Medicine, Science & Technology

The Production of Head-Initial and Head-Final Languages

Description
The Production of Head-Initial and Head-Final Languages
Published
of 17
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Chapter 6 The Production of Head-Initial and Head-FinalLanguages Mikihiro N. Tanaka, Holly P. Branigan, and Martin J. Pickering 6.1 Introduction Speakers can sometimes express the same message in many ways. For instance,anevent in which apirate knocks over aswing can be described in Englishusingthe active as ‘‘ The pirate knocked over the swing’’   or using the passive as ‘‘ Theswing was knocked over by the pirate’’  . In Japanese, the same message can beexpressed in at least four ways, either as an active (1-a, 1-c) or a passive (1-b,1-d) and with either the subject before the object (SOV; 1-a, 1-b) or the objectbefore the subject (OSV; 1-c, 1-d):(1) a. SOV-Active 海 賊 がブランコを 倒 した。 Kaizoku-ga buranko-o taoshita.Pirate-NOM Swing-ACC knock over-PAST   ‘The Pirate knocked over the swing.’b. SOV-passive ブランコが 海 賊 によって 倒 された。 buranko -ga Kaizoku -niyotte taosareta.Swing -NOM Pirate-Oblique knock over-PAST-Passive‘The swing was knocked over by the pirate.’c. OSV-Active ブランコを 海 賊 が 倒 した。 Buranko-o Kaizoku-ga taoshita.Swing-ACC Pirate-NOM knock over-PAST‘The Pirate knocked over the swing.’ M.N. Tanaka ( * )Faculty of Arts and Sciences, Showa University, 4562 Kamiyoshida Fujiyoshida-shi,Yamanashi-ken, 403-0005, Japane-mail: mtanaka@cas.showa-u.ac.jpH. Yamashita et al. (eds.),  Processing and Producing Head-final Structures ,Studies in Theoretical Psycholinguistics 38, DOI 10.1007/978-90-481-9213-7_6,  Springer Science þ Business Media B.V. 2011113  d.OSV-passive 海 賊 によってブランコが 倒 された。 Kaizoku -niyotte Buranko -ga taosareta.Pirate -Oblique Swing -NOM knock over-PAST-Passive‘The swing was knocked over by the pirate.’All four sentences have the same denotational meaning but clearly differ inemphasis. Speakers must choose one sentence over the others. How do they dothis?Most studies that investigate these issues have been carried out in English, alargely head-initial language. They suggest that, for instance, English speakerstend to place conceptually more accessible (e.g., animate or concrete) entitiesearlier than less accessible ones. Depending on which entity is more accessible,such tendencies can lead to a preference to produce either passives or actives(e.g., Bock & Warren, 1985). In addition, many studies suggest that people tendto repeat the syntactic structure that they have previously used ( structural  priming : Bock, 1986; Pickering & Branigan, 1998; see Pickering & Ferreira, 2008). Other studies have suggested that speakers also tend to place shorterphrases earlier than longer phrases (e.g., Arnold, Wasow, Losongco, &Grinstrom, 2000; Stallings, MacDonald, & O’Seaghdha, 1998; Hawkins, 1994). However, there has been very little work on how speakers choose syntacticstructure in head-final languages such as Japanese and even less comparingsentence production in head-initial and head-final languages. It is important totest both types of language, in part because some grammatical alternationsoccur in one type of language but not the other. In addition, the central role of the verb in production may mean that processing is fundamentally affected bythe position of the verb. Hence there may be important differences in theproduction system between head-initial and head-final languages. Therefore,in this chapter, we examine three factors – the conceptual accessibility of referents, structural priming and the length of noun phrases – which havebeen hypothesized to affect production, and compare their influence on pro-duction in both types of language. We then consider how such results mightinform theories of language production and in particular whether commonmechanisms can be postulated for head-initial and head-final languages. 6.2 Models of Language Production Most current models of language production distinguish three different stages: conceptualization , determining what message is expressed;  formulation , realizingthis message as a series of linguistic representations; and  articulation , convertingthese representations into motor movements (e.g., Levelt, 1989). Formulation is 114 M.N. Tanaka et al.  generallydividedintotwodifferentprocessinglevels:  grammaticalencoding ,duringwhich the syntactic component ( lemma ) of appropriate lexical entries is retrievedand used to generate syntactic structure; and ( morpho)phonological encoding ,during which the morphophonological component of those entries is retrievedand used to generate morphological and phonological representations.Although the details of grammatical encoding are controversial(e.g., Hartsuiker, Kolk, & Huiskamp, 1999; Pickering, Branigan, & McLean,2002), most models of production (e.g., Bock & Levelt, 1994; Garrett, 1975; Levelt, 1989) assume that it involves two different types of processing:  func-tional processing , which maps from the conceptual representation to an un-ordered set of lemmas that are tagged for grammatical function (e.g., subject,direct object); and  positional processing , which determines constituent structure(including word order). The processing tasks at each level are thought to bedifferent.At thefunctionallevel, conceptual informationistransformed intoanunordered representation that incorporates information about grammaticalfunction. In contrast, the positional level is a constituent-structure representa-tion that is specified for linear order.Thus, processing at the two levels could be influenced by different typesof information. Since functional processing involves the transformation of aconceptual representation into a sentence, such processing might be affected byconceptual information (e.g., animacy or imageability; Bock & Warren, 1985;McDonald, Bock, & Kelly, 1993). In contrast, positional processing deals withconstituent assembly (determination of word order) and might plausibly beinfluenced by lexical or phonological information (e.g., Bock, 1986, 1987; Kelly, Bock, & Keil, 1986; McDonald et al., 1993). However, some researchers have claimed that conceptual representations might influence positionalprocessing as well (Branigan & Feleki, 1999; Kempen & Harbusch, 2004; Prat-Sala & Branigan, 2000).More importantly, many models assume that language production is highlyincremental (Bock, 1982; Kempen & Hoenkamp, 1987; Levelt, 1989), so that as soon as minimal information is available, it will be passed down to the nextlevel. In this sense, the processor does not need to wait until all the informationrelevant for production of the entire utterance is available but can deal withfragments of information simultaneously. Incremental processing is theoreti-cally attractive because it can explain how speakers generally manage toproduce language fluently and efficiently without frequent pausing to plan thenext part of their utterance. Evidence that speakers’ production processes aresensitive to the accessibility of different kinds of information (such as the acces-sibilityofparticularconceptsorparticularsyntacticstructures,asnotedabove)istakenassupportforthisassumption.Sucheffectsareeasilyexplainedifinforma-tion is processed as and when it becomes available; it is in contrast difficult toexplain why accessibility affects sentence production if the processor does notproceed incrementally but rather waits until all relevant information has beenretrieved before initiating further processing of any of that information. 6 The Production of Head-Initial and Head-Final Languages 115  6.3 Factors Affecting Syntactic Choice in Head-Initialand Head-Final Languages Research on language production has investigated a number of factors thataffect choice of syntactic structure. In this section, we consider three factorswhoseinfluencehasbeenexploredinbothhead-initialandhead-finallanguagesand examine the extent to which they exert comparable effects in both languagetypes.As noted above, many researchers have suggested that language productionis processed incrementally (e.g., Kempen & Hoenkamp, 1987; Levelt, 1989). That is, the processor deals with information as soon as it becomes availableandat thesame time thatdifferent aspects of processingtake place. In this sensespeakers do not need to wait until all the information is retrieved but cangenerate an utterance as soon as minimal input is available.In keeping with an incremental model of production, language production ishypothesized to be influenced by the relative accessibility of information. Withrespecttosyntacticprocessing,relevantdimensionsthatmightbeimportantarethe relative accessibility of concepts associated with particular entities and therelative accessibility of syntactic information. 6.3.1 Conceptual Accessibility Many researchers have suggested that conceptual representation influencesgrammatical encoding during production. Some evidence comes from anexperimental study by Bock and Warren (1985). Bock and Warren proposedthat the choice of grammatical function assignment is influenced by what theyterm  conceptual accessibility . They claimed that conceptual accessibility influ-ences grammatical function assignment in two ways. Firstly, the ease of wordretrieval from the mental lexicon influences function assignment, so lemmasthat are retrieved faster will be assigned a grammatical function before lemmasthat are retrieved less quickly. Secondly, grammatical functions are assignedaccording to Keenan and Comrie’s (1977) Noun Phrase (NP) accessibilityhierarchy, such that the subject function is assigned first, direct object second,then indirect object and oblique object. Thus lemmas that are retrieved morequickly tend to be assigned the subject function and lemmas that are retrievedless quickly tend to be assigned an object function. Such a process will partlydetermine the grammatical structure of the sentence (e.g., active or passive).Totestthis,BockandWarren(1985)conductedasentencerecalltask,asking participants to memorize sentences and recall them in a randomized order.They used imageability as an index of conceptual accessibility and demon-strated that participants tended to recall active and passive sentences in a waythat allowed more concrete (hence more imageable) entities to appear as sen-tence-initial subjects. For example, they were more likely to recall passive 116 M.N. Tanaka et al.  sentences with more imageable agents and less imageable patients (e.g.,  Theshock was administered by the doctor ) as active sentences (e.g.,  The doctoradministered the shock ) than to recall passive sentences with less imageableagents and more imageable patients (e.g.,  The doctor was administered by theshock ) as actives (e.g.,  The shock administered the doctor ). Similarly, they weremore likely to recall active sentences with less imageable agents and moreimageable patients as passive sentences than to recall active sentences withmore imageable agents and less imageable patients as passives. In addition tothis, Bock and Warren (1985) tested for effects of conceptual accessibility on noun-phrase conjuncts (e.g.,  The pirate and the swing were gone  vs.  The swingand the pirate were gone ). They used this structure because the nouns that makeup the conjuncts in NP conjunctions have the same grammatical function andso word order could be changed freely without any associated variations ingrammaticalfunctionassignment.Bock andWarren failedtofindanytendencyfor participants to produce more imageable nouns earlier than less imageableones in NP conjunctions. Thus they argued that conceptual accessibility influ-ences the choice of grammatical function assignment but not word order.Further supporting evidence for this conclusion was reported by McDonaldet al. (1993), who found a tendency for animate nouns to precede inanimatenouns in the transitive sentences but not in the conjuncts. When both animatenouns and inanimate nouns appeared in the same subject position of conjunctphrases, the effect of animacy disappeared. This finding was consistent with thehypothesis by Bock and Warren (1985) where conceptual accessibility influ-ences the choice of grammatical function assignment.However, since Bock and Warren (1985) and McDonald et al. (1993) inves- tigated this issue in English, whichhas a relatively rigid order, it is hard to makeaclear dissociationbetween the effects ofgrammaticalfunctionassignment andword order determination. In fact, further studies using different head-initiallanguages showed that conceptual accessibility can also influence the choice of word order. For instance, Branigan and Feleki (1999) tested the effect of animacy in Greek using sentence recall and found that speakers were morelikelytorecallsentencesinaformthatallowed theconceptuallymoreaccessibleentity to precede the less accessible entity, irrespective of grammatical function.Thus, their participants tended to recall SVO sentences (e.g.,  Sta dimokratika politevmata, to sindagma sevete ton politi  : ‘‘In democratic regimes, the law-NOM respects the citizen-ACC’’) as OVS sentences (e.g.,  Sta dimokratika politevmata, ton politi sevete to sindagma : ‘‘In democratic regimes, the citizen-ACC respects the law-NOM’’) more when the subject was inanimate and theobject was animate, than when the subject was animate and the object wasinanimate. Equally, their participants were likely to recall OVS orders (e.g.,  Stadimokratika politevmata, to sindagma sevete o politis : ‘‘In democratic regimes,the law-ACC respects the citizen-NOM’’) as SVO (e.g.,  Sta dimokratika poli-tevmata, o politis sevete to sindagma : ‘‘In democratic regimes, the citizen-NOMrespectsthelaw-ACC’’)morewhentheanimatenounwasthesubjectthanwhenthe inanimate noun was the subject. 6 The Production of Head-Initial and Head-Final Languages 117

Bab I

Jan 24, 2019

TUgas Agama

Jan 24, 2019
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks