| The present study investigated some syntactic aspects of the main form of idiomatic expressions in Chinese,Chengyu(CY).CYs are meaning-concise lexical phrases with stereotyped forms.Most CYs are joined structures of two two-character words(referred to as two-word CYs)or four one-character words(referred to as four-word CYs).A two-word CY forms a subject–verb unit(SV),a verb–object unit(VO),a structure of modification(SM),or a verb–verb unit(VV).Both the first and the second two one-character word in most four-word CYs make up an SV,a VO,or an SM.Just as Van Lancker Sidtis(2006)claimed that the meaning of idioms typically involves a conflation of social,political,contextual,attitudinal and emotional factors,CYs are also linguistic constructions that have gone through a sociolinguistic process of conventionalization.Thus the force of convention makes the established idiomatic meanings unavailable if merely comprehended by their literal meaning or syntactic rules.People must see through the literal meaning to get what a CY really means(Liu & Gao,1990).The unique external form and protean internal semantic relations of Chinese idioms make the study of comprehension of them a heated issue in the field of linguistics.1.RationaleWestern psycholinguists probe into the idiomatic representation models since a long time ago.The representation models of idioms can be grouped into three types,namely,non-compositional,compositional and hybrid views.Non-compositional approaches assume that idioms are stored in the lexicon and retrieved from it as whole long words(Bobrow & Bell,1973;Swinney & Cutler,1979;Weinreich,1969).Like words,idioms appear to be arbitrary learned mappings between linguistic form and meaning.The logic is that idioms are stored and accessed as whole units,and that this sort of direct access is computationally less expensive than the process of accessing and integrating the meanings of individual words into a structure.These notions have been opposed to explain the influence of transparency of idioms on their comprehension.However,they failed to construe why the meaning of lemma in idioms can be automatically activated,and why some idioms have syntactic flexibility.More recent work has shed doubt upon the whole word view.Investigations into idiom meaning have challenged the assumptions that all idioms are compositionally opaque and that the words comprising an idiomatic expression bear no relation to its meaning(Gibbs,Bogdanovich,Sykes & Barr,1997;Gibbs,Nayak & Cutting,1989).Structural investigations have demonstrated that individuals are aware of the grammatical properties of idiom-internal words(Peterson,Burgess,Dell & Eberhard,2001),and that idioms behave like structures in many ways,such as exhibiting structural priming(Konopka & Bock,2009).Compositional models oppose the theory of long words and assume that individual idiom components contribute to the overall sense of an idiom(Gibbs et al.,1989;Cacciiari & Tabossi,1988).Nevertheless,they were not able to explain how lemmas help to comprehend the opaque idioms,and what if those idioms contain lemmas without literal meaning.Presently,the hybrid model is accepted by most linguists.These results mentioned above support a hybrid representation for idiomatic expressions that preserves their structural properties while still accounting for their more word-like meaning(Cacciari & Tabossi,1988;Cutting & Bock,1997;Sprenger,Levelt & Kempen,2006).Hybrid models as a combination of non-compositional and compositional models reflect unitary features of syntactic and lexical-conceptual nodes in the mental lexicon(Giora & Fein,1999;Cutting & Bock,1997).However,they all converge upon the notion that literal processing plays a causal role in the access of idiomatic meaning.They were unable to illuminate whether the lemma processing and whole long words processing take place at homogeneous level.Cacciari and Tabossi(1988),drawing evidence from idiom comprehension,proposed the Configuration Hypothesis.Under this model the parser proceeds with literal interpretation until it has incrementally accumulated sufficient evidence that the string in question is idiomatic.At this point the idiomatic meaning is retrieved.Crucially,how far down the literal parse the processor goes is dependent upon the degree to which the idiomatic string can be plausibly taken literally,and how well the parser can predict the intended meaning during on-line comprehension.They refer to this tipping point as the idiom key(Tabossi & Zardon,1993).Specifically speaking,the Configuration Model assumes that a word combination(a potential idiomatic expression)is initially processed literally until a configuration “key” is recognized,and the idiomatic meaning is activated.Subsequently,literal and figurative(idiomatic)processing run in parallel until the literal sense is definitely rejected and the idiomatic one is accepted as the intended interpretation(Cacciary & Tabossi,1988;Cacciari & Glucksberg,1991).Nevertheless,the model is not committed to the idea that idioms are stored as lexical items.In a similar vein,Cutting and Bock(1997)proposed a distributed representation for idioms based on speech error data.In a series of experiments,they found that incidence of speech errors increased as a function of overlap in meaning or structure regardless of whether the structure in question was idiomatic.They proposed that idioms are represented as structural phrasal frames directly associated with conceptual meaning.The problem of Cutting and Bock's Conceptual Metaphor Model is that an abstract syntactic structure cannot recognize the relationship between concepts and individual active lemmas,which results in troubles in the production system inability to identify the speaker's intention.This is particularly very probable in cases in which a potential idiomatic expression contains open slots for more than just one NP or VP leaving it unclear how the system decides where particular NPs and VPs shall be inserted.Nonetheless,the hybrid representation model did not specify the understanding relationship between syntax and semantics of idioms.Sprenger et al.(2006)found that idiom production was facilitated by priming phonological and semantic associates of the idiom's literal component words,providing further evidence in favor of this view.The Superlemma Model(Sprenger et al.,2006)is based on Levelt's theory of lexical access in speech production(Levelt,1989).The model posits that an idiom has its unitary idiomatic concept that activates individual lemmas it is composed of,but the lemmas are not exclusively bound to one idiomatic meaning.They proposed that rather than phrasal frames,idioms are represented as superlemmas,which operate as a grammatical function over the component lemmas of the idiom.These superlemmas provide a representation for idioms,and potentially other multi-word collocations,which can account for the idiosyncratic nature of idiom meaning and their varying degrees of structural flexibility while allowing for normal lexical competition between idioms and normal lemmas(e.g.,kick the bucket versus die)with minimal storage redundancy(Kuiper,Egmond,Kempen & Sprenger,2007).The idiomatic expression is represented in the lexicon and activated by a superlemma that relates to a specific lexical concept which in turn activates the single lemmas composing the superlemma(Sprenger et al.,2006).This model of idiom processing simultaneously unites the features of idiom production as well as comprehension,and is an extension of the hybrid model(Cutting and Bock,1997).In comparison to Cutting and Bock's theory of lexical concept nodes,it avoids the troublesome explanations for the existence of many syntactic idiosyncrasies present in many idioms since the superlemma theory suggests that “the syntactic relationships and idiosyncratic constraints that characterize an idiom are directly applied to the lemmas involved;no additional operation is required(Kuiper et al.,2007).This model proposes that lemma processing and long word processing run in parallel.There exist competition,activation and inhibition between lemma and long word processing,that invoking greater impact on idiom comprehension will be activated earlier.The superlemma hypothesis provides a detailed and coherent model for idiom representation and representational machinery to cope with the heterogeneous nature of the idiom class.However,the precise representational content of these superlemmas is the subject of debate(Tabossi,Wolf,& Koterle,2009).In addition,applying this model to idiom comprehension is not trivial.In production,activation spreads from the conceptual layer down to the individual component lemmas.Thus for an idiom such as kick the bucket,the superlemma representation will enter normal competition with other semantically associated lemmas and superlemmas(e.g.,die,pass away).Once selected,activation will then spread to the individual component lemmas(e.g.,kick and bucket)with the superlemma providing precomputed phrasal configurations for the string.During comprehension,activation spreads upward from the component lemmas to the conceptual level.The role of the superlemma in this process is less clear.One possibility is that the superlemma representation acts as a sort of gate to the conceptual layer.Thus during on-line processing,encountering a syntactic environment incompatible with an idiomatic reading would preclude activation spreading to the conceptual layer.Under this view,strings such as “????” should not result in consideration of idiomatic meaning,as the superlemma representation would not contain an appropriate function that maps the component lemmas onto an incompatible structure and meaning.Another possibility is that superlemma as participate in spreading activation during comprehension in the same way as other lemmas.Thus,any partial activation of the superlemma representation will necessarily spread to the conceptual layer.If this is the case we would predict that an incompatible syntax would not be sufficient to prevent some consideration of the idiomatic interpretation during comprehension.Unlike syntactic incompatibility,however,semantic bias,for example in the form of semantically driven expectations to expect an idiom or literal sense,is expected to have different effects on how comprehenders interpret potentially idiomatic strings.During comprehension,the superlemma hypothesis predicts that some activation of the literal component lemmas is obligatory for the superlemma to become active and for the idiomatic meaning to be accessed.Thus we predict that bias that encourages an idiomatic interpretation should still result in some consideration of the literal meaning,owing to its obligatory status in the system.According to Cutting and Bock(1997),an idiom has its own representation of the lexical concept as a whole,the activation of which spreads to the representations for the constituent words at the lexical–syntactic(lemma)level.However,it violates the notion that idiom comprehension is not different from phrasal comprehension.Sprenger et al.(2006)revised the hybrid model by superlemma theory.Sprenger et al.(2006)extended this theory,arguing that as separate representations for the syntactic properties of idioms,superlemmas are connected to the idioms' building blocks.However,very few studies on how the lemmas with distinct properties such as meaning,part of speech,syntax and so on influence the representation of idioms have been conducted.Consequently,we are not clear about the precise function of superlemma in the representation of idioms,when it is processed from literal meaning to figurative meaning.In our study the paper posits that the hybrid view,especially superlemma model,is more precise to account for how human mind deals with linguistic phenomena involved in syntactic aspects and is adopted as the general model to comprehend CYs.We predict that participants will comprehend CYs in the hybrid model and the superlemma model.Our results will be interpreted in favor of models of idiom processing that propose obligatory literal processing.2.Research aimsThis present research aims at investigating how the morpheme of CYs with different syntactic structures impacts the descriptive norms and representation of CYs.Very few researches have been done in this respect so far.In psycholinguistics,a large number of studies have been done to investigate how idiomatic expressions are comprehended(Caillies & Butcher 2007;Declercq & Lendvai,2011)and processed(Conner,Hyun,O'Connor Wells,Anema,Goral,Monéreau-Merry,...& Obler,2011;Sprenger et al.,2006)with different tasks.The neural correlates of idiom processing have also been investigated(Hillert & Bura?as,2009).However,only a few studies have been dedicated to the collection of norms for idiomatic expressions.Tabossi et al.(2011)reported descriptive normative measures for 245 Italian verbal idiomatic expressions.Caillies(2009)reported norms for 300 English idioms with off-line experiments.Maya R.Libben and Debra A.Titone(2008)studied the multidetermined nature of English idiom processing.Bonin et al.(2013)reported psycholinguistic norms for 305 French idioms.But these researches merely pertained to some dimensions of idiom norms.No research has been done concerning the descriptive norms of Chinese idioms so far.Besides,the specified role of superlemma has not been interpreted explicitly in the CYs comprehension.We attempt to illustrate the role of lemmas with distinct syntactic structures on superlemma activation.During the experiments we further probe into the structural symmetry effects,the relationship between unitary and morpheme representation,between literal and figurative representation,between syntactic and semantic representation.Consequently,we interpreted the precise function of lemma on the superlemma in the representation of CYs,when it is processed from literal meaning to figurative meaning.The paper attempts to interpret the representation models of CYs with distinct structures on the basis of hybrid model and superlemma theory,hoping to revise the superlemma theory according to the characteristics of CYs representation.Thus,we also revised the precise function of superlemma in the representation of CYs with unavailable syntactic structures or semantics.3.Research questionsThe experiments consist of two steps.Step one focuses on the impact of syntactic structures on CYs descriptive norms.We conducted four experiments in step two.Step two emphasizes the impact of syntactic structures on CYs comprehension.In step 1 we will answer two questions:(1)will there be salient significance between the 5 CYs descriptive norms and 7 syntactic structures?(2)are the 7 CYs descriptive norms related with their reaction time?In step 2 we try to reply four questions: under the influence of syntactic structures in CYs comprehension(1)is there structural symmetry?(2)what's the relationship between whole word representation and morpheme representation?(3)between literal meaning representation and figurative meaning representation?(4)between syntactic representation and semantic representation?(5)what's the role of lemmas with distinct syntactic structures in word representation?(6)what's the specified role of supperlemms in the CYs comprehensuin with unavailable syntactic structures or semantics?4.ExperimentsWe used the undergraduates of XXX University as participants.In experiment 1,we used questionnaires in experiment 1.Normative measures were obtained for knowledge,familiarity,subjective frequency,age of acquisition,predictability,literality,and compositionality for 350 CYs,and the influences of the CYs syntactic structures on the descriptive norms were analyzed.Consistent with previous studies,all of the norms yielded a high reliability,and there were strong correlations between knowledge,familiarity,subjective frequency,and age of acquisition,and between familiarity and predictability.Unlike in previous studies(Libben & Titone,2008),however,we observed a strong correlation between literality and compositionality.Our normative results for familiarity,knowledge,predictability,AoA,subjective frequency,literality,and compositionality among 350 CYs with seven structures(VO,SM,SV,VV,VOVO,SMSM,and SVSV)seemed to support the hybrid view of idiom representation and comprehension.The influence of the CYs syntactic structures on the descriptive norms did not have the same pattern for all variables.We concluded that CYs with an SM composition are less likely than VOVO CYs to be decomposable,to be recognized through their constituent words,or to be familiar to,known by,or encountered by users.CYs with an SMSM structure are also less likely than VOVO CYs to be decomposable or to be known or encountered by users.The factor analysis illustrated that CYs knowledge,familiarity,subjective frequency,AoA(age of acquisition)and predictability play a major role when idioms are processed as a whole unit.Whereas,CYs compositionality and literality acts as main norms when idioms are accessed as morphemes.Experiment 1 convincingly discovered that CYs knowledge,familiarity,subjective frequency,compositionality,literality have salient significance under the influence of syntactic structures.The present study provides an outline of how CYs characteristics are intercorrelated and how their features interact with their syntactic structures,which can be used as a reference for further studies on idioms in Chinese.Empirical studies can investigate how the syntactic structure of a CY influences its representation and comprehension,so that the hybrid view on idiom representation and comprehension can be finely specified.Experiment 2 investigated the relationship between descriptive norms and reaction time.The distinct relationship between them was found.It demonstrates that AoA is the highest credible norm to predict CYs comprehension time.Those CYs with less AoA can be easier comprehended by the participants.On this basis,we assume tentatively that all 11-year-olds were equal in terms of their language exposure,particularly CYs exposure.One possibility is that participants might simply be taught to learn the idiomatic sequence as a whole and to memorize the example sentences by heart(Wang & Liu,2003).Other norms as CYs knowledge,familiarity,subjective frequency,compositionality,literality and predictability can also predict CYs comprehension time.This phenomenon manifests that descriptive norms are highly correlated with CYs on-line reaction time.Experiments 3,experiment 4,exprriment5 and experiment 6 emphasize the influence of syntactic structures on CYs representation of comprehension.Experiment 3 and experiment 4 were conducted to examine the characteristics of cognition of symmetrical structure of CYs.Experiment 3 applied meaning decision paradigm and experiment 4 used CYs meaning decision task in reverse order.Both experiments got identical results in the respects of structural symmetry,syntactic structure and semantics.Experiments 3 and 4 illustrate that there is significant structural symmetry effect in CYs comprehension,as was illustrated by the results that symmetrical CYs were processed faster than asymmetrical CYs.The reaction time of symmetrical CYs of VOVO composition and SMSM composition was faster than that of asymmetrical CYs of VO composition and SM composition.It seems that asymmetric syntax poses greater challenges for CYs understanding.However,the structural symmetry does not exist in the CYs of SVSV composition and SV composition.Another,the significance of structural symmetry effect is salient in the CYs literal process;while the significance of structural symmetry effect is not salient in the CYs figurative process.It indicates that CYs with different syntactic structures are inclined to have distinct representation in the aspect of structural symmetry.Different from alphabetic language,CYs has its own properties of structural symmetry as mentioned above.The results are further discussed based on different models of Chinese compound words recognition(Peng,Li,Liu,1994;Ding & Peng,1997;Zhang & Peng,1997;Peng,Ding,Wang,Taft,Zhu,1999;Wang & Peng,2000).Just like IIC(Inter and Intra Connection Model),the superlemma theory interprets that symmetrical CYs can be regarded as a combination of two compound words that consist of two groups of morphemes with the same function and property.These two compound words form an exciting link and facilitate each other in the CYs comprehension.Meanwhile,the morphemes in each compound word are also activated.These morphemes with the same syntactic structure are apt to form an exciting link.The identical syntactic structure facilitates the activation of superlemma.As a result,when the superlemma comes up in the participants' mind,the syntactic collocation comes up automatically.Thus the processing time from literal meaning to figurative meaning is significantly faster.Experiment 4 used CYs meaning decision task in reverse order to further testify the syntactic symmetry of CYs.The results showed that the reaction time of CYs of SMSM composition in reverse order was slower than that of CYs with orthographic order.The results indicate that the individual morphemes in reverse order are not activated one by one,but the CYs are extracted as a whole word during the understanding process.On Consensus with the experimental hypothesis,it manifests that CYs of SMSM composition in reverse order tends to be retrieved as a whole.On the other hand,the reaction time of CYs of SV composition in reverse order was faster than that of CYs with orthographic order.The morphemes of CYs of SV composition are activated individually and CYs of SV composition are inclined to be processed in morpheme representation model.Both experiment 3 and experiment 4 support the hybrid model and supperlemms theory.The outcome proves that superlemmas with distinct syntactic structures have various impacts on CYs representation.The results manifest that the CYs with distinct syntactic structures contain different representation models.Crucially,the lemmas with distinct syntactic properties have facilitating or inhibiting effects on the superlemma activation.The experiment 5 took space segmentation paradigm to further explore the relationship between morpheme and whole word representation during symmetrical CYs comprehension processing.For one thing,the research field is limited in symmetrical CYs.For another,the experiment 3 and experiment 4 may lead to similar results due to identical decision task.The results showed that CYs of VOVO composition,SMSM composition and SVSV composition tend to process variously even under the same case of symmetrical structure.The CYs of SMSM composition are apt to be retrieved as whole word on account of the high semantic tenacity of individual constitutes.Our results explicitly showed that the reaction time of non-space segmentation was saliently faster than that of morpheme segmentation,while the reaction time of non-space segmentation was not significant as that of non-morpheme segmentation.The CYs of VOVO composition tend to be extracted by morpheme access.The morphemes are liable to be activated one by one since the semantic tenacity is low.The results evidently illustrated that the reaction time of morpheme segmentation and non-space segmentation was faster that of non-morpheme segmentation.The CYs of SVSV composition are likely to be processed as a whole word as the reaction time of non-space segmentation and morpheme segmentation was the fastest,and there was no salient significance between that of morpheme segmentation and morpheme segmentation.The results of experiment 5 support the superlemma theory.The results of space segmentation experiment manifest that the representation of individual word is essential in idiom comprehension.Experiment 5 confirms the representation of CYs as the hybrid and superlemma model.It also testifies that CYs with different syntactic structures present distinct representation models.The results indicate that there is no absolute whole idiom representation or absolute morpheme representation.They coexist in the idiom comprehension processing.The morpheme representation among constitutes in CYs comprehension is obligatory.Experiment 6 used correct-or-wrong decision paradigm.Reaction time was graded in our experimental conditions involving correct CYs,synonyms,semantic violations,and combined semantic and syntactic violations.As mentioned earlier,the fixed-form idioms in Chinese have a structural stability,and the characters/morphemes for any given idiom are fixed and not subject to substitution or paraphrase.These characteristics of CYs give us a hint that the last character in each test material has an extremely high cloze probability(i.e.,98%,according to our pretest).In other words,items in the three incongruous conditions are unlikely to occur in the language.Those correct CYs were recognized the fastest.This result manifests that the syntax and semantics compete with each other in the CYs comprehension.The correct syntax conduces to the access of CYs figurative meaning.The CYs of semantic violations and CYs of combined semantic and syntactic violations were processed the second faster.There was no difference between the semantic only and the combined violation conditions.As the superlemma theory predicts that the function of the superlemma is too strong that even the figurative meaning of CYs with incompatible syntax or semantics can also be accessed.As to the CYs of semantic violations,the participants might ignore the incompatible syntax and got access to the figurative meaning directly via superlemma.Thus the less amount of activation led to faster reaction time.Concerning the CYs of combined semantic and syntactic violations,the superlemma did not activate the concept of CYs.The very economical way of decision work condensed the reaction time as well.The CYs of synonyms were the last to be recognized.This result indicates that synonyms activated the supperlma,and the superlemma in turn activated the CY figurative meaning.However,there comes a competing relationship between the synonymous meaning and the figurative meaning.Thus the reaction time was delayed.The results are consistent with some previous studies.In this study we argue that the processing of idioms is more like that of regular phrases or short sentences.Experiment 6 displayed the stimuli character-by-character,which encouraged participants to process the individual characters in an idiom successively.The four words in CYs were presented individually by mask paradigm.Interestingly,the results are the same as that of whole word presentation(Holsinger,2013).These results manifest that the activation of certain literal meaning of morphemes is prerequisite in CYs comprehension.Furthermore,facilitation to the semantic activation of literal morpheme can affect the extraction of figurative meaning in CYs processing.Just as the superlemma hypothesis predicts that some activation of the literal component lemmas is obligatory for the superlemma to become active and for the idiomatic meaning to be accessed,these results prove that the activation of literal meaning of the individual morpheme is obligatory in CYs comprehension in access from literal to figurative meaning.Experiment 6 supports the hybrid model and superlemma theory as well.The results showed that the structural symmetry also exists in CYs with certain syntactic structures under different syntactic or semantic conditions.In accordance to the superlemma theory,the conceptual meaning of a CY arises from but more than the literal meaning via the media of superlemma.The semantic integration of Chinese language,not like alphabetic languages,does not fully rely on the intact syntactic structures as indicated by prime ERP researches.Based on superlemma theory,the retrieval of superlemma has restricted the syntactic position of individual constitute.Therefore,CYs can still be processed on the premise of syntactic and semantic transformation.Although the superlemma theory offers a specified mechanism for the nature of CYs various comprehension and representation,it does not clarify the exact role the superlemma takes from literal to figurative meaning extraction.Experiment 6 indicates that the semantic processing still exists even though there are unavailable transformations in syntax and semantics.The semantic representation takes a priority during CYs understanding process.Accordingly,an incompatible syntactic structure does not fully prevent CYs consideration.The figurative meaning may be active even in the syntactically unavailable condition.If this is correct,revision of the representational content of the superlemma and its role during CYs comprehension is required.One possible revision is that during comprehension the participants do not actively check the current syntactic structure against the superlemma representation.Given that many idioms have some degree of structural flexibility,it may be uneconomical for the parser to check against the potentially large set of possibilities immediately.During production,however,the superlemma may be capable of playing a more direct role in shaping the structural configuration of the CYs.If this is the case we predict that syntax may play a stronger role when the CY is less flexible and hence there are fewer possibilities to check against or more processing resources are available.5.ConclusionIn a word,the results of these experiments mentioned above are compatible with the contemporary models of idiom representation.Our results support the hybrid model and superlemma theory in CYs comprehension.Experiment 1,2,3,4,5 and 6 explicitly illustrate that superlemmas with distinct syntactic structures have different impact on the CYs norms and representation.The off-line experiments show that the syntactic structures have an influence on the CYs' descriptive norms.The on-line experiments indicate the CYs demonstrate its own processing properties under the influence of distinct syntactic structures.CYs embrace structural symmetry effects.CYs contain semantic representation priority effects.On the basis of superlemma theory,there is a parallel relationship between whole word and morpheme representation,between literal and figurative meaning representation,between syntax and semantics representation during CYs comprehension.There is a competitive relationship between these representations.Literal processing is obligatory for retrieval of figurative meaning.The CYs with distinct syntactic structures are processed with different representation tendency.Experiment 6 supports the strong superlemma theory.The strong superlemma theory revised the inadequacy of superlemma theory.If the syntactic structure is incompatible,we have structural representations associated directly with conceptual content;thus,the syntactic process can be omitted.If the semantics of lemma is unavailable,the activation of partial component lemmas may lead to the activation of superlemma so that the conceptual meaning of CY can be accessed.In many ways the representation of CYs behaves like any other member of the lexicon.Our research specifies the role of superlemma with distinct syntactic structures in the CYs string.These results render us to further study the mechanism of CYs representation of comprehension.In turn,it will encourage us to probe more into the nature of Chinese language comprehension mechanism. |
PDF Full Text Request