as a ‘continuum that lacks spatial dimensions’ (‘Time’ 2006). Due to a lack of physical depiction, human beings have had to relate to time with perceivable physical experiences in order to explicate it. In English language, time has been compared with motion, an adversary (Kövecses 2002: 285), a container (Lakoff & Johnson 1980b: 29), as well as money (p. 7). The phrase “spending time” is an illustration of money (use) being compared to time. Since a comparison is being made between concepts that evidently do not share physical properties (money-use, for example, is in no way physically similar to time that does not possess physical properties), these comparisons are utterly ‘metaphorical’.In marketing, a service, as an intangible product, lacks spatial dimensions; however, a product manufactured by a firm is a tangible entity with spatial dimensions. Numerous advertisements, nonetheless, are meant to seed value creation in customers, so market brands rather than products. Brands which lacks spatial dimensions are metaphorically expressed.2. LITERATURE REVIEWMetaphors in advertising literature has been primarily conceptual. Ward and Gaidis (1990) reviewed several models of metaphor comprehension and quality which were grounded in work from psychology and linguistics. Furthermore, Scott (1994) has argued for a theory of visual rhetoric to help researchers frame how meaning is constructed via visual arguments in advertisements. These contributions have provided both a call for, and a fertile ground for research into, the effects of metaphors in the marketing communication context.Those responding to the call have provided valuable insight into the effects of these advertising message strategies. Findings suggest that consumers spend more time looking at and processing ads that contain metaphors (Gray and Snyder 1989). MacInnis, Moorman and Jaworski (1991) argue that as executional cues, metaphors are interesting, they stimulate curiosity about the brand, and consequently, they result in deeper levels of processing. Pawlowski, Badzinski and Mitchell (1998) found that children's cognitive development plays a role in the comprehension of metaphors in ads, which in turn affects memory. They found that although young readers may have difficulty interpreting metaphors, there was a slight advantage in recall and perceptions compared to literal advertisements.A crucial consideration for advertising research is determining if metaphors are correctly interpreted or even understood by consumers. According to Ward and Gaidis (1990), comprehensibility is an important variable in the study of metaphors: "To be effective, a promotional metaphor must be minimally comprehended by its intended audience" (p. 636). Stern (1988) suggests that a significant proportion of the intended audience does not always "get" the intended meaning of the metaphor. Work by Phillips (1997) highlights the magnitude of metaphor comprehension in advertising. Phillips (1997) found that while strong pictorial implicatures (metaphors in which central meaning is manifest and difficult to misinterpret) were interpreted as the advertising creator intended, weak implicatures (those which require "work" by the viewer to interpret) were either misinterpreted, or solicited multiple divergent interpretations.3. ONTOLOGICAL AND STRUCTURAL METAPHORS According to Lakoff and Johnson (1980b), conceptual metaphors penetrate our understanding of the world around us. Metaphors are present in everyday speech, in every language, and are to a certain extent, culture specific (p. 3).76This perspective maintains that one conceptual domain, is comprehended via another conceptual domain, and is expressed as DOMAIN A IS DOMAIN B (Kövecses 2002: 4). DOMAIN A that refers to any abstract concepts (or source domains) is related in some way to DOMAIN B, concrete objects (or target domains). Our understanding of the workings or features of the concrete domains will help us partially relate to the abstract ones (p. 4). This process is unidirectional, or non-reversible; this implies that the target domain can not normally be comprehended via the source domain (p. 6). This is logical, as individuals need to be able to relate the concrete to the abstract in order to understand the less physical world. In addition, this is a highly automated process that is used unconsciously; as Lakoff and Turner (1989) put it,Metaphor is a tool so ordinary that we use it unconsciously and automatically… it isirreplaceable: metaphor allows us to understand ourselves and our world in ways that no othermodes of thought can. (xi)The use of capital letters, when referring to the domains, signifies that conceptual metaphors, being mental categories, are not necessarily expressed in a language. Nevertheless, all metaphorical expressions are written in lower-case letters (Kövecses 2002: 4). This signifies that conceptual metaphors in a language are expressed through metaphorical expressions.Even though all conceptual metaphors function according to the cognitive theory of metaphor illustrated above, there are different categories of classification. The metaphorical expressions structured by conceptual metaphors that will be investigated in this paper fall into two categories: structural and ontological metaphors. The disparity between these two types of metaphor is that structural metaphors include an extremely well-defined target domain that will help structure the abstract source domain (Kövecses 2002: 33). Ontological metaphors, on the other hand, do not have such a well-defined target domain (p. 34). Ontological metaphors purely categorize the abstract source domain into objects, substances, and containers and our understanding of the three is rather limited and quite general (Lakoff & Johnson 1980b: 25). If the mappings of the ontological metaphors were richer than they are, they would be considered structural metaphors (Kövecses 2002: 35).3.1 Ontological metaphorsOntological metaphors are one of the least noticeable types of conceptual metaphor (Lakoff & Johnson 1980b: 28). They enable us to understand our own experiences as regards concrete objects, by picking out certain parts of our own experiences from the whole in order to categorize, group and quantify them. In other words, we ‘impose artificial boundaries’ on abstract entities (Kövecses 2002: 25).BRAND IS A CONTAINEROne concrete reality encountered among the ontological metaphors that relates to brand, is a container, BRAND IS A CONTAINER. In the ensuing advertisement, whereas Yomost yoghurt contains ingredients derived from nature, Yomost brand contains more than that: the brand is a container of rhythms of life.Bản hòa tấu của những hương vị thảo nguyên(A concert of the flavors from the pasture)(Advertisement of Yomust yoghurt)The brands “Vinaphone” and “HTC” contain the entire world:Cùng Vinaphone và HTC Wildfire mang cả thế giới vào điện thoại di động của bạnVinaphone and HTC Wildfire bring the whole world into your cellphone.BRAND IS A VALUABLE RESOURCEAnother conceptual metaphor that structures the view of brand in Vietnamese advertising discourse is the BRAND IS A VALUABLE RESOURCE conceptual metaphor. This conceptual metaphor can be implicitly expressed in advertisements:77Du lịch Việt Nam - Vẻđẹp tiềm ẩn.(Vietnam tourism – the hidden charm)It can be explicitly expressed to highlight that brand value is linked to life:La Vie - Một phần tất yếu của cuộc sống(La Vie – the integral part of life)Ởđâu có sóng ởđó có sự sống.(Where there’s air – there will be life)(Mobifone’s advertisment of 3G)Or via a valuable commodity such as gold:Đầu tư ngay, tay chạm vàng(Invest instantly to touch gold)(Advertisement of The EverRich II apartments)Ensure Gold – Sống khỏe với trái tim vàng(Ensure Gold – Live healthily with a gold heart)BRAND IS A COMPANIONBrand, furthermore, is metaphorized as a companion who understands and supports customers in their quest for their own values in their life as well as their career. Take the following advertisements as examples: Tay trong tay trên đường thành công(Hand in hand on the journey towards success)(Advertisement of Fiat car in Tuoi Tre (Youth) Newspaper on October 16,2001)Người bạn đường dũng mãnh đang đến(A strong companion is approach)(Advertisement of Ford car in Tuoi Tre (Youth) Newspaper on June 5, 2001) In the examples above, even though “tay trong tay” can be understood as “Fiat car user’s hand on the car’s steering wheel” (in Vietnamese language, “tay” has two homonyms: “tay” implying “hand” and “tay” implying “steering wheel”) and “người bạn đường dũng mãnh” (a strong companion) can be discerned as “Ford car”, both Fiat and Ford mean to transfer their brand values to their customers and promote them to higher levels in society as well as push them towards opportunities. “Dũng mãnh” (strong) here denotes the power of the brands.The subsequent examples show that brand values are made prominent and obscure the concreteness of the products (FPT, S.J.C.) or services (AIA).FPT - Cùng đi tới thành công.(FPT – Towards success together)Gửi trọn niềm tin cho người dẫn đường tận tụy(Place all the trust on the guide)(Advertisement of AIA Insurance Company in Tuoi Tre (Youth) Newspaper onApril 13, 2001)S.J.C. – Người bạn chuẩn mực và tin cậy(S.J.C – A trustworthy friend)BRAND IS A GLADIATOR78An extension of ontological metaphors subsists in the form of personification. Personification is expressed in conceptual metaphors through a source domain that has been ascribed human characteristics (Lakoff & Johnson 1980b: 34).In the conceptual metaphor BRAND IS A GLADIATOR, we have traveled from comprehending BRAND IS A PERSON to comprehending brand in a specific way; we now acknowledge how to think about and act towards it (Lakoff & Johnson 1980b: 34). Brand values in the ensuing advertisements are personificated into gladiators fighting against any harms towards the beauty and the peace of human life: Mosfly – Hạ gục nhanh, tiêu diệt gọn(Mosfly – Defeat briskly, eradicate thoroughly.)OMO- Đánh bật 99 vết bẩn(OMO – Eliminate 99 sorts of dirt)Xóa đi dấu vết thời gian(Erase the traces of time)(Advertisement of DeBon in Heritage (Di Sản) magazine, issue December and January2001)3.2 Structural metaphorsUnlike ontological metaphors, the target domains in structural metaphors are understood through the intricate makeup and detailed knowledge we have of the source domains (Kövecses 2002: 33).The three versions of the BRAND IS MOTION conceptual metaphor that will be investigated in this paper embrace:1) The brand is deemed a point towards which the observer (customer) moves, as illustrated in the following ad in which customers move toward Vfresh brand:Trái tim ơi, để chắc ai đó thật sự quan tâm đến bạn, hãy tìm tôi(Heart! To be sure that person whole-heartedly cares for you, please find me.)(Vfresh soy milk)2) The observer (customer) is a point towards which the brand moves, as San Miguel brand is boldly rushing towards customers shaking things around:Long trời lở tuyết(Heaven shakes and snow falls)or Ajinomoto travels around the world to visit every customer:Vòng quanh thế giới, Ajinomoto.(Around the world, Ajinomoto)and Yomost brand moves towards you, carrying his love:Hãy để cánh bướm Yomost nối nhịp yêu thương(Let the Yomost butterfly connect heartbeats)3) The brand, in certain cases, moves, and customers move along. The advertisement below suggests consumers should move along the continuous technological innovation of Sony:Sony – luôn đi trước thời đại(Sony – always go ahead of time)794.CONCLUDING REMARKSMetaphors are the mappings of the abstract world into the concrete world through human senses or experiences. Similar to time metaphors, brand metaphors can be categorized into ontological and structural metaphors. BRAND IS MOTION is a structural metaphor, and BRAND IS A CONTAINER, BRAND IS A VALUABLE RESOURCE, BRAND IS A COMPANION, and BRAND IS A GLADIATOR are instances of ontological metaphors. Cognitive theory, along with views on ontological and structural metaphors not merely help explore time metaphors and brand metaphors, but also metaphors of other abstract concepts by mapping them into the dimensions of space or the dimensions of human experiences, so abstract concepts become no longer abstract through human cognitive lens.REFERENCESBarthes, R. (1986 [1964]). Rhétorique de l’image. Communications4,40–51.Goldman, A.I. (1986). Epistemology and Cognition. Cambridge, MA: Harvard University Press. Goodstein, R.C. (1993). Category-based Applications and Extensions in Advertising: Motivating More Extensive Ad Processing. Journal of Consumer Research, 20 (June), 87-99.Gray, S.A., and Snyder, R. (1989). Metaphors in Advertising: Effects on Memory. In Gardner, M.P. (ed.), Proceedings of the Society for Consumer Psychology. Washington, DC: American Psychological Association, 85-87.Kövecses, Z. (2002). Metaphor: a practical introduction. New York: Oxford University Press.Lakoff, G. (1993). The contemporary theory of metaphor. In A. Ortony (Ed.), Metaphor and Thought. 2nd ed., (202-251). Cambridge: Cambridge University Press.Lakoff, G. and Johnson, M. (1980a). Conceptual Metaphor in Everyday Language. The Journal of Philosophy, 77 (8), 453-486.Lakoff, G. and Johnson, M. (1980b). Metaphors we live by. Chicago: University of Chicago Press. Lakoff, G. and Turner, M. (1989). More then cool reason: a field guide to poetic metaphor. Chicago: University of Chicago Press.MacInnis, D.J., Moorman, C., and Jaworski, B.J. (1991). Enhancing and Measuring Consumers' Motivation, Opportunity, and Ability to Process Brand Information from Ads. Journal of Marketing, 55(October), 32-53.McQuarrie, E., & Mick, D.G. (1999). Visual rhetoric in advertising: text-interpretive, experimental and reader-response analyses. Journal of Consumer Research26, 37–53.McQuarrie, E., & Phillips, B. (2005). Indirect Persuasion in Advertising: How Consumers Process Metaphors Presented in Pictures and Words. Journal of Advertising 34 (2), 7–21.Mick, D.G. (1992). Levels of subjective comprehension in advertising processing and their relations to ad perceptions, attitudes, and memory. Journal of Consumer Research, 18(4), 411-424.Oliver, R.L., Robertson, T.S., & Mitchell, D.J. (1993). Imaging and analyzing in response to new product advertising. Journal of Advertising, 22(4), 35-49.Pawlowaki, D.R., Badzinski, D.M., and Mitchell, N. (1998). Effects of Metaphors on Children's Comprehension and Perception of Print Advertisements. Journal of Advertising, 27(Summer), 83-98. Phillips, B.J. (1997). Thinking Into It: Consumer Interpretation of Complex Advertising Images. Journal of Advertising, 26(Summer), 77-87.80Scott, L.M. (1994). Images in Advertising: The Need for a Theory of Visual Rhetoric. Journal of Consumer Research, 21(September), 252-273.Stern, B.B. (1988). Medieval Allegory: Roots of Advertising Strategy for the Mass Market. Journal of Consumer Research, 52(JuIy), 84-94.Time (2006). In Encyclopœdia Britannica. Retrieved November 18, 2006, from Encyclopædia Britannica Online: /eb/article-9108686Toncar, M., & Munch, J. (2001). Consumer responses to tropes in print advertising. Journal of Advertising 30, 55–65.Walter, E., and Woodford, K. (Eds.). (2005). Cambridge Advanced Learner’s Dictionary (CALD) [CD-ROM]. Cambridge: Cambridge University Press.Ward, J., and Gaidis, W. (1990). Metaphor in Promotional Communication: A Review of Research on Metaphor Comprehension and Quality. Advances in Consumer Research, V ol. 17, Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, eds., Provo, UT Association for Consumer Research, 636-642. Zaltman, G., and Coulter, R.H. (1995). Seeing the voice of the customer: Metaphor-based advertising research. Journal of advertising research, , July/August, pp. 35-51.81。

第⼀部分：基本传播学理论词汇媒介事件 Media Events民族志 Ethnography传播⽣态 Ecology of Communication真实/虚构 Reality/Fiction拟态环境 Pseudo-Environment刻板成见 Stereotyping晕轮效应 Halo Effects⼆元价值评判 Two-Valued Evaluation公共关系 Public Relation阐释理论 Interpretive Theory⾮语⾔符号 Nonverbal Sign⾮语⾔传播 Nonverbal Communication 意指 Signification话语理论 Theories of Discourse⽂化期待 Culture Expectations⽂化批判 Culture Criticizing范式 Paradigm叙事范式 Narrative Paradigm强语境 High Context弱语境 Low Context功能理论 Functionalism话语分析 Discourse Analysis传播的商品形式 the Commodity Forms of Communication受众商品 Audience Commodity商品化 Commodification空间化 Spatialization结构化 Structuration媒介集中化 Media Conglomeration传媒产业 Media Industry注意⼒经济 Attention Economy媒介竞争 Media Competition受众分割 Audience Segmentation媒介资本 Media Capital传播政治经济学 Political Economy of Communication传播研究 Communication Research抽样 Sampling 调查研究⽅法 Survey Research内容分析法 Content Analysis实验分析法 Experimental Research定性研究法 Qualitative Research Methods个案研究法 Case Study效度与信度 Validity/Reliability变量 Variables实地观察法 Field Observation虚拟社群 Virtual Community扩散研究 Diffusion Research传播 Communication内向/⾃我传播 Intrapersonal Communication ⼈际传播 Interpersonal Communication群体传播 Group Communication组织传播 Organization Communication⼤众传播 Mass Communication单向传播 One-Sided Communication双向传播 Two-Sided Communication互动传播 Interactive Communication媒介 Media⼤众传播媒介 Mass Media新媒介 New Media新闻洞 News Hold新闻价值 News Value传播者 Communicator主动传播者 Active Communicator受传者/受众/阅听⼤众 Audience受众兴坤 Audience Interest受众⾏为 Audience Activity信息 Information信号 Signal讯息 Message信息熵 Entropy冗余/冗余信息 Redundancy传播单位 Communication Unit奥斯古德模式 Osgood Model编码 Encoding解码 Decoding信源 Source传播的数学理论 Mathematical Theory of Communication传播渠道 Communication Channel有效传播 Effective Communication传播效果 Effects知识沟 Knowledge-Gap使⽤与满⾜模式 Uses and Gratifications Model 使⽤与依从模式 Uses and Dependencys Model ⼜传系统 System of Oral Communication地球村 Global Village内爆 Implosion全球化 Globalization本⼟化 Localization电⼦空间 Cyber Space数字化 Digitalization⽂化帝国主义 Culture Imperialism跨⽂化传播 Intercultural Communication守门⼈ Gatekeeper新闻采集者 News Gatherers新闻加⼯者 News Processors模式 Model有线效果模式 Limited Effects Model适度效果模式 Moderate Effects Model强⼤效果模式 Powerful Effects Model⼦弹论 Bullet Theory两级传播模式 Two-Step Flow Model多级传播模式 Multi-Step Flow Model沉默的螺旋模式 Spiral of Silence Model劝服传播 Persuasive Communication议程设置模式 the Agenda-Setting Model时滞 Time Lag最合适效果跨度 Optimal Effects Pan时间跨度 Time Span公众舆论 Public Opinion选择性接触 Selective Exposure选择性注意 Selective Attention选择性理解 Selective Perception选择性记忆 Selective Retention可信性提⽰ Credibility Heuristic喜爱提⽰ Liking Heuristic 共识提⽰ Consensus Heuristic市场驱动新闻学 the Market-Driven Journalism 意识形态 Ideology霸权 Hegemony权⼒话语 Power Discourse视觉⽂本 Visual Text⽂本 Text超级⽂本 Hypertext结构主义 Constructionism解构主义 Deconstructionism⽂化⼯业 Culture Industry⼤众⽂化 Mass Culture⽂化研究 Cultural Studies批判学派/批判理论 Critical Theory法兰克福学派 Frankfurt School⼥权主义/⼥性主义 Feminism符号学 Semiotics/Semiology符号 Sign能指与所指 Signified/SignifierFourth Estate 第四等级(新闻界的别称) freedom of the Press 新闻⾃由free-lancer n．⾃由撰稿⼈full position 醒⽬位置Good news comes on crutches. 好事不出门。

Conditional Random Fields:Probabilistic Modelsfor Segmenting and Labeling Sequence DataJohn Lafferty LAFFERTY@ Andrew McCallum MCCALLUM@ Fernando Pereira FPEREIRA@ WhizBang!Labs–Research,4616Henry Street,Pittsburgh,PA15213USASchool of Computer Science,Carnegie Mellon University,Pittsburgh,PA15213USADepartment of Computer and Information Science,University of Pennsylvania,Philadelphia,PA19104USAAbstractWe present,a frame-work for building probabilistic models to seg-ment and label sequence data.Conditional ran-domfields offer several advantages over hid-den Markov models and stochastic grammarsfor such tasks,including the ability to relaxstrong independence assumptions made in thosemodels.Conditional randomfields also avoida fundamental limitation of maximum entropyMarkov models(MEMMs)and other discrimi-native Markov models based on directed graph-ical models,which can be biased towards stateswith few successor states.We present iterativeparameter estimation algorithms for conditionalrandomfields and compare the performance ofthe resulting models to HMMs and MEMMs onsynthetic and natural-language data.1.IntroductionThe need to segment and label sequences arises in many different problems in several scientificfields.Hidden Markov models(HMMs)and stochastic grammars are well understood and widely used probabilistic models for such problems.In computational biology,HMMs and stochas-tic grammars have been successfully used to align bio-logical sequences,find sequences homologous to a known evolutionary family,and analyze RNA secondary structure (Durbin et al.,1998).In computational linguistics and computer science,HMMs and stochastic grammars have been applied to a wide variety of problems in text and speech processing,including topic segmentation,part-of-speech(POS)tagging,information extraction,and syntac-tic disambiguation(Manning&Sch¨u tze,1999).HMMs and stochastic grammars are generative models,as-signing a joint probability to paired observation and label sequences;the parameters are typically trained to maxi-mize the joint likelihood of training examples.To define a joint probability over observation and label sequences, a generative model needs to enumerate all possible ob-servation sequences,typically requiring a representation in which observations are task-appropriate atomic entities, such as words or nucleotides.In particular,it is not practi-cal to represent multiple interacting features or long-range dependencies of the observations,since the inference prob-lem for such models is intractable.This difficulty is one of the main motivations for looking at conditional models as an alternative.A conditional model specifies the probabilities of possible label sequences given an observation sequence.Therefore,it does not expend modeling effort on the observations,which at test time arefixed anyway.Furthermore,the conditional probabil-ity of the label sequence can depend on arbitrary,non-independent features of the observation sequence without forcing the model to account for the distribution of those dependencies.The chosen features may represent attributes at different levels of granularity of the same observations (for example,words and characters in English text),or aggregate properties of the observation sequence(for in-stance,text layout).The probability of a transition between labels may depend not only on the current observation, but also on past and future observations,if available.In contrast,generative models must make very strict indepen-dence assumptions on the observations,for instance condi-tional independence given the labels,to achieve tractability. Maximum entropy Markov models(MEMMs)are condi-tional probabilistic sequence models that attain all of the above advantages(McCallum et al.,2000).In MEMMs, each source state1has a exponential model that takes the observation features as input,and outputs a distribution over possible next states.These exponential models are trained by an appropriate iterative scaling method in the 1Output labels are associated with states;it is possible for sev-eral states to have the same label,but for simplicity in the rest of this paper we assume a one-to-one correspondence.maximum entropy framework.Previously published exper-imental results show MEMMs increasing recall and dou-bling precision relative to HMMs in a FAQ segmentation task.MEMMs and other non-generativefinite-state models based on next-state classifiers,such as discriminative Markov models(Bottou,1991),share a weakness we callhere the:the transitions leaving a given state compete only against each other,rather than againstall other transitions in the model.In probabilistic terms, transition scores are the conditional probabilities of pos-sible next states given the current state and the observa-tion sequence.This per-state normalization of transition scores implies a“conservation of score mass”(Bottou, 1991)whereby all the mass that arrives at a state must be distributed among the possible successor states.An obser-vation can affect which destination states get the mass,but not how much total mass to pass on.This causes a bias to-ward states with fewer outgoing transitions.In the extreme case,a state with a single outgoing transition effectively ignores the observation.In those cases,unlike in HMMs, Viterbi decoding cannot downgrade a branch based on ob-servations after the branch point,and models with state-transition structures that have sparsely connected chains of states are not properly handled.The Markovian assump-tions in MEMMs and similar state-conditional models in-sulate decisions at one state from future decisions in a way that does not match the actual dependencies between con-secutive states.This paper introduces(CRFs),a sequence modeling framework that has all the advantages of MEMMs but also solves the label bias problem in a principled way.The critical difference between CRFs and MEMMs is that a MEMM uses per-state exponential mod-els for the conditional probabilities of next states given the current state,while a CRF has a single exponential model for the joint probability of the entire sequence of labels given the observation sequence.Therefore,the weights of different features at different states can be traded off against each other.We can also think of a CRF as afinite state model with un-normalized transition probabilities.However,unlike some other weightedfinite-state approaches(LeCun et al.,1998), CRFs assign a well-defined probability distribution over possible labelings,trained by maximum likelihood or MAP estimation.Furthermore,the loss function is convex,2guar-anteeing convergence to the global optimum.CRFs also generalize easily to analogues of stochastic context-free grammars that would be useful in such problems as RNA secondary structure prediction and natural language pro-cessing.2In the case of fully observable states,as we are discussing here;if several states have the same label,the usual local maxima of Baum-Welch arise.bel bias example,after(Bottou,1991).For concise-ness,we place observation-label pairs on transitions rather than states;the symbol‘’represents the null output label.We present the model,describe two training procedures and sketch a proof of convergence.We also give experimental results on synthetic data showing that CRFs solve the clas-sical version of the label bias problem,and,more signifi-cantly,that CRFs perform better than HMMs and MEMMs when the true data distribution has higher-order dependen-cies than the model,as is often the case in practice.Finally, we confirm these results as well as the claimed advantages of conditional models by evaluating HMMs,MEMMs and CRFs with identical state structure on a part-of-speech tag-ging task.2.The Label Bias ProblemClassical probabilistic automata(Paz,1971),discrimina-tive Markov models(Bottou,1991),maximum entropy taggers(Ratnaparkhi,1996),and MEMMs,as well as non-probabilistic sequence tagging and segmentation mod-els with independently trained next-state classifiers(Pun-yakanok&Roth,2001)are all potential victims of the label bias problem.For example,Figure1represents a simplefinite-state model designed to distinguish between the two words and.Suppose that the observation sequence is. In thefirst time step,matches both transitions from the start state,so the probability mass gets distributed roughly equally among those two transitions.Next we observe. Both states1and4have only one outgoing transition.State 1has seen this observation often in training,state4has al-most never seen this observation;but like state1,state4 has no choice but to pass all its mass to its single outgoing transition,since it is not generating the observation,only conditioning on it.Thus,states with a single outgoing tran-sition effectively ignore their observations.More generally, states with low-entropy next state distributions will take lit-tle notice of observations.Returning to the example,the top path and the bottom path will be about equally likely, independently of the observation sequence.If one of the two words is slightly more common in the training set,the transitions out of the start state will slightly prefer its cor-responding transition,and that word’s state sequence will always win.This behavior is demonstrated experimentally in Section5.L´e on Bottou(1991)discussed two solutions for the label bias problem.One is to change the state-transition struc-ture of the model.In the above example we could collapse states1and4,and delay the branching until we get a dis-criminating observation.This operation is a special case of determinization(Mohri,1997),but determinization of weightedfinite-state machines is not always possible,and even when possible,it may lead to combinatorial explo-sion.The other solution mentioned is to start with a fully-connected model and let the training procedurefigure out a good structure.But that would preclude the use of prior structural knowledge that has proven so valuable in infor-mation extraction tasks(Freitag&McCallum,2000). Proper solutions require models that account for whole state sequences at once by letting some transitions“vote”more strongly than others depending on the corresponding observations.This implies that score mass will not be con-served,but instead individual transitions can“amplify”or “dampen”the mass they receive.In the above example,the transitions from the start state would have a very weak ef-fect on path score,while the transitions from states1and4 would have much stronger effects,amplifying or damping depending on the actual observation,and a proportionally higher contribution to the selection of the Viterbi path.3In the related work section we discuss other heuristic model classes that account for state sequences globally rather than locally.To the best of our knowledge,CRFs are the only model class that does this in a purely probabilistic setting, with guaranteed global maximum likelihood convergence.3.Conditional Random FieldsIn what follows,is a random variable over data se-quences to be labeled,and is a random variable over corresponding label sequences.All components of are assumed to range over afinite label alphabet.For ex-ample,might range over natural language sentences and range over part-of-speech taggings of those sentences, with the set of possible part-of-speech tags.The ran-dom variables and are jointly distributed,but in a dis-criminative framework we construct a conditional model from paired observation and label sequences,and do not explicitly model the marginal..Thus,a CRF is a randomfield globally conditioned on the observation.Throughout the paper we tacitly assume that the graph isfixed.In the simplest and most impor-3Weighted determinization and minimization techniques shift transition weights while preserving overall path weight(Mohri, 2000);their connection to this discussion deserves further study.tant example for modeling sequences,is a simple chain or line:.may also have a natural graph structure;yet in gen-eral it is not necessary to assume that and have the same graphical structure,or even that has any graph-ical structure at all.However,in this paper we will be most concerned with sequencesand.If the graph of is a tree(of which a chain is the simplest example),its cliques are the edges and ver-tices.Therefore,by the fundamental theorem of random fields(Hammersley&Clifford,1971),the joint distribu-tion over the label sequence given has the form(1),where is a data sequence,a label sequence,and is the set of components of associated with the vertices in subgraph.We assume that the and are given andfixed. For example,a Boolean vertex feature might be true if the word is upper case and the tag is“proper noun.”The parameter estimation problem is to determine the pa-rameters from training datawith empirical distribution. In Section4we describe an iterative scaling algorithm that maximizes the log-likelihood objective function:.As a particular case,we can construct an HMM-like CRF by defining one feature for each state pair,and one feature for each state-observation pair:. The corresponding parameters and play a simi-lar role to the(logarithms of the)usual HMM parameters and.Boltzmann chain models(Saul&Jor-dan,1996;MacKay,1996)have a similar form but use a single normalization constant to yield a joint distribution, whereas CRFs use the observation-dependent normaliza-tion for conditional distributions.Although it encompasses HMM-like models,the class of conditional randomfields is much more expressive,be-cause it allows arbitrary dependencies on the observationFigure2.Graphical structures of simple HMMs(left),MEMMs(center),and the chain-structured case of CRFs(right)for sequences. An open circle indicates that the variable is not generated by the model.sequence.In addition,the features do not need to specify completely a state or observation,so one might expect that the model can be estimated from less training data.Another attractive property is the convexity of the loss function;in-deed,CRFs share all of the convexity properties of general maximum entropy models.For the remainder of the paper we assume that the depen-dencies of,conditioned on,form a chain.To sim-plify some expressions,we add special start and stop states and.Thus,we will be using the graphical structure shown in Figure2.For a chain struc-ture,the conditional probability of a label sequence can be expressed concisely in matrix form,which will be useful in describing the parameter estimation and inference al-gorithms in Section4.Suppose that is a CRF given by(1).For each position in the observation se-quence,we define the matrix random variableby,where is the edge with labels and is the vertex with label.In contrast to generative models,con-ditional models like CRFs do not need to enumerate over all possible observation sequences,and therefore these matrices can be computed directly as needed from a given training or test observation sequence and the parameter vector.Then the normalization(partition function)is the entry of the product of these matrices:. Using this notation,the conditional probability of a label sequence is written as, where and.4.Parameter Estimation for CRFsWe now describe two iterative scaling algorithms tofind the parameter vector that maximizes the log-likelihood of the training data.Both algorithms are based on the im-proved iterative scaling(IIS)algorithm of Della Pietra et al. (1997);the proof technique based on auxiliary functions can be extended to show convergence of the algorithms for CRFs.Iterative scaling algorithms update the weights asand for appropriately chosen and.In particular,the IIS update for an edge feature is the solution ofdef.where is thedef. The equations for vertex feature updates have similar form.However,efficiently computing the exponential sums on the right-hand sides of these equations is problematic,be-cause is a global property of,and dynamic programming will sum over sequences with potentially varying.To deal with this,thefirst algorithm,Algorithm S,uses a“slack feature.”The second,Algorithm T,keepstrack of partial totals.For Algorithm S,we define the bydef,where is a constant chosen so that for all and all observation vectors in the training set,thus making.Feature is“global,”that is,it does not correspond to any particular edge or vertex.For each index we now define thewith base caseifotherwiseand recurrence.Similarly,the are defined byifotherwiseand.With these definitions,the update equations are,where.The factors involving the forward and backward vectors in the above equations have the same meaning as for standard hidden Markov models.For example,is the marginal probability of label given that the observation sequence is.This algorithm is closely related to the algorithm of Darroch and Ratcliff(1972),and MART algorithms used in image reconstruction.The constant in Algorithm S can be quite large,since in practice it is proportional to the length of the longest train-ing observation sequence.As a result,the algorithm may converge slowly,taking very small steps toward the maxi-mum in each iteration.If the length of the observations and the number of active features varies greatly,a faster-converging algorithm can be obtained by keeping track of feature totals for each observation sequence separately. Let def.Algorithm T accumulates feature expectations into counters indexed by.More specifically,we use the forward-backward recurrences just introduced to compute the expectations of feature and of feature given that.Then our param-eter updates are and,whereand are the unique positive roots to the following polynomial equationsmax max,(2)which can be easily computed by Newton’s method.A single iteration of Algorithm S and Algorithm T has roughly the same time and space complexity as the well known Baum-Welch algorithm for HMMs.To prove con-vergence of our algorithms,we can derive an auxiliary function to bound the change in likelihood from below;this method is developed in detail by Della Pietra et al.(1997). The full proof is somewhat detailed;however,here we give an idea of how to derive the auxiliary function.To simplify notation,we assume only edge features with parameters .Given two parameter settings and,we bound from below the change in the objective function with anas followsdefwhere the inequalities follow from the convexity of and.Differentiating with respect to and setting the result to zero yields equation(2).5.ExperimentsWefirst discuss two sets of experiments with synthetic data that highlight the differences between CRFs and MEMMs. Thefirst experiments are a direct verification of the label bias problem discussed in Section2.In the second set of experiments,we generate synthetic data using randomly chosen hidden Markov models,each of which is a mix-ture of afirst-order and second-order peting models are then trained and compared on test data.As the data becomes more second-order,the test er-ror rates of the trained models increase.This experiment corresponds to the common modeling practice of approxi-mating complex local and long-range dependencies,as oc-cur in natural data,by small-order Markov models.OurFigure3.Plots of error rates for HMMs,CRFs,and MEMMs on randomly generated synthetic data sets,as described in Section5.2. As the data becomes“more second order,”the error rates of the test models increase.As shown in the left plot,the CRF typically significantly outperforms the MEMM.The center plot shows that the HMM outperforms the MEMM.In the right plot,each open square represents a data set with,and a solid circle indicates a data set with.The plot shows that when the data is mostly second order(),the discriminatively trained CRF typically outperforms the HMM.These experiments are not designed to demonstrate the advantages of the additional representational power of CRFs and MEMMs relative to HMMs.results clearly indicate that even when the models are pa-rameterized in exactly the same way,CRFs are more ro-bust to inaccurate modeling assumptions than MEMMs or HMMs,and resolve the label bias problem,which affects the performance of MEMMs.To avoid confusion of dif-ferent effects,the MEMMs and CRFs in these experiments use overlapping features of the observations.Fi-nally,in a set of POS tagging experiments,we confirm the advantage of CRFs over MEMMs.We also show that the addition of overlapping features to CRFs and MEMMs al-lows them to perform much better than HMMs,as already shown for MEMMs by McCallum et al.(2000).5.1Modeling label biasWe generate data from a simple HMM which encodes a noisy version of thefinite-state network in Figure1.Each state emits its designated symbol with probabilityand any of the other symbols with probability.We train both an MEMM and a CRF with the same topologies on the data generated by the HMM.The observation fea-tures are simply the identity of the observation symbols. In a typical run using training and test samples, trained to convergence of the iterative scaling algorithm, the CRF error is while the MEMM error is, showing that the MEMM fails to discriminate between the two branches.5.2Modeling mixed-order sourcesFor these results,we usefive labels,(),and26 observation values,();however,the results were qualitatively the same over a range of sizes for and .We generate data from a mixed-order HMM with state transition probabilities given byand,simi-larly,emission probabilities given by.Thus,for we have a standardfirst-order HMM.In order to limit the size of the Bayes error rate for the resulting models,the con-ditional probability tables are constrained to be sparse. In particular,can have at most two nonzero en-tries,for each,and can have at most three nonzero entries for each.For each randomly gener-ated model,a sample of1,000sequences of length25is generated for training and testing.On each randomly generated training set,a CRF is trained using Algorithm S.(Note that since the length of the se-quences and number of active features is constant,Algo-rithms S and T are identical.)The algorithm is fairly slow to converge,typically taking approximately500iterations for the model to stabilize.On the500MHz Pentium PC used in our experiments,each iteration takes approximately 0.2seconds.On the same data an MEMM is trained using iterative scaling,which does not require forward-backward calculations,and is thus more efficient.The MEMM train-ing converges more quickly,stabilizing after approximately 100iterations.For each model,the Viterbi algorithm is used to label a test set;the experimental results do not sig-nificantly change when using forward-backward decoding to minimize the per-symbol error rate.The results of several runs are presented in Figure3.Each plot compares two classes of models,with each point indi-cating the error rate for a single test set.As increases,the error rates generally increase,as thefirst-order models fail tofit the second-order data.Thefigure compares models parameterized as,,and;results for models parameterized as,,and are qualitatively the same.As shown in thefirst graph,the CRF generally out-performs the MEMM,often by a wide margin of10%–20% relative error.(The points for very small error rate,with ,where the MEMM does better than the CRF, are suspected to be the result of an insufficient number of training iterations for the CRF.)HMM 5.69%45.99%MEMM 6.37%54.61%CRF 5.55%48.05%MEMM 4.81%26.99%CRF 4.27%23.76%Using spelling featuresFigure4.Per-word error rates for POS tagging on the Penn tree-bank,usingfirst-order models trained on50%of the1.1million word corpus.The oov rate is5.45%.5.3POS tagging experimentsTo confirm our synthetic data results,we also compared HMMs,MEMMs and CRFs on Penn treebank POS tag-ging,where each word in a given input sentence must be labeled with one of45syntactic tags.We carried out two sets of experiments with this natural language data.First,we trainedfirst-order HMM,MEMM, and CRF models as in the synthetic data experiments,in-troducing parameters for each tag-word pair andfor each tag-tag pair in the training set.The results are con-sistent with what is observed on synthetic data:the HMM outperforms the MEMM,as a consequence of the label bias problem,while the CRF outperforms the HMM.The er-ror rates for training runs using a50%-50%train-test split are shown in Figure5.3;the results are qualitatively sim-ilar for other splits of the data.The error rates on out-of-vocabulary(oov)words,which are not observed in the training set,are reported separately.In the second set of experiments,we take advantage of the power of conditional models by adding a small set of or-thographic features:whether a spelling begins with a num-ber or upper case letter,whether it contains a hyphen,and whether it ends in one of the following suffixes:.Here wefind,as expected,that both the MEMM and the CRF benefit signif-icantly from the use of these features,with the overall error rate reduced by around25%,and the out-of-vocabulary er-ror rate reduced by around50%.One usually starts training from the all zero parameter vec-tor,corresponding to the uniform distribution.However, for these datasets,CRF training with that initialization is much slower than MEMM training.Fortunately,we can use the optimal MEMM parameter vector as a starting point for training the corresponding CRF.In Figure5.3, MEMM was trained to convergence in around100iter-ations.Its parameters were then used to initialize the train-ing of CRF,which converged in1,000iterations.In con-trast,training of the same CRF from the uniform distribu-tion had not converged even after2,000iterations.6.Further Aspects of CRFsMany further aspects of CRFs are attractive for applica-tions and deserve further study.In this section we briefly mention just two.Conditional randomfields can be trained using the expo-nential loss objective function used by the AdaBoost algo-rithm(Freund&Schapire,1997).Typically,boosting is applied to classification problems with a small,fixed num-ber of classes;applications of boosting to sequence labeling have treated each label as a separate classification problem (Abney et al.,1999).However,it is possible to apply the parallel update algorithm of Collins et al.(2000)to op-timize the per-sequence exponential loss.This requires a forward-backward algorithm to compute efficiently certain feature expectations,along the lines of Algorithm T,ex-cept that each feature requires a separate set of forward and backward accumulators.Another attractive aspect of CRFs is that one can imple-ment efficient feature selection and feature induction al-gorithms for them.That is,rather than specifying in ad-vance which features of to use,we could start from feature-generating rules and evaluate the benefit of gener-ated features automatically on data.In particular,the fea-ture induction algorithms presented in Della Pietra et al. (1997)can be adapted tofit the dynamic programming techniques of conditional randomfields.7.Related Work and ConclusionsAs far as we know,the present work is thefirst to combine the benefits of conditional models with the global normal-ization of randomfield models.Other applications of expo-nential models in sequence modeling have either attempted to build generative models(Rosenfeld,1997),which in-volve a hard normalization problem,or adopted local con-ditional models(Berger et al.,1996;Ratnaparkhi,1996; McCallum et al.,2000)that may suffer from label bias. Non-probabilistic local decision models have also been widely used in segmentation and tagging(Brill,1995; Roth,1998;Abney et al.,1999).Because of the computa-tional complexity of global training,these models are only trained to minimize the error of individual label decisions assuming that neighboring labels are correctly -bel bias would be expected to be a problem here too.An alternative approach to discriminative modeling of se-quence labeling is to use a permissive generative model, which can only model local dependencies,to produce a list of candidates,and then use a more global discrimina-tive model to rerank those candidates.This approach is standard in large-vocabulary speech recognition(Schwartz &Austin,1993),and has also been proposed for parsing (Collins,2000).However,these methods fail when the cor-rect output is pruned away in thefirst pass.。

外刊每日精读 | Clause for thought文章脉络【1】新式解码器首次让人们可以通过非入侵方式读懂思想。

【2】最新的人工智能技术终于将读心术带入了现实世界，但有些硬性限制。

【3】解码器解读大脑活动的过程。

【4】大约有一半时间，文本与原词的意思非常接近，有时甚至完全吻合。

【5】解码器能够利用大脑活动准确描述其中的一些内容。

【6】科学家们认为非入侵性读心术是一种真正的飞跃，但也在努力减轻人们对这种新技术的担忧。

【7】大阪大学大脑活动视觉图像重构的先驱西本真司教授认为这项重要的发现，可以为脑机接口的发展奠定基础。

经济学人原文Clause for thought: first non-invasive way to read minds as AI turns brain activity into text【1】An AI-based decoder that can translate brain activity into a stream of text has been developed, in a breakthrough that allows thoughts to be read non-invasively for the first time. The decoder could reconstruct speech with uncanny accuracy while people listened to a story – or even silently imagined one – using only fMRI scan data. Previous language decoding systems have required surgical implants, and the latest advance raises the prospect of new ways to restore speech in patients strugglingto communicate as a result of stroke or motor neurone disease. Dr Alexander Huth, a neuroscientist who led the work at the University of Texas at Austin, said: “We were kind of shocked that it works as well as it does. I’ve been working on thisfor 15years … so it was shocking and exciting when it finally did work.”【2】Mind-reading has traditionally been the preserve of sci-fi, in characters such as the X-Men’s Jean Grey, but the latest AI technology has finally taken the concept into the real world. This decoder’s achievement overcomesa fundamental limitation of fMRI: while the technique can map brain activity toa specific location with incredibly high resolution, there is an inherent time lag, which makes tracking activity in real time impossible. The lag exists because fMRI scans measure the blood-flow response to brain activity, which peaks and returnsto baseline over about 10 seconds, meaning even the most powerful scanner cannot improve on this. “It’s this noisy,sluggish proxy for neural activity,” said Huth. This hard limit has hampered the ability to interpret brain activity in response to natural speech because it gives a “mishmash of information” spread over a few seconds.【3】However, the advent of large language models – the kind ofAI underpinning OpenAI’s ChatGPT – provided a new way in. These models are able to represent, in numbers, the semantic meaning of speech, allowing the scientists to look at which patterns of neuronal activity corresponded to strings of words with a particular meaning rather than attempting to read out activity word byword. The learning process was intensive: three volunteers were required to lie ina scanner for 16 hours each, listening to podcasts. The decoder was trained to match brain activity to meaning using the large language model GPT-1, a precursor to ChatGPT. Later, the same participants were scanned listening to a new storyor imagining telling a story and the decoder was used to generate text from brain activity alone.【4】About half the time, the text closely – and sometimes precisely– matched the intended meanings of the original words. “Our system works at the level ofideas, semantics, meaning,” said Huth. “This is the reason why what we get out is not the exact words, it’s the gist.” For instance, when a participant was played the words: “I don’t have my driver’s licence yet,” the decoder translated as: “She has not even started to learn to drive yet.”【5】In another case, the words: “I didn’t know whether to scream, cry or run away. Instead, I said: ‘Leave me alone!’” was decoded as: “Started to scream and cry, and then she just said: ‘I told you to leave me alone.’” The participants were also askedto watch four short, silent videos while in the scanner, and the decoder was able to use their brain activity to accurately describe some of the content, the paper in Nature Neuroscience reported.【6】“For a non -invasive method, this is a real leap forward compared to what’s been done before, which is typically single words or short sentences,” Huth said. Jerry Tang, a doctoral student at the University of Texas at Austin and co-author, said: “We take very seriously the concerns that it could be used for bad purposes, and have worked to avoid that. “We want to make sure people only use these typesof technologies when they want to and that it helps them.”【7】Prof Shinji Nishimoto of Osaka University, whohas pioneered the reconstruction of visual images from brain activity, described the paper as a “significant advance”. He said: “This is a non-trivial finding and can be a basis for the development of brain-computer interfaces.” The team now hope to assess whether the technique could be applied to other, more portable brain-imaging systems, such as functional near-infrared spectroscopy (fNIRS).。

托福阅读中的一句长难句讲解This nascent world system developed as a result of insatiable demands for nonlocal raw materials in different ecological regions where societies were developing along very similar evolutionary tracks toward greater complexity.本句话是来自tpo63：The Sumerians and Regional Interdependence 苏美尔人与区域相互依存这一句的难点在于：词汇难，术语多，结构复杂. 上下文逻辑理困难.第一步：句子成分划分和翻译同学们可以自己先划分句子成分，再看老师的划分。

This nascent world system + developed[ as a result of insatiable demands for nonlocal raw materials/ in different ecological regions]（where societies + were developing /along very similar evolutionary tracks/ toward greater complexity. )这句话主要有2部分：（一）主干句：主语：This nascent world system 谓语：developed 状语：as a result of insatiable demands for nonlocal raw materials in different ecological regions.这是一个有3个介词短语组成的状语结构：① as a result of insatiable demands由于贪得无厌的需求② (demands) for nonlocal raw materials对非本地的原材料的需求③ in different ecological regions（原材料）来自不同的生态区域⭐️难点2：这一些较难的生词/术语：1️⃣ nascent：“初生的、萌芽的”。

判断题1.Interlanguage is neither the native language nor the second language.(T)2.Krashen assumed that there were two independent means or routesof second language learning: acquisition and learning. (T)3.There are two interacting factors in determining language transfer insecond language learning. (F)4.Three important characteristics of interlanguage: systemacticity ,permeability and fossilization. (T)5.Intrinsic motivation:learners learn a second language for externalpurposes. (F)6.Neurolinguistics is the study of two related areas: language disordersand the relationship between the brain and language. (T)7.The brain is divided two sections: the higher section called the brainstem and the lower section called the cerebrum. (F)8.An interesting fact about these two hemispheres is that eachhemisphere controls the opposite half of the body in terms of muscle movement and sensation. (T)9.Most right-handed individuals are said to be right lateralized forlanguage. (F)10.C T scanning uses a narrow beam of X-ray to create brain images thattake the form of a series of brain slices. (T)11.1 Right hear advantage shows the right hemisphere is not superior forprocessing all sounds, but only for those that are linguistic in nature, thus providing evidence in support of view that the left side of the brain is specialized for language and that's where language centers reside. （f）12.2 Evidence in support of lateralization for language in left hemispherecomes from researches in Dichotic listening tasks（t）13.3interpersonal communications is the process of using languagewithin the individual to facilitate one’s own thought and aid the formulation and manipulation of concepts. （t）14.4 Linguistic lateralization is hemispheric specialization or dominancefor language. （t）15.5 Dichotic Listening is a research technique which has been used tostudy how the brain controls hearing and language, with which subjects wear earphones and simultaneously receive different sounds in the right or left ear, and are then asked to repeat what they hear.（f）16.6 Dichotic Listening is a research technique which has been used tostudy how the brain controls hearing and language, with which subjects wear earphones and simultaneously receive different soundsin the right and left ear, and are then asked to repeat what they hear.（t）17.7 Input refers to the language which a learner bears and receives andfrom which he or she can learn. （f）18.8 Fossilization ，a process that sometimes occurs in language learningin which incorrect linguistic features (such as the accent of a grammatical pattern) become a permanent part of the way a person speaks or writes in the target language.（f）19.9 The different languages have a similar level of complexity anddetail, and reflect general abstract properties of the common linguistic system is called Universal Grammar . （t）20.10 Acculturation a process of adapting to the culture and valuesystem of the second language community.（t）21.I n socialinguistic studies,speakers are not regarded as members ofsocial groups (F)22.n ew words maybe coined from already existing words by substractingan affix thought to be part of the old world (T)23.a ll languages make a distinction between the subject and directobject,which can be illustrated in word order (T)24.I t has been noticed that in many communities be language used bythe older generation differs from that used by the elder generation in certain ways (F)25.A pidgin is a special language variety that mixes or blends languagesand it isn’t used by people who speak different languages for restricted purposes such as trading(F)26.I t is interesting to know that the language used by men and womenhave some special features of others (F)27.I t is an obvious facts that people who claim to be speakers of thesame language don’t speak the language in the different manner (T)28.A regional dialect is a linguistic variety used by people living (T)29.F usion refers to this type of grammatication in which words developinto affixes (T)30.H istorical linguistics,as a branch of linguistics is mainly coverned withboth the description and explanation of language changes that occurred over time (T)选择题Chapter 71.Which one is not right about Blenging?(b)A:disco-discotheque B:brunch-breakfast+luchC:B2B-Business-to-Business D:videophone-video+cellphone2.Semantic changes contains three processes ,which one is ture?(a)A:namely widening ,narrowing and shift in meaningB:semantic broadening ,narrowing and semantic dispearingC:semantic shift ,narrowing and semantic lossingD:namely widening ,narrowing and not shift in meaning3.Science and technology influence English language in these aspects(d) A:space travelB:compnter and internet languageC:ecdogyD:above of allnguage changes can be found at different linguistic levels,such as in the<D>A:phonology and morphologyB:syntax and lexiconC:semantic component of the grammarD:ABC5,Morphological and syntactic change contian<D>A:addition or loss of affixesB:change of word ordenC:change in regation ruleD:abrove of allChapter 81.Which is not Halliday's social variables that determine the register? (D) A:field of discourseB:tenor of discourseC:mode of discouseD:ethnic dialect2.Which is not dialectal varieties?(C)A:regional dialect and idiolectB:language and genderC:registerD:ethnic dialect3.To some extent,language especially the structure of its lexicon,refects___of a sociey.(C)A:physical B:social environmentC:both AandB D:social phenomenon4.____,refers to the linguistic variety characteristic of a particular social class.(D)A:Social-class dialect B:sociolectC:A andB D:A or B5.Two languages are used side by side with each having a ____role to play;and language switching occurs when the situation ____.(A)A:different,changesB:similar,changesC:different,unchangingD:similar,unchangingChapter 91.which is not the component of culture ?<D>nguageB.ideasC.beliefD.soil2.in a word,language express<D>A.factsB.events which represent similar world knowledge by its peopleC.peoples' attitudes.beliefsD.cultural reality3.any linguistic sign may simultaneously have a <D>A.denotativeB.connotativeC.iconicD.denotative,connotative,or iconic kind of meanings4.what's the meaning of"a lucky dog"in english?<B>A.a clever boyB.a smart ladC.a lucky personD.a silent person5.traditionally,curture contact consists of three forms.which is wrong below<A>A.acquisitionB.acculturationC.assimilationD.amalgamation Chapter 101.The interavtionist view holds that language as a result of the complex interplay between the___A__of a child and the __A__in which he grows .A: human chracteristics environmentB: chracteristics environmentC: language acquisition placeD: gift place2.The atypical language development includes__A___A: hearing impairment mental retardationB: autism stutteringC: aphasia dyslexia dysgraphiaD: Both A ,B and C3.Children's language learning is not complete by the time when they enter school at the age of _C__A: 3 or 4 B: 4 or 5C: 5 or 6 D: 6or 7Chapter 111.A distinction was made between ( ) and ( ).The former would facilitate target language learning,the later would interfere. < A >A positive transfer negative transferB negative transfer positive transferC contrastive analysis error analysisD error analysis contrastive analysis2.( ) are learners' consious,goal-oriented and problem-solving based efforts to cahieve desierable learning efficiency. < A >A Learning strategiesB Cognitive strategiesC Metacognitive strategiesD Affect strategiesnguage acquisition device(LAD) came from( ). < D >A John B.WatsonB B.F. SkinnerC S.D. KrashenD ChomskyChapter 121.____is the study of two related areas:language disorders and the relationship between the brainand language.A.neurolinguisticsB.linguisticsC.neuronsD.modern linguistics2.Psycholingusitics is the study of _____and mental activity associated with the use of languageA.psychobiologyB.psychological statesC.physical statesD.biological states3._____uses a narrow beam of X-ray to create brain images that the form of a series of brainslices.A.PETB.MRIC.CT scanningD.fMRI4.The brain is divided into two sections:the lower section called the____and the higher sectioncalled____.A.brain stem,cerebrumB.brain stem,neuronsC.cerebrum,brain stemD.cerebrum,neurons5.Damage to parts of the left cortex behind the central sulcus results in a type of aphasia called_____.A.Wernicke's aphasiaB.Broca'saphasiaC.Acquires dyslexiaD.fluent aphasia填空题第七章1.In addition to the borrowed affixes,some lexical forms become grammaticalized over time,this process is called ______________2.Generally speaking,there are mainly two possible ways of lexical changes: ________and ________,which often reflects the introduction of new objects and notions in social practices.3.New words may be coined from already existing words by "subtracting"an affix thought t be part of the old word ,such words are thus called____________.4.Over the time many words remain in use,but their meanings have changed,three mainly processes of semantic change,___________,____________, ____________.5.While the "_________"and "__________ "do seem to account for some linguistic changes,it may not be explanatory enough to account for other changes.KEYS:1.grammaticalization2.the addition and loss of words3.back-formation4.widening, narrowing, shift5.theory of least effort, economy of memory第八章1·-------is the sub-field of linguistics that studies the relation between language and society,between the uses of language and the social structures in which the users of language live. 答案Sociolinguistics 2·The social group that is singled out for any special study is called th e ----------.答案speech community3A------------is a linguistic variety used by people living in the same geographical region.答案regional dialect4he Ttype of language which is selected as appropriate to the type of situation is a---------.答案register5A-------is a special language variety thatmixes or blends languages ang it is used by people who speak different languages for restricted purposes such as trading.答案pidgin第九章1. anguage and culture,intrinsically interdependent on each other,have_through history (evolved together)2. ulture reflects a total way of life of a people in a_(community)3.in a word,_expreses culture reality (language)4.culture differences are also evident in the way_ and compliments are expressed (gratitude)nguage as the_of culture is tightly intertwined with culture (keystone)第十章1 ( ) refers to a child’s acquisition of his mother tongue.2 Generally speaking, there are mainly three different theories concerning how language is learned,namely the behaviorist,the interactionist ,( ) views.3 All child language acquisition theories talk about the roles of twofactors to different degrees the age ang ( ).4 Lexical contrast and ( ) theories are also proposed to explain how children acquire their vocabulary or lexicon.5 The atypical language development includes hearing impairment,mental retardation, autism,stuttering,( ),dyslexia,dysgraphia.答案：nguage acquisition2.the innatist3.the linguistic environment4.prototype5.aphasia第十一章1.（）refers to the systematic study of how one person acquiresa second language subsequent to his native language (NL or L1) .2.Contrastive analysis compares the ( ) cross these twolanguages to locate the mismatches or differences so that people can predict the possible learning difficulty learners may encounter .3.In addition, because of its association with an outdated modellanguage description (structuralism) and the increasingly discredited learning theory (behaviorism) , the once predominant contrastive analysis was gradually replaced by ( ).4.The interlingual errors mainly result from （）interferenceat different levels such as phonological , lexical , grammatical ordiscoursal , etc .5.Krashen assumed that there were two independent means or routesof second language learning : acquisition and （）。

热奈特叙事话语重点概念精简版目录Narrative Discourse (3)Chapter One order (6)Chapter Two Duration (10)Chapter Three Frequency (14)Chapter Four Mood (16)Chapter Five Voice (20)Narrative DiscourseBy Gerard Genette Pbed in 1980 Narratology denotes both the theory and the study of narrative and narrative structure and the ways that these affect our perception. As a matter of fact, this word is an anglicisation of French word narratologie, coined by Tzvetan Todorov (Grammaire du Décaméron, 1969). Since the 1960’s, the contemporary narrative theory has been rapidly developing towards maturity, in which French structuralist critic Gerard Genette plays a pivotal role. On the basis of absorbing the other s’ research results, he constructs his own narrative theory, whose origin mainly includes Saussure Linguistics, Structuralism, Russian Formalism, and New Criticism.Russian formalists argue that the literary characteristic is not what to write but how to write. Literary narrative mainly includes “sjuzhet” (plot) and “fa bula” (usually refers to story. Fabula and sjuzhet (also syuzhet, sujet, sjužet, or suzet) are terms originating in Ru ssian Formalism and employed in narratology that describe narrative construction. Sjuzhet is an employment of narrative and fabula is the order of retelling events. They were first used in this sense byVladimir Propp and Shklovsky.). And the plot determines the story. On the basis of this formalist concept, Propp places emphasis on form and structure of works in his Morphology of the Folktale (1928). But he only takes note of the syntactic relationship of the surface of the story. Later Gremas, Levi Strauss, and French narratologist Bremond carry out a series of comprehensive researches on the relationship between the surface and deep structures of the story, and sum up a wide variety of grammatical patterns of the story. Though large and vague, these narrative structure models provide a great reference for Genette. New Criticism representatives Brooks and Warren collaborate on Understanding Fiction(1979), put forward the question of “who speaks”, and hence draw forth the concept of “focus of narration”, which lays a solid foundation for Genette’s "focalization" theory. Moreover, as far as the study of the narrative formula, it is obvious that Genette is influenced by Wayne Booth's The Rhetoric of Fiction (1983). However, in terms of the narrative form, Booth has a similar view to New Critics.Accepting and absorbing the above-mentioned scholars’ advantages and strengths, Genette published Narrative Discourse in 1972, which makes Marcel Proust’s In search of Lost Time the research object and proposes his own unique narrative outlook. In the book, at first he indicates that narrative contains three distinct notions, namely, narrative, story and narrating, and further distinguishes them. Narrative refers to narrative discourse, which means “the narrative statement, the oral or written discourse that undertakes to tell of an event or a series of events” (Genette, 1980: 25). Story means an event or a series of events told in narrative discourse, real of fictitious. Narrating is the act of someone recounting something. To analyzing narrative discourse is, essentially,to study the relationship between narrative and story, between narrative and narrating, and between story and narrating.Excluding Introduction and Afterword, Narrative Discourse is divided into five chapters, which are Order, Frequency, Duration, Mood, and Voice in turn. In the five chapters, Genette at length analyzes the artistic techniques of In search of Lost Time, and hence summarizes and establishes a set of his own narratology. Genette incorporates French structuralist narrative theories, constructs rather comprehensive and systematic narrative theory, and thus lays a solid foundation for contemporary narratology. It is under the influence of his narrative discourse that many subsequent scholars and experts such as Miede Bal, Gerald Prince, and Rimmon-Kenan further explore and deeply dig the narrative theories. These scholars speak highly of his narrative discourse, and in the meantime put forward some doubts and challenges, in view of which Genette also published Nouveau discours du récit (new narrative discourse) in 1983 as a response. In this new narrative discourse, he discusses such questions as the classification of person, the application of the present tense, the interrelation between mood and voice, and focalization, and consequently interprets and perfects his narrative theory.In short, Genette presents a lot of concepts which has become the standard terms of classic in the narrative field. Besides, the publication of his Narrative Discourse has aroused strong reaction and sensation in the literary theory circle. According to his narrative theory, many analyze and interpret the specific works and bear great fruit.Chapter One orderTime is thought of as a uni-directional and irreversible flow, a sort of one-way street, just as Heraclitus said early in western history: “You cannot step twice into the same river, for other waters and yet other waters go ever flowing on.”However, as far as narrative activity is concerned, “the time of even the simplest story escapes the ordinary notion of time conceived as a series of instants succeeding one another along an abstract line oriented in a single direction” (Ricoeur, 1980: 169). Narrative is the art of TIME, which is the main subject that the majority of structuralist narratological works dwell on. In narratives, TIME can be defined as the relations of chronology between story and text, possessing the duality, namely, the time of the thing told and the time of the narrative. German theoreticians refers to this kind of temporal duality as the opposition between “erzählte Zeit” (story time) and Erzählzeit (narrative time). In Les catégories du récit littéraire, Todorov divides the narrative into three categories: tense, aspect, and mood. Here the tense means the relationship between the story time and the discourse time.In Narrative Discourse, Genette spends almost half of the book researching TIME( from p.33 to p.160). According to Genette, time can be viewed in three respects: order, duration and frequency, under which he sets out to examine the relations between the story time and the text time.Definition of OrderAccording to Genette, to “study the temporal order of a narrative is to compare the order in which events or temporal sections are arranged in the narrative discourse withthe order of succession these same events or temporal segments have in the story” (Genette, 1980: 35). Actually, the order is intended for exploring the relations between the story time and the narrative time. In Genette’s terms, the main types of discrepancy between them are called anachronies, which mainly include three types: analepsis, prolepsis, and achrony.An analepsis is “any evocation after the fact of an event that took place earlier than the point in the story where we are at any given moment” (Genette, 1980: 40). That is to say, it is a narration of story-event at a point in the text after later events have been told. This narration goes back to a past point in the story. A prolepsis is “any narrative maneuver that consists of narrating or evoking in advance an event that will take place later” (Genette, 40). It is a narration of story-event at a point before earlier events have been mentioned. The narration takes an excursion into the future of the story. In order to determine the anachrony, Genette introduces two concepts:reach and extent.The former refers to the temporal distance far from the “present”moment, when an anachrony appears, whether analeptic or proleptic. The latter means the duration of story covered by the anachrony. If reach and extent of an (or mainly isolated) event can not be clearly determined, the event is dateless and ageless. This kind of anachrony deprived of temporal connection is called an achrony.According to Genette, every anachrony is made up of a narrative that is temporally second, namely, “second narrative”. With respect to the anachrony, “the totality of the context can be taken as first narrative” (Genette, 1980: 49). Based on the differences between an analepsis and the first narrative in reach an extent, Genette classifies analepses into three types: external analepsis, internal analepsis and mixed analepsis.External analepsis means its “entire extent remains external to the extent of thefirst narrative” (Genette, 1980: 49). The second analepsis is, in Genette’s terms, internal analepsis, whose temporal departure and extent are within the first narrative. The third analepsis referred to by Genette is called mixed analepsis, whose “reach goes back to a point earlier and whose extent arrives at a point later than the beginning of the first narrative” (Genette, 1980: 49). In others words, if the period covered by the analepsis begins before the starting point of the first narrative but at a later stage either joints it or goes beyond it, then the analepsis is considered “mixed”. On the whole, analepses can add the narrative capacity in unit of time. It, more often than not, contains the rich and long train of thoughts and the diverse and confused past.world.Apart from analepses, the second common form of anachronies is prolepses, which can be defined as “any evocation after the fact of an event that took place earlier than the point in the story where we are at any given moment” (Genette, 1980: 40). Prolepses are a kind of anticipation or a hint at the future event. Like analepses, prolepses are also divided into external prolepses and internal prolepses.The limit of the temporal field of the first narrative is clearly marked by the last non-proleptic scene and some events take place after this scene. As opposed to external prolepses, internal prolepses can be designated as some episodes told earlier than the last non-proleptic scene of the story.AchronyIn such anachronies as analepses and prolepses, their reach and extent can essentially be confirmed. However, in the achrony, the events “express the narrative’s capacity to disengage its arrangement from all dependence on the chronological sequence of the story it tells.”(Genette, 1980: 84) From the context, readers cannot obtain any inference. It is an anachrony deprived of every temporal connection. Inachronies, the most outstanding way is “synchrony”, which brings the story time upon the same plane, blurs the linear relation of time and highlights the spatiality of the event. Genette insists, “Proleptic analepses and analeptic prolepses are so many complex anachroies, and they somewhat disturb our reassuring ideas about retrospection and anticipation…some events not provided with any temporal reference whatsoever, events that we cannot place at all in relation to the events surrounding them…they need only be attached not to some other event but to the (atemporal) commentarial discourse that accompanies them.” (Genette, 1980: 83)Another form of achronies is the atemporal commentarial discourse. Expounding the achronies in Marcel Proust’s In search of Lost Time, Genette points out that the narrative order of the story has no connection to the temporal order of the events, or only a partially coincidental connection. “The truth is that the narrator had the clearest of reasons for grouping together, in defiance of all chronology, events to be connected by spatial proximity, by climatic identity, or by thematic kinship; he thus made clear, more than anyone had done before him and better than they had, narrative’s capacity for temporal autonomy.” (Genette, 1980: 85)Chapter Two DurationDefinition of DurationDuration refers to the relations between the time the events are supposed to have taken to occur and the amount of the text devoted to their narration in the novel. According to Genette, there are totally four basic forms of narrative movements, which are two extremes, namely, ellipsis and descriptive pause, and two intermediaries: scene and summary. All these four narrative movements signify the narrative speed or pace. Genette proposes to employ constancy of speed to examine the degrees of duration. The maximum speed is ellipsis, where there is zero textual space corresponding to some story duration. On the other hand, the minimum speed is indicated as a descriptive pause, in which a certain segment of the text corresponds to zero story duration. In theory, there are infinite possibilities of speed between the two extremes while in practice all of them can be conventionally reduced to summary and scene. In summary, the speed is accelerated through a textual “condensation” or “compression” of a given story-period into a relatively short statement of its main features. The degree of condensation can, of course, vary from summary to summary, producing multiple degrees of acceleration. In scene, story duration is conventionally equal to text duration. The purest scenic form is dialogue, in which story time is identical to narrative pseudo-time. In addition, a detailed narration of an event can also be regarded as scenic.Genette schematizes the temporal values of such four movements as Pause, Scene, Summary and Ellipsis with the following formulas, with ST designating story time and NT the pseudo-time or conventional time, of the narrative:Ellipsis: NT=0, ST=n. Thus: NT＜∞STSummary: NT＜STScene: NT=STPause: NT=n, ST=0. Thus: NT∞＞ST(Genette,1980: 94-95)Here, ＞means longer than＜means shorter than∞ means infiniteEllipsisAn ellipsis can be called “omission”. Here, it refers to temporal ellipses. As far as temporality is concerned, “the analysis of ellipses comes down to considering the story time elided” (Genette, 1980: 106). The first question that should be known is whether duration is indicated or not. If duration is clearly indicated, it is definite ellipses. If not, it is indefinite ellipses. Generally speaking, from the formal point of view, according to Genette, ellipses can be distinguished as the following three types: Explicit ellipses, Implicit ellipses, and Hypothetical ellipses.Generally speaking,explicit ellipses mainly arise from two forms of the lapse of time. The former refers to the indication (definite or not) of the lapse of the time they elide, which, by and large, is equal to the quick summaries to be mentioned hereinafter of the “some years passed” type. Implicit ellipses are “those whose very presence is not announced in the text and which readers can infer only from some chronological void or gap in narrative continuity”(Genette, 1980: 108). Hypothetical ellipsis is the most implicit form of ellipsis, which is “impossible to localize, even sometimes impossible to place in any spot at all, and revealed after the event by an analepsis”(Genette, 1980: 109).SummaryAmong the four basic forms of narrative movements, summary is only second to ellipsis in narrative rhythms. It is a “narration in a few paragraphs or a few pages of several days, months, or years of existence, without details of action or speech” (Genette, 1980: 95-96). In the tradition of the novel, insignificant events which do not greatly influence the course of the plot are quickly summarized, while the turning points or the dramatic climaxes which have a strong influence on the course of the plot are presented extensively in scenes. Generally speaking, summary is the most usual transition between two scenes, the basic background which makes scenes stand out, and thus the most excellent connective tissue of novelistic narrative. The narrative speed is accelerated through a textual “condensation” or “compression” of a story time into a relatively short statement. The degree of compression can, of course, vary from summary to summary, which will contribute to different degrees of acceleration.In addition to accelerating the narrative, summary is also a suitable instrument for presenting background information, or for connecting various scenes. Another function of summary is to connect different scenes.SceneIn a scene, story time and narrative time are conventionally considered identical. In this situation, an event is described in detail, almost in extenso. Thus it gives us a feeling of participating in the scene in person. As mentioned in the last section, the fundamental rhythm of the novelistic narrative is defined by the alternation of summary and scene. Obviously, in so doing, the aim is neither to overtire readers with too rapid a tempo nor to bore them with one that was too slow. According to Genette, the purest scenic form is dialogue, which basically realizes the equality of time between story and narrative. Of course, a detailed description of an event is also considered scenic. Thus,what characterizes a scene is the quantity of narrative information and the relative effacement of the narrator.PauseAs mentioned above, compared to ellipses, pause is the minimum speed in these two extremes of the narrative movements. In a pause, some segment of the text corresponds to zero story time while the narrative time can infinitely go on. There appear two kinds of pauses: the descriptive pause and the commentary pause.Chapter Three FrequencyDefinition of FrequencyNarrative frequency is one of the main aspects of narrative temporality. It refers to, according to Genette, “the relations of frequency (or, more simply, of repeti tion) between the narrative and the diegesis”(Genette, 1980:113). An event is not only capable of happening; it can also happen again, or be repeated. The “repetition” is in fact a mental construction that is attained by the elimination of the peculiar aspects of each occurrence itself, and a preservation of only those qualities shared by all the others of the same class. From the multiplication of the two possibilities given on both sides: the event repeated or not, the statement repeated or not, we can reduce the system of relationships between the narrated events of the story and the narrative statements of the text to three virtual types: singulative narrative, repeating narrative and iterative narrative.Singulative NarrativeSingulative narrative is far and away the most ordinary and normal form in the three kinds of frequency, and it means “narrating once what happened once (or, if we want to abbreviate with a pseudo-mathematical formula: 1N/1S)” or “narrating n times what happened n times (nN/nS)” (Ge nette, 1980: 114-115).Repeating NarrativeIn repeating narrative, the recurrences of the statement do not correspond to any recurrence of events, and it refers to “narrating n times what happened once” (Genette, 1980: 115). On the one hand, the same event can be told several times with stylisticvariations; on the other hand, it can also repeat with variation in “point of view”.Iterative NarrativeIterative narrative is a type of narrative, where a single narrative utterance takes upon itself several occ urrences together within the same event, and it means “narrating one time (or rather: at one time) what happened n times (1N/nS)” (Genette, 1980:116). Every iterative narrative is a synthetic narrating of the events that occur and reoccur in the course of an iterative series that is composed of a certain number of singular units (Genette, 1980:127).Chapter Four MoodDefinition of MoodThe term “mood” invokes a grammatical category in a metaphorical way. To be more strictly, it categorizes verb forms into indicative, imperative, interrogative and subjunctive according to whether they state a fact, give an order, tell a possibility or express a wish. Metaphorically, Genette defines mood from “degrees of affirmation” and “the different points of view from which the life or action is looked at” (Genette, 1980:161). Narrative mood aims at the capability of telling things from whose point of view. Narrative representation, or narrative information, has its degrees according to Genette. He believes that “the narrat ive can furnish the reader with more or fewer details, and in a more or less direct way, and can thus seem (to adopt a common and convenient spatial metaphor, which is not to be taken literally) to keep at a greater or lesser distance from what it tells” (Genette, 1980:162). The narrative can also choose to regulate the information it delivers according to the capacities of knowledge of one of another participant in the story, with the narrative adopting or seemingly adopting the participant’s vision or poi nt of view. Thus narrative mood involves two aspects, in which way the narrative information is regulated and the different degrees of its provision. The word “mood” in French is the same as the word for “music”. Genette said that narrative mood is depende nt on the “distance” and “perspective” of the narrator, and like music, narrative mood has predominant patterns.MoodDistanceBefore we come to the analysis of distance in this novel, it is necessary to have a look at when the term of “distance” was addr essed for the first time and its essential meaning. In Book III of The Republic, Plato contrasts two narrative modes. They include whether the poet “himself is the speaker and does not even attempt to suggest to us that anyone but himself is speaking”, which Plato calls pure narrative, or the poet “delivers a speech as if he were someone else”, which is called imitation or mimesis(Genette, 1980:162). Indirection and condensation are two distinctive features of pure narrative, and it is more distant than imitation: it says less, and in a more mediated way. At the end of the nineteenth century and the beginning of the twentieth in the United States and England, the contrast surged forth in novel theory, in the barely transposed terms of showing vs. telling. Showing in a novel, with little or no narratorial meditation, allows readers to experience the story through a character’s action, words, thoughts, senses and feeling, while in a telling mode, the narrator controls the action, the characterization and the point-of-view arrangement as well as readers instead through the narrator’s exposition, summarization and description. To study distance, there are other two terms that should be distinguished: narrative of events and narrative of words.Narrative of events is a narrative in which the nonverbal is transcribed into the verbal. It involves mimesis and diegesis (or we can say “showing”and “telling”). The mimetic factors come down to those two data: the large quantity of narrative information and the effacement of the narrator. Generally speaking, mimesis leaves no trace of the narrator telling. On the other hand, diegesis can be defined by a minimum of information and a maximum of the informer. That is, the quantity of information and the presence of the informer are in inverse ratio. Actually, the verbal “imitation”of nonverbal events is just an illusion of mimesis, not a real mimesis. It is not realistic that in the narrative of events there is no interference of the narrator or no traces of thenarrator telling.Narrative of words involves the distance between the narrator and the character s’speech. According to Genette, it can be divided into three types: narrated speech, transposed speech, and reported speech.Among them, narrated speech is the most distant, and the most reduced; transposed speech is more mimetic and capable of exhaustiveness; the most “mimetic” form is reported speech.PerspectiveGenette defines narrative perspective, after distance, “the second mode of regulating information” (Genette,1980:185) in narrative fictions. It is also called “focalization”. According to Mieke Bal, focalization refers to “the elements presented and the vision through which they are presented” (Bal, 1997:100). Focalization, then, becomes a matter of whose point of view orients the narrative perspective. Since the end of the nineteenth century, among all the questions related to narrative techniques, “point of view” is the one most frequently studied, with indisputable critical results. Genette’s replacement of i t to the slightly more abstract term “focalization” makes an attempt to avoid the specifical visual connotations of the terms “vision”, “field” and “point of view”.Each narrative has both a narrator and a focalizer. A narrator is the speaker of the narrative, the agent who establishes communication with a narratee and decides what and how the story will be told. The narrator tells what the focalizer sees. A focalizer is the agent whose point of view orients the narrative text. A text is anchored on the point of view of a focalizer when it presents (and does not transcend) the thoughts, reflections and knowledge of the focalizer, his or her actual and imaginary perceptions, as well as his or her cultural and ideological orientation. It goes without saying that the narrator and the focalizer do not always overlap within the same person or narrative agent, andhence the different kinds of narrative situations. Genette divides focalization into three kinds: zero focalization, external focalization and internal focalization.Zero focalization has the same features with the traditional omniscient point of view, where the narrator knows more than the character, or to be more exact says more than any of the characters knows. It can be symbolized by the formula: Narrator＞Character by Todorov.Internal focalization takes place when events or thoughts are mediated through the point of view of the focalizer, and it can be symbolized by Narrato r＝Character, signifying that the narrator says only what a given character knows. It is always employed by modern narrators to see an event or to experience a feeling from a character’s perspective, emphasizing the description of the thoughts, feelings of characters, as well as analysis and interpretation of their actions. This narrative type includes three sub-types: fixed internal focalization, referring to that the presentation of narrative facts and events from the constant point of view of a single focalizer, variable internal focalization, which means the presentation of different episodes of the story as seen through the eyes of several focalizers, and multiple internal focalization, a technique of presentation of an episode repeatedly, each time through the eyes of a different focalizer.External focalization occurs when the narrator presents the aspects of the story using solely observable, external information, and it denotes a focalization that is limited to what the observer could actually have observed from the outside. This focalization has the narrator focusing on some visible and external aspects of the events and characters in the narrative, and the narrator merely relates physically ascertainable facts to the reader. It could be formulated as: Narrator＜Character.Chapter Five VoiceDefinition of VoiceThe term “voice” easily reminds readers of its grammatical definition that refers to a verb either active or passive. More generally, it indicates the relation of the subject of the verb to the action which the verb expresses. On the narrative level, the subject is not only the person who carries out or submits to the action, but also the person (the same one or another) who reports it, and, if needed, all those people who participate, even though passively, in this narrating activity (Genette, 1980:213). In narratology, the basic question involving voice is “Who speaks?” It means that who narrates this story. Voice has a great deal to do with characterization and consciousness of the narrator or narrators. According to Genette, narrative voice concentrates on the study of the narrating instance from three aspects: time of narrating, narrative levels and person.VoiceTime of NarratingTime of narrating in a text can’t be avoided, since a writer must necessarily tell the story in a tense, no matter whether it is present, past, or future, and it can keep various temporal relations with the events of the story. The narrator is always in a specific temporal position relative to the story he or she is telling. From the point of view of temporal position, Genette describes four types of narrating.The most frequently used one far and away is called subsequent narrating, which is the classical position of the past-tense narrative, telling readers the events after they happen; the second type is a kind of predictive narrating, which is called prior narrating。

《社会语言学视角下《私人订制》幽默语言的多模态话语分析》摘要：电影语言是在特定的时间、空间内发生的源于真实语境的话语交际行为，是研究多模态话语分析最适合的“温床”，影片中巧妙地把娜破仑的名言进行换序，偏离了话语的正常“轨道”，让观众心理期待“扑空”，从而增添了影片的幽默情趣，在电影《私人订制》中，就有如此用例电影语言是在特定的时间、空间内发生的源于真实或预想真实语境的话语交际行为的展示方式，是研究多模态话语分析最适合的“温床”。

“社会语言学”起源于20世纪60年代的美国，原名为sociolinguistics。

显而易见，它是“社会学”（ sociology）和“语言学”（linguistics）组合而成的一门交叉学科。

语言本身就具有社会属性，把语言再回归到社会生活中从社会的角度，用社会学的方法去研究其使用、发展和演变，这是社会语言学学科本质。

多模态话语分析兴起于20世纪末，利用媒体把媒介符号引入到语言学研究领域，把语言和言语、面部表情和身势语、语音识别和语音综合等多种模态结合。

它的出现有效地弥补了单模态话语分析的不足，成为了现今语言学研究领域的一大热门。

电影语言是在特定的时间、空间内发生的源于真实语境的话语交际行为，是研究多模态话语分析最适合的“温床”。

冯小刚导演的喜剧电影一直是“票房良药”。

与电影学、文学领域相比，电影语言学角度的研究成果则非常少，而且大都是从语言基本层面的语音、词汇、语法角度做静态分析，语言深层结构方面的分析并不多。

一、多模态话语形式分析（一）语音层面语音是语言的物质外衣，也是话语交际的重要手段。

利用方言语调来获取幽默效果，是《私人订制》电影的一大特色。

作为主流文化的普通话一直是电影语言的主体，使用“原生态方言”一方面能够凸显人物的典型性性格特征，另一方面能够借助演员的个性表演来彰显主题的喜剧性效果。

比如在电影第一幕《性本善》中扮演“领导老乡”的小璐说着一口“洋泾浜”的陕西话：小璐：ni（你）可bo（別）怪三婶儿she（说）ni（你）。

江苏省南京市第一中学2024届高三上学期暑期阶段性测试英语试题一、阅读理解1．Which unique technological feature of the Apple Vision Pro enables an unprecedented mixed reality experience for users?A．Eye-tracking system.B．Dual-chip design.C．Optical sensors.D．High-definition display.2．Which of the following statements accurately compares the capabilities of the four products mentioned?A．Apple Vision Pro is the only one with a dualchip design.B．Meta Quest 3 offers the highest graphics processing performance.C．PS VR2 detects eye movements for online interaction.D．PICO 4 lacks a high-precision tracking system.3．If Simon wanted to buy a product to play a game, which of the following products would be his first choice?A．Apple Vision Pro.B．Meta Quest 3.C．PS VR2.D．PICO 4.“Digital art with iPad and Apple Pencil has helped to expand my creative thinking.”Apichaya "Bim" Wannakit is starting her fourth year in painting, sculpture, and graphic art at Silpakorn University in Bangkok, in the heart of the modern Thai art scene. Growing up in Northeast Thailands creative community in Ubon Ratchatani, Bim developed her passion for art at a young age.She took inspiration from anime like "Dragon Ball Z" cartoons she saw on television and the "Manga" comic books she collected. She put years of hands-on practice into her illustrations using brushes, watercolors, and oil paints. In high school, she discovered even more ways to express herself and expand her creative palette when she started using iPad with Apple Pencil.“I quickly adapted to the precision and flexibility of digital painting using Apple Pencil,” Bim says. “It brought my entire art kit into one magical tool, without sacrificing the natural feel of traditional mediums. It allows me a new way to bring my ideas to life.”When it was time to get ready for university, Bim used iPad to bring her art portfolio together, capturing and storing all her mixed media digitally in iCloud. Bim says, “I initially chose Apple technology because of its cloud service for file management and simple-to-learn design interface — iPad has made my life easier.”Today she uses Procreate on iPad with Apple Pencil for all her creative workflows. Features such as QuickShape and StreamLine with Apple Pencil enable her to quickly add layers, outlines, colors, and shadows in an immersive and playful way. “My love for traditional painting and digital art brings me a new sense of balance. My Apple devices help to make this possible from anywhere.”iPad also supports her daily academic work. Bim uses Apple Pencil with GoodNotes for note-taking, and Microsoft Word on iPad for creating art descriptions. “In class, I use iPad Pro to present my work,” Bim says. “And the 12.9-inch screen is large enough for my professor to view my art comfortably.”“Through my art, I want to tell a story that inspires others and creates a space for experimental ideas.”4．What does Bim Wannakit value most about using Apple devices for her art?A．The cost-effectiveness of the devices.B．The ability to create a balance between traditional and digital art.C．The simplicity of the design interface.D．The ease of file management through cloud services.5．Why did Bim choose to use iPad and Apple Pencil for her artistic work?A．They replicated the feel of traditional art materials exactly.B．They provided a cheaper alternative to professional art tools.C．They allowed her to combine digital flexibility with a natural drawing feel.D．They were the only tools available to her at the time.6．According to the article, how has Bim's artistic practice been enhanced by digital tools?A．She can now create larger artworks than before.B．She has access to a wider range of colors and textures.C．She can share her work instantly with a global audience.D．She can experiment with new techniques and styles more easily.7．Bim mentioned that GoodNotes and Microsoft Word have been helpful in her academic work. What are the main uses of these tools?A．Creating detailed sketches for her art projects.B．Organizing her research and writing academic papers.C．Editing photos for her online art gallery.D．Designing promotional materials for art exhibitions.We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios. We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of high fidelity video. Our results suggest that scaling video generation models is a promising path towards building general purpose simulators of the physical world.We train a network that reduces the dimensionality of visual data. This network takes raw video as input and outputs a latest representation that is compressed both temporally and spatially. Sora is trained on and subsequently generates videos within this compressed latent space. We also train a corresponding decoder model that maps generated latents back to pixel space.All of the results above and in our landing page show text-to-video samples. But Sora can also be prompted with other inputs, such as pre-existing images or video. This capability enables Sora to perform a wide range of image and video editing tasks—creating perfectly looping video, animating static images, extending videos forwards or backwards in time, etc.Sora is also capable of generating images. We do this by arranging patches of Gaussian noise in a spatial grid with a temporal extent of one frame. The model can generate images of variable sizes—up to 2048*2048 resolution.We find that video models exhibit a number of interesting emergent capabilities when trained at scale. These capabilities enable Sora to simulate some aspects of people, animals and environments from the physical world. These properties emerge without any explicit inductivebiases for 3D, objects, etc.—they are purely phenomena of scale.We believe the capabilities Sora has today demonstrate that continued scaling of video models is a promising path towards the development of capable simulators of the physical and digital world, and the objects, animals and people that live within them.8．What is the meaning of the underlined words in the text?A．the original raw video data.B．a compressed version of the video data.C．the process of reducing video quality.D．the spatial and temporal dimensions of a video.9．What is the main content of Paragraph 3?A．Sora's ability to generate high-resolution images.B．The process of training Sora on compressed latent space.C．Sora's various applications in image and video editing.D．The emergence of interesting capabilities in video models.10．Which of the following best describes the overall goal of the research described in the passage?A．To create realistic images and videos using only text prompts.B．To develop a general-purpose simulator capable of simulating various aspects of thephysical world.C．To train a network that can compress video data without losing quality.D．To explore the potential of transformer architectures in video and image generation tasks. 11．What is the author's attitude towards the future development of Sora?A．Skeptical.B．Optimistic.C．Neutral.D．Uncertain.Electric vehicles (EVs) are a strong weapon in the world's efforts against global warming. But the effects of EVs depend on what country you are in. In some nations, electric vehicles lead to the release of more carbon gasses than gasoline cars, new research shows.The Radiant Energy Group (REG) compared gas emissions caused by a gasoline vehicle and from charging an electric vehicle. The study compared the emissions caused by charging a Tesla Model 3 to drive 100 kilometers with the emissions coming from an average gasoline cardriven the same distance.Countries where charging an electric vehicle is cleaner than driving a gasoline-powered car use a lot of hydroelectric, nuclear or solar power.Sales of electric cars are rising the fastest in Europe. Data from REG suggests that EVs in Poland and Kosovo actually create more carbon emissions because their electric systems depend so much on coal.In other European countries, however, EVs result in reduced emissions. The carbon gas reduction depends on what energy supplies electricity systems and the time of day vehicles are charged.The countries with the biggest carbon gas savings from EVs use a lot of nuclear and hydroelectric power. An EV driver in Germany reduces greenhouse gas emissions by 55 percent over a gasoline car. Germany uses a mix of renewable energy and coal to produce electricity.Germany and Spain create a lot of electricity from the sun and wind. But the sun and wind do not add to a country's electricity system equally throughout the day.For this reason, the amount of carbon emissions saved by driving an EV depends on the time of day it is being charged. Charging in the afternoon, when there is more sun and wind, saves 16 to 18 percent more carbon than at night when electricity systems are more likely to be using gas or coal.Automakers including General Motors, Stellantis and V olkswagen have set targets to sell mainly electric vehicles in Europe in the coming years. U.S. car manufacturer General Motors said it will have all new electric cars by 2022.12．Which of the following statements is NOT true according to the passage?A．The amount of carbon emissions saved by EVs depends on the source of electricity used for charging.B．Charging EVs during daylight hours with renewable energy sources can cause more carbon savings.C．The time of day when EVs are charged can significantly affect their carbon footprint.D．General Motors plans to sell only gasoline-powered cars by 2022 in the United States. 13．What can be inferred from the fact that car manufacturers like General Motors, Stellantis, and V olkswagen have set targets to sell mainly electric vehicles in Europe?A．The demand for EVs in Europe is expected to decrease in the near future.B．These manufacturers believe that EVs will become the norm in Europe in the coming years.C．Europe has banned the sale of gasoline-powered cars entirely.D．These manufacturers are not confident in the long-term viability of EVs.14．Based on the information in the passage, which of the following is a potential challenge for the widespread adoption of EVs?A．The limited range of EVs compared to gasoline-powered cars.B．The high initial cost of EVs compared to traditional vehicles.C．The inconsistency of renewable energy sources for EV charging.D．The lack of charging stations in rural areas.15．Which of the following would be the most suitable title for the passage?A．The Environmental Impact of Electric Vehicle Charging.B．The Global Shift to Electric Vehicle Adoption.C．The Economics of Electric Vehicle Ownership.D．The Future of Renewable Energy in Automobiles.In the rapidly advancing world of today, the concept of lifelong learning has become increasingly relevant. 16 Rather, it is essential to continuously adapt and learn throughout one's life. This essay aims to explore the significance of lifelong learning in the modern era, discussing its impact on personal growth, career development, and societal progress.17 As technology and knowledge continue to evolve, it is important to stay updated with new information and ideas. By engaging in lifelong learning, individuals can enhance their cognitive abilities, broaden their horizons, and develop a growth mindset. This, in turn, leads to a more fulfilling and enriching life experience.Lifelong learning is essential for career development. In the modern job market, employers increasingly value candidates who possess the ability to adapt to change and learn new skills quickly. Lifelong learners are more likely to stay relevant and competitive in their field, as they are constantly improving their knowledge and skills. 18Moreover, lifelong learning contributes significantly to societal progress. As individualscontinuously acquire new knowledge and skills, they are able to contribute more effectively to their communities and countries. 19 Lifelong learning also fosters a culture of inclusivity and diversity, as it encourages individuals from different backgrounds and experiences to learn and grow together.In conclusion, lifelong learning is an indispensable aspect of life in the modern era. 20 By committing to lifelong learning, individuals can stay updated with new knowledge and ideas, enhance their cognitive abilities, and contribute more effectively to their communities and countries. Therefore, it is important for us to cultivate a habit of lifelong learning and encourage it among our peers and future generations.A．Additionally, they demonstrate a commitment to personal and professional development. B．With the development of science and technology, humans will continue to be replaced by AI. C．Lifelong learning is crucial for personal growth.D．This can lead to innovations, economic growth, and cultural richness.E．It is essential for personal growth, career development, and societal progress. F．Therefore, only lifelong learning, in order to maintain the leading position.G．It is no longer sufficient to acquire knowledge and skills during one's formal education.二、完形填空Once upon a time, there was a small village hidden deep in the mountains. The villagers led a simple life, depending mostly on farming for their 21 . One day, a young man named Jack arrived in the village. He was 22 and eager to explore the surrounding mountains.Jack quickly 23 with the villagers and became a part of their community. He was fascinated by the 24 tales of the mountains and the mysterious creatures that were said to inhabit them. Determined to find out the truth, Jack decided to embark on a journey of 25 .Before leaving, the villagers warned him of the dangers that awaited him in the mountains. They told him about the 26 weather, the treacherous paths, and the wild animals that roamed the area. But Jack was not 27 . He packed his belongings, bid farewell to the villagers, and started his journey.As he climbed higher and higher, the weather became increasingly 28 . The pathswere slippery and difficult to navigate. But Jack persevered, determined to reach the top of the mountain.After days of 29 , Jack finally reached the summit. The view was breathtaking, with rolling hills and dense forests stretching out as far as the eye could see. But as he was taking in the 30 , he heard a strange sound coming from the bushes nearby.Curious, Jack approached the bushes and peeked inside. To his surprise, he saw a small, 31 creature with big eyes and furry ears. The creature seemed 32 at first, but then it slowly approached Jack, sniffing the air curiously.Jack realized that this was one of the mysterious creatures the villagers had talked about. He felt a sense of 33 and excitement wash over him. He gently reached out his hand and the creature cautiously sniffed it, 34 allowing Jack to pet it.From that day on, Jack and the creature became close friends. They explored the mountains together, 35 each other's company. The villagers were amazed when Jack returned with his newfound friend and shared his adventures with them.The experience taught Jack the value of courage, perseverance, and friendship. It also showed him that there was more to the world than what he had imagined, and that there was always something new and exciting to discover.21．A．income B．entertainment C．inspiration D．experience 22．A．adventurous B．cautious C．lazy D．timid 23．A．argued B．competed C．bonded D．disagreed 24．A．boring B．ordinary C．amazing D．strange 25．A．discovery B．escape C．adventure D．research 26．A．unpredictable B．pleasant C．calm D．mild 27．A．discouraged B．frightened C．delighted D．satisfied 28．A．severe B．comfortable C．warm D．changeable 29．A．hiking B．struggling C．resting D．searching 30．A．scene B．creature C．challenge D．danger 31．A．fierce B．ugly C．cute D．powerful 32．A．aggressive B．nervous C．confident D．happy 33．A．wonder B．relief C．pride D．pity34．A．eventually B．immediately C．suddenly D．frequently 35．A．seeking B．enjoying C．avoiding D．tolerating三、语法填空阅读下面短文，在空白处填入1个适当的单词或括号内单词的正确形式。