Mandarin Chinese compounds, their representation, and processing in the visual modality

This article was downloaded by: [University of Toronto Libraries]On: 10 August 2014, At: 16:16Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office:Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Writing Systems ResearchPublication details, including instructions for authors and subscriptioninformation:http://www.tandfonline.com/loi/pwsr20

Mandarin Chinese compounds, theirrepresentation, and processing in the visualmodalityVedran Dronjic aa University of Toronto, CanadaPublished online: 20 Dec 2011.

To cite this article: Vedran Dronjic (2011) Mandarin Chinese compounds, their representation, and processingin the visual modality, Writing Systems Research, 3:1, 5-21, DOI: 10.1093/wsr/wsr005

To link to this article: http://dx.doi.org/10.1093/wsr/wsr005

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and ourlicensors make no representations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressed in this publicationare the opinions and views of the authors, and are not the views of or endorsed by Taylor &Francis. The accuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liable for anylosses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantialor systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, ordistribution in any form to anyone is expressly forbidden. Terms & Conditions of access and usecan be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/pwsr20

http://www.tandfonline.com/action/showCitFormats?doi=10.1093/wsr/wsr005

http://dx.doi.org/10.1093/wsr/wsr005

http://www.tandfonline.com/page/terms-and-conditions

Correspondence: Vedran Dronjic, Centre for Educational Research in Languages and Literacies, OISE/University of Toronto, 252 Bloor Street West, Toronto, ON M5S 1V6, Canada. E-mail: [email protected]

5Writing Systems Research, Vol. 3, No. 1, 2011, 5–21. © The Author 2011. Published by Oxford University Press. For Permissions, please email: [email protected] doi:10.1093/wsr/wsr005

Mandarin Chinese compounds, their representation, and processing in the visual modality

Vedran Dronjic University of Toronto, Canada

Abstract This article reviews a number of key linguistic and psycholinguistic findings pertaining to Mandarin Chinese compound words, their representation, and their processing in the visual modality. A summary of a number of key facts about compounds and compounding in general is followed by a linguistic overview of Mandarin compounds. Next, the issue of the basic unit of representation in the Mandarin mental lexicon is discussed, and the conclusion is reached that this unit is the word rather than the morpheme. A number of factors influencing visual access to compounds in languages in general and Mandarin in particular are identi-fied and empirical findings on these factors are summarized. The article concludes by examining the relationship between Chinese orthography and compound pro-cessing in Mandarin and by identifying a number of issues which await to be addressed in future research.

The present article is an overview of what is currently known about how Mandarin Chinese compounds are represented and processed (in the visual modality) by adult speakers. The morphosyllabic nature of the Chinese script has long fascinated researchers. Unlike most of the world's scripts currently in use, which typically represent language at the phone-mic level (Turkish), a combination of the phonemic and subsyllabic levels (English), or, less frequently, at the syllabic level (Ge'ez or Brahmic abugidas; Japanese kanas), the Chinese script consists of char-acters which by and large represent morphemes. The fact that the Chinese script is the oldest writ-ing system still in use is an important reason for its morphosyllabic character; scripts representing syl-lables and individual sounds evolved later. Also, as is often pointed out, this solution is favoured over an alphabet because it enables speakers of various

Sinitic languages (Mandarin, Yue, Wu, etc.) to share a written language, albeit based on Mandarin gram-mar and vocabulary, thereby bringing a degree of cohesion to the vast and heterogeneous Han Chinese community. (For more on the Chinese script, see Hanley et al ., 1999 ; McNaughton and Ying, 1999 .)

The psycholinguistics of Mandarin compounds was chosen as the focus of this survey due to the fact that over 70 % of Mandarin words are disyllabic, dimorphemic compounds ( Institute of Language Teaching and Research 1986 ). In Xing's estimate (2006) , as many as 80 % of Mandarin words are compounds. Compounds also predominate among Mandarin neologisms (95 % ), whereas derivatives (words created from other words, usually through the addition of prefixes or suffixes, e.g. English ‘teach’—‘teacher’) account for only a little > 2 % of all newly coined words ( Ceccagno and Basciano, 2007 ).

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic

6 Writing Systems Research, Vol. 3, No. 1, 2011, 5–21

Thus, compounding can be said to be the main mor-phological process in Mandarin ( Arcodia, 2007 ), both diachronically and synchronically. As a rela-tively simple morphological phenomenon (often involving no more than the combination of two independent lexical items), compounding is prob-ably the oldest morphological process in the world's languages ( Libben, 2007 ). Compounds exist in all languages investigated to date ( Dressler, 2007 ). It should be noted, however, that compounding is not equally common across languages; English, although compound-friendly, relies on compound-ing less extensively than Mandarin ( Janssen et al ., 2008 ). Affix-rich languages such as Spanish, Polish, or Hebrew do so even less.

As in all discussions of morphological representa-tion and processing, the question we will ask is how morphologically complex words (here compounds) are represented and accessed in the mental lexicon, and, more specifically, whether they are represented and accessed as wholes or in decomposed form, with an emphasis on representation and processing in the written modality. While the processes of recognizing words in speech and writing are certainly partially different, the lexical and conceptual representations that these processes ultimately tap are shared and the mental lexicon is, for very good reason, pre-sumed to be modality independent. What makes Mandarin compounds particularly interesting to researchers is the fact that Chinese characters make Mandarin morphemes salient to the reader (relative to, say, written English). It is, then, worth asking what impact morphological structure has on read-ing Chinese and, conversely, what impact (if any) being literate in Chinese might have on the mental lexicon. Due to the salience of the morpheme in written Chinese and the fascination which the script holds for psycholinguists, the bulk of the research on Mandarin compound representation and processing focuses on the written modality ( Myers, 2007 ).

The article is organized as follows: a discussion of the most important facts about compounding in general is followed by an overview of Mandarin morphology, with a strong emphasis on compounds. Next, a major issue in the psycholinguistics of Chi-nese is addressed, ‘What is represented in the lexi-con?’. This is followed by an overview of a number

of factors known to play a role in the reading of compounds in general and in Mandarin in particu-lar. Whenever there is reason to believe that a spe-cific effect is due to the nature of the Chinese script, this will be pointed out. Finally, a number of con-clusions regarding the surveyed research are drawn and, as customary, gaps in the existing research are identified.

1 Compounds across Languages

A common way of defining compounds is to say that they are words consisting of at least two other words (e.g. Bauer, 2006 ; Fabb, 2005 ). For instance, the English compound ‘primary school’ contains ‘primary’ and ‘school’, both potentially independent words. The compound ‘primary school’ resembles the phrase ‘an interesting school’. What differenti-ates the two is that we would expect the former, but not the latter, to have a separate entry in the mental lexicon. Numerous criteria have been proposed for distinguishing compounds from phrases, but such criteria are, as a rule, mere tendencies and either do not apply universally and/or are not exclusive to compounds. Therefore, compound status is a mat-ter of degree rather than a strict yes/no dichotomy. For instance, compounds typically do not allow the insertion of elements: A black widow spider from Canada is referred to as a ‘Canadian black widow’, not a ‘*black Canadian widow’, although colour adjectives usually precede adjectives denoting origin (c.f. ‘a black Chinese car’). Thus, ‘black widow’ is a compound while ‘black car’ is not. However, Mandarin features a class of compounds which do allow their constituents to be separated (e.g. bāngmáng 帮忙 ‘help + busy = help’; bāng le yī gè dà máng 帮了一个大忙 ‘help + perfec-tive aspect + one + classifier + big + busy = helped a lot’; Siewierska et al. , 2010 ). For a more detailed discussion of various criteria for identifying com-pounds, see Dressler (2007) and Packard (2000) .

Apart from containing potentially independent forms, compounds may also contain forms incapable of independent functioning. For instance, English ‘nephrology’ contains the morpheme {NEPHR}

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

Writing Systems Research, Vol. 3, No. 1, 2011, 5–21 7

Mandarin Chinese compounds

‘kidney’, which never appears independently. Thus, it may be more accurate to adopt a definition of compounds as words containing at least two roots (i.e. morphemes which carry a substantial amount of lexical meaning and are not affixes). In languages like English, roots are typically free ( Carstairs-McCarthy, 2002 ), so most compounds indeed contain words. While autonomous words (i.e. forms which can fill syntactic slots on their own) are universally preferred bases of compounds ( Dressler, 2007 ), in languages like Serbo-Croatian, free roots are less common, and compounds often (1) contain affixes in addition to roots; and (2) contain bound roots, which never appear as independent words. Consider Serbo-Croatian vetrokaz ‘weather vane’, which con-sists of the root {VET_R} ‘wind’, the interfix {O}, and the root {KAZ} ‘tell’. Note that neither vetr nor kaz ever appear as independent words. While a definition allowing words like vetrokaz to be clas-sified as compounds seems to be a better fit for the variation found in the world's languages than one insisting on independent words only, there is no consensus on this issue in theoretical linguistics.

Compound meanings tend to be compositional, with each of the constituents contributing its seman-tics to the whole. For instance, a ‘scarecrow’ is an object used to scare crows. However, while the rela-tionship between the parts is evident to those famil-iar with this compound, inferring the meaning of ‘scarecrow’ would not be straightforward for some-body unfamiliar with the word. Would a scarecrow be something which scares crows, a crow which is easily scared, or perhaps a crow that scares peo-ple? Compound meanings are not fully predictable (1) because of semantic drift, that is, factors such as polysemy (a ‘chairperson’ is not a person who is a chair, but a person who presides over something such as a committee) or metaphor (a ‘bottlenose dol-phin’ does not have a bottle for a nose, but rather a bottle-shaped nose); and (2) because there are mul-tiple possible semantic relations between the parts of a compound, but these are not fully spelled out as they would be in a phrase or clause ( Fabb, 2005 ). Moreover, not all of a compound's meaning is con-tained in its constituent parts. Thus, a ‘wheelchair’ is not just any chair with wheels, but only a specific kind of chair with wheels ( Bauer, 2006 ).

Compounds are often classified based on headed-ness. The head of a compound is its part related in meaning to the whole word such that, semantically, the compound becomes a subordinate of the head. Also, the head belongs to the same syntactic class as the entire compound. In endocentric compounds, one of the constituents is the semantic head. For instance, in ‘espresso cup’, ‘cup’ is the head, since an espresso cup is a kind of cup and both ‘cup’ and ‘espresso cup’ are nouns. In exocentric compounds, the semantic head is not contained within the compound. ‘Copycat’ is an exocentric (headless) compound, as a copycat is not a kind of cat. Note, however, that ‘cat’ still has head-like characteristics in ‘Copycat’; it assigns its syntactic class to the compound and accepts inflectional morphemes such as {S PLURAL } on behalf of the entire word. The meanings of exocen-tric compounds are generally more difficult to infer than those of endocentric compounds, and this may be why endocentric compounds are more common in languages ( Dressler, 2007 ; Katamba and Stonham, 2006 ). A further distinction is made between sub-ordinate compounds, in which one of the elements is syntactically or semantically subordinated to the other (as in the foregoing examples), and coordinate compounds, in which no such relationship is dis-cernible. One example of a coordinate compound is ‘bittersweet’, where both parts resemble the head; something which is bittersweet is both bitter and sweet; in addition, both ‘bitter’ and ‘sweet’ are adjectives, and so is ‘bittersweet’. Note, however, that there are also coordinate compounds in which neither of the elements appears to be the semantic head, such as the Mandarin compound dàxǐao 大小 ‘big + small = size’. In other words, coordinate com-pounds too may be endocentric or exocentric.

In terms of the syntactic classes that serve as input to compounding, noun–noun compounds which are themselves nouns ([N + N] N ; e.g. ‘mailbox’) are the most common ( Dressler, 2007 ). Other, smaller, types include adjective–noun compounds (e.g. ‘loud-mouth’), noun–adjective compounds (e.g. ‘navy blue’), verb–noun compounds (e.g. ‘breakneck’), adjective–adjective compounds (e.g. ‘dark red’), and others. Identical syntactic classes may com-bine to yield different outputs. For instance, an English verb and noun may combine to yield

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


a noun ([V + N] N , ‘drawbridge’), an adjective ([V + N] A , ‘breakneck’), or perhaps even a verb ([V + N] V , ‘shunpike’). Note, however, that certain types of compounds are more common in some lan-guages than in others. For instance, [V + V] com-pounds (e.g. ‘typewrite’) are rare in English, but common in Mandarin. For more on compounds, see Dressler (2007) , Fabb (2005) , Katamba and Stonham (2006) , and Selkirk (1982) .

In some languages, compounds may be marked prosodically in speech, which helps signal that the compound is a single lexical item. For instance, many English compounds have primary stress on the first constituent (compare ‘bláckboard’ = ‘a wall board in a classroom used for writing, regardless of its actual colour’ and ‘bláck bóard’ = ‘a board which is black’). Depending on the orthographic rules of particular languages, compounds may be written without ortho-graphic spacing (e.g. ‘keyhole’), with a hyphen or a similar orthographic symbol (e.g. ‘bleary-eyed’), or with a space (e.g. ‘prime minister’). While many orthographies feature more than one of these spacing solutions and may even allow certain compounds to be written in more than one way, others consistently apply just one (e.g. all Mandarin compounds are written without spacing because orthographic spaces are not used intrasententially other than after certain punctuation marks).

2 Mandarin Compounds

Unlike Classical Chinese, morphologically an essen-tially isolating language ( Branner, 2001 ) with a strong tendency towards monosyllabicity ( Arcodia, 2007 ), present-day Mandarin features a number of different morphological phenomena and prefers disyllabic words ( Lin, 2001 ). The commonly held view that Mandarin has almost no morphology is a misconception. Truth be told, Mandarin morphology is certainly simple compared to the morphological systems of Georgian, Finnish, or Inuktitut.

Most Mandarin morphemes are represented by a single syllable in speech and a single charac-ter in writing. There are around 5,000 morphemes in everyday use in Mandarin ( Zhou et al. , 1999 ), but only 1,300 syllables, out of which 297 have

a single meaning ( Yin, 1984 ), while the rest repre-sent polysemous morphemes or stand for multiple morphemes. Homophones such as míng 明 ‘bright’ and míng 名 ‘name’ are extremely common, and homophonous morphemes can even share a char-acter ( huá 华 ‘Chinese’ and huá 华 ‘magnificent’). Conversely, a single character can be used to write two or more morphemes with different pronuncia-tions, for instance, 了 can stand for liǎo ‘know’ and le ‘perfective aspect’.

Although compounds predominate in Mandarin vocabulary ( Xing, 2006 ), there is more to Manda-rin morphology than compounding. First, let us point out that Mandarin morphemes can be both free (allowed to fill syntactic slots independently) or bound (only found in morphologically complex words). An orthogonal distinction, as in descriptions of other languages, is made between lexical (con-tent) and grammatical (function) morphemes. Free morphemes can thus be lexical (e.g. māo 猫 ‘cat’) and grammatical (e.g. hé 和 ‘and’). Also, certain free lexical morphemes are polysyllabic. Some of these polysyllabic morphemes feature characters which do not occur in other words (e.g. pútao 葡萄 ‘grape’ or qīuyǐn 蚯蚓 ‘earthworm’) and are known as ‘bind-ing words’ ( liánmiáncí 连绵词 ). In other polysyl-labic morphemes (e.g. shāfā 沙发 ‘ sofa ’), characters normally used to represent other morphemes (here, ‘sand’ + ‘distribute’) serve to represent the mor-pheme's sound (e.g. shāfā for English ‘sofa’) while being completely semantically unrelated to the mor-pheme. All of the above are free morphemes and can be considered to be roots.

Nonetheless, compounding of bound roots is the most typical word-formation process in Mandarin ( Packard, 2000 ; Dong, 2004 in Arcodia, 2007 ). There are both lexical and grammatical bound morphemes in Mandarin. Bound roots are prototypical bound lexical morphemes. Fáng 房 ‘house’ is one example. It only appears in complex words (e.g. fángzi 房子 ‘house/building/room’; píngfáng 平房 ‘bungalow’). Apart from bound roots, Mandarin also has a num-ber of bound affixes, although considerably fewer than a typical Indo–European language ( Li and Thompson, 1981 ). These can be both derivational and inflectional. Derivation is the morphological process of adding affixes to bases in order to form

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



new lexical items (e.g. ‘teach’—‘teacher’). Inflec-tion consists of adding affixes to stems to encode grammatical information such as aspect or plural-ity (e.g. ‘car’—‘cars’). Derivational affixes appear in Mandarin as both prefixes (e.g. wú 无 ‘without’) and suffixes (e.g. zhě 者 ‘one who does/is X’), while bound inflectional morphemes are suffixes (e.g. le 了 ‘perfective aspect’ or men 们 ‘human-plural’). There are more suffixes than prefixes (Li and Thompson, 1981). While the free/bound distinction in Manda-rin may appear problematic, Packard (2000) points out that it is relatively straightforward when the following is borne in mind: (1) certain morphemes (e.g. yán 言 ́ speak´) were free in Classical Chinese (and may still be used as such in expressions inher-ited from this language), but are synchronically bound; and (2) certain polysemous morphemes are free with some meanings and bound with others. For instance, gōng 工 is free when it means ‘work’ or ‘job’ and bound when it means ‘labour’, ‘art’, or ‘industry’.

The distinction between bound roots and deriva-tional affixes is particularly important for determin-ing whether a Mandarin word is a compound or a derivative. Two of the three criteria Packard (2000) lists for drawing this distinction seem particularly useful: (1) derivational affixes have more general meanings than bound roots, and compounds contain-ing bound roots appear more lexicalized than deriva-tives; and (2) derivational affixes are more productive than bound roots. (There is no universally adopted definition of productivity, but this term generally refers to how widely a morphological rule applies in the lexicon, whether it applies to novel forms, and what restrictions it is subject to. See Aronoff, 1976 ; Baayen, 1992 ; Booij, 1977 ; and Nishimoto, 2003 , for more details). In addition, Mandarin suffixes often carry the neutral tone ( Lin, 2001 ). Packard cites the example of the bound root yuán 员 ‘person whose job/position is X’ and the derivational suffix zhě 者 ‘one who does/is X’. While the former, simi-lar to the English ‘person’ in ‘salesperson’, tends to appear in words signifying occupations or positions such as jiàoyuán 教员 ‘instructor’ and chūnàyuán 出纳员 ‘cashier’, the latter, apart from being more productive, is more general in meaning and can appear in words such as lǎozhě 老者 ‘senior’ or

zìyuànzhě 自愿者 ‘volunteer’. Still, distinguish-ing derivation and compounding remains a prob-lematic issue in linguistics (see Bauer, 1983 ) and, similar to the phrase/compound contrast, the deriva-tive/compound contrast is best viewed as fuzzy, continuous, and probabilistic rather than strictly dichotomous.

Mandarin compounds come in a variety of types and can be classified in multiple ways. A recent clas-sification ( Ceccagno and Basciano, 2007 ; a refine-ment of Bisetto and Scalise, 2005 ) identifies three macro-types: subordinate, attributive, and coordi-nate, each further subdivided into endocentric and exocentric sub-types. In this taxonomy, subordi-nate compounds contain constituents whose rela-tion resembles that between head and argument (e.g. verb and object). Examples of this type would be dúfàn 毒贩 ‘drugs + peddle = drug trafficker’ (endocentric) or jiānshì 监事 ‘supervise + business/work = supervisor’ (exocentric). Constituents of attributive compounds have a modifier-head rela-tion, exemplified by tiānjià 天价 ‘sky + price = prohibitive price’ (endocentric) or lánlǐng 蓝领 ‘blue + collar = blue-collar’ (exocentric). Finally, in coordinate compounds, constituents display a relation of coordination. Thus shūguǒ 蔬果 ‘vegetable + fruit = fruit and vegetables’ is an endo-centric coordinate compound, while dàxǐao 大小 ‘big + small = size’ is an exocentric coordinate compound.

In terms of the syntactic category of their con-stituents, the largest types of Mandarin compounds are [N + N], [V + N], [V + V], and [A + N] ( Huang, 1998 ). With regard to productivity, all three types of compounds mentioned above are still productive, as attested by Ceccagno and Basciano's (2007) investigation of Mandarin neologisms, with the attributive type being the most productive, followed by the subordinate type. Coordinate compounds are currently the least frequent among newly coined words. While languages are often said to prefer either right-headed or left-headed compounding (cf. Williams’ (1981) controversial Right-Hand Head Rule), Mandarin evades such characteriza-tions. Packard (2000) acknowledges the variation found in Mandarin data, but still proposes a Head-edness Principle, stating that there is a statistical

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


tendency for disyllabic nouns to have a noun on the right and for verbs to have a verb on the left. Ceccagno and Basciano (2007) remark that this generalization only applies to subordinate com-pounds and side with Huang (1998) in conclud-ing that there is no clear headedness preference in Mandarin compounding, with endocentric com-pounds being either left-headed, right-headed, or two-headed. They also stress that left-headedness in Mandarin compounding cannot be relegated to peripheral status (as right-headed compounding per-haps could in Romance languages) and note that all three types of headedness are currently productive. Ceccagno and Basciano conclude that the position of the head in a Mandarin compound is determined by a combination of factors, namely input and out-put categories and relations between the constitu-ents. It is also worth pointing out that headedness is a complex issue even in languages commonly claimed to have a universal preference for either right-headed (English) or left-headed (French) com-pounds. Upon closer scrutiny, such preferences are typically found to be mere statistical tendencies, with numerous examples of compounds not con-forming to the putative preferred headedness model (for a detailed discussion of English, see Bauer and Renouf, 2001 ).

To sum up, Mandarin compounds can be linguis-tically classified in numerous ways, for instance, based on the syntactic category of the whole word, its constituents, or the meaning and structural relations between the constituents (including head-edness). The adoption of specific taxonomies will largely depend on the goals of individual analy-ses. We may rightly ask whether linguistic theo-ries have any relevance for language representation and processing. Bearing in mind the complex and often uneasy relationship between theoreti-cal linguistics and psycholinguistics ( Phillips and Wagers, 2007 ), the psychological reality of linguistic models cannot be taken for granted. However, to the extent that most current linguistic models do not arise in a theoretical vacuum, but are based on mentalist and cognitivist assump-tions about language and mind, they are often a plausible starting point for psycholinguistic investigations.

3 The Psycholinguistics of Mandarin Compounds

In this section, I summarize the most important findings regarding Mandarin compound process-ing in the visual modality. First, I consider whether words are at all represented in Mandarin. Then, I identify a number of factors which have been shown to influence compound representation and processing in Mandarin and other languages studied to date.

3.1 Reality of the word level and the role of morphemes A fundamental question in any consideration of the psycholinguistics of compounds is, ‘What is the basic unit of lexical representation?’ To an English speaker, the answer seems straightforward — the word. However, it is often emphasized in the litera-ture ( Chao, 1968 ; Hoosain, 1992 ; Li, 2007 ; Packard, 2000 ) that the z ì 字 (character or morpheme) is the unit viewed by many speakers as the basic building block of Mandarin (‘the sociological word’) rather than the cí 词 (word). In reading Mandarin, it is important to distinguish between lexicalized units ( hóngchá 红茶 ‘red + tea = black tea’) and phrasal combinations ( hóngchē 红车 ‘red + vehicle = a red vehicle’). Studies which ask Mandarin speakers to divide text into words ( Bassetti, 2005 ; Hoosain, 1992 ; Tsai et al. , 1998 in Myers, 2007 ) demonstrate that, at least under certain experimental conditions (offline judgments), there is no perfect agreement among speakers in their (explicit) awareness of word boundaries. This is largely attributable to the nature of the Chinese script, which makes morphemes visu-ally salient while obscuring word boundaries (no spaces are used between words in a sentence), and may be further enhanced by the fact that, despite the preponderance of compounds in Mandarin in type counts, monosyllabic words comprise more than half the tokens in a running text ( Xiao et al ., 2009 ).

Although words are not visually salient in Chi-nese text, the existence of a word level in Mandarin is demonstrable both through linguistic argumen-tation and experimentation. For an overview of the linguistic arguments (the most important being

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



that Mandarin syntactic rules operate on words, not morphemes), see Packard (2000) . Packard also points out that a model of lexical access in spoken Mandarin exclusively based on morphemes would be extremely inefficient (if at all possible). Fur-thermore, since even fully transparent, endocentric compounds cannot be reduced to the meanings of their components, compounds must have (a mini-mum of) a whole-word conceptual representation. Also, because a compound's syntactic category is not always predictable from its constituents, as in the [N + N] adverb tiāntiān 天天 ‘day + day = every day’ (note that there is no obvious reason why this word would not be a noun meaning ‘a few days’ or ‘a long time’), at least some compounds must be represented at the lemma level, which, among other things, specifies a word's syntactic properties ( Levelt, 1989 ).

Apart from thought–experiment evidence, numer-ous experimental findings point to the psychological reality of whole-word representations of Mandarin compounds. For instance, paralleling findings from other languages ( Reicher, 1969 ), there is a word-superiority effect in Mandarin; when participants are asked to verify a target character in visual displays consisting of two real Chinese characters, responses are significantly faster for real words than non-words ( Mattingly and Xu, 1994 ). A similar advan-tage for real words is found when participants are asked to pronounce a character in two-character combinations ( Liu, 1988 ).

Taft (2003) found that Mandarin speakers are equally fast and accurate in verifying characters regardless of whether these characters represent sub-morphemic units, bound morphemes, or free mor-phemes. However, when asked to judge whether a character can be a word, participants perform best (in terms of accuracy) with free roots, demonstrat-ing the role of word-status in the Mandarin mental lexicon. Taft and Zhu (1997) found that masked single-character primes identical to the second character of a target (binding word or compound) speeded up word naming, thereby pointing to the existence of whole-word representations on some level in the mental lexicon. Since most of the tar-get compounds were not semantically transparent, it is not likely that the observed facilitation effect

was due to the semantic priming of morphemic rep-resentations without the existence of whole-word representations which bind these morphemic repre-sentations at some level in the lexicon. An additional indicator of the psychological reality of whole-word representations in Mandarin is Janssen et al .'s (2008) finding that in Mandarin (as in English) only whole-word frequency influences the speed of com-pound production in picture-naming tasks (but see their article for findings that constituent frequen-cies may also play a role in compound production). Further evidence comes from Liu and Peng's (1997) study, in which compound targets were preceded by compound primes with either a short (43 ms) or long (143 ms) stimulus-onset asynchrony (SOA). When the whole words were semantically related, there was a processing speed-up at both SOAs, while primes in which only one constituent was semantically related to the entire target only had an effect at the longer SOA. This not only implies that whole-word semantic representations exist, but also that they can be accessed before corresponding morphemic representations.

There is, then, both theoretical and experimental evidence for the psychological reality of Manda-rin words. Of course, the existence of whole-word representations does not preclude the existence of morphemic representations; morphemic effects in the processing of spoken and written Mandarin are substantial, and any model of lexical representa-tion must account for these effects (e.g. Zhou and Marslen-Wilson, 1995 ; Liu and Peng, 1997 ; Taft and Zhu, 1997; Zhou et al ., 1999 ).

There may well not be a single answer to the long-standing debate regarding morphological com-positionality in lexical representation and access in languages (cf. Taft and Forster, 1975 ; Taft, 1981 ; Butterworth, 1983 , for early examples of oppos-ing views), although there are strong indications that parsimony is not the operating principle of the human morphological processor. Humans seem to maximize opportunities for word recognition by being sensitive to both whole words and their pos-sible constituents ( Libben et al. , 1999 ; Kuperman et al. , 2009 ). The research on Mandarin summarized in continuation best fits a position that the amount of (de)composition in morphological storage and

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


processing is modulated by factors such as fre-quency, semantic relationships between a complex word and its constituents ( Libben, 1998 ), and oth-ers. It should be noted, however, that most psy-cholinguistic research on Mandarin compounds is done with noun–noun compounds ( Myers, 2007 ), which may limit the generalizability of the existing findings.

3.2 Statistical properties of the lexicon Word frequency has long been known to influ-ence lexical access ( Foss, 1969 ; Rayner and Duffy, 1986 ; Whaley, 1978 ). Apart from the frequency of a word form, the summed frequencies of a word's inflected forms (e.g. a noun's singular and plural) have also been shown to play a role in lexical access ( Baayen et al ., 1997 ). Another important variable is morphological family size (the number of inflected and derived forms containing a word). In a study of monomorphemic Dutch nouns, Schreuder and Baayen (1997) observed faster lexical decisions for words with larger families. They found no effect of family frequency (the combined frequency of a morphological family's members). Extending these findings (in Dutch and English), de Jong and col-leagues (2002) found that the family sizes of both constituents of a compound were correlated to lexical decision times (faster for larger families). They also found that position family frequency (family frequency constrained by position in the string) was a better predictor of decision laten-cies. In fact, for English compounds written with a space (‘high school’), there was only an effect of position family frequency. Kuperman and cowork-ers’ (2009) recent lexical decision and eye-tracking findings with Dutch compounds cast further light on the role of various statistical properties of compounds in lexical access: while whole-word frequency played the most important role in access to compounds, this role was greater with low-frequency left constitu-ents. The effects of whole-word and left-constituent frequency occurred simultaneously. The frequency of the right constituent had a weaker and later effect. Also, the right constituent was less important in the recognition of compounds whose left constituents

occurred in a small number of words. Apart from once again underscoring the reality of whole-word compound representations, these findings echo the foregoing conclusion about the parser's preference for maximization of opportunity. Also, they con-stitute a refinement of Liu and Peng's (1997) find-ings about the time course of access to morphemic representations.

In Mandarin, the calculation of morpheme fre-quency is not straightforward ( Myers, 2007 ) because a written character often has different meanings in different words and because it is not always easy to tell whether these divergent meanings are cases of polysemy (one morpheme with various meanings) or homonymy (different morphemes sharing a form). Therefore, as Myers warns, studies purporting to manipulate morpheme frequency in Mandarin often actually manipulate character frequency. This is a limitation of the research summarized in this sec-tion and will need to be addressed in the future, as it is a conflation of spoken and written modalities as well as of character and morpheme.

Moving on to studies of Mandarin compound processing, a study by Zhang and Peng (1992) found that both whole-word and character frequency affected lexical decisions to visually presented com-pounds. Another study using the visual lexical deci-sion task ( Peng et al. , 1999 ; Experiment 1) revealed a more complex pattern of results: both the whole-word (or surface) frequency of the compound and its cumulative frequency (the summed frequency of the compound and its constituent morphemes) were related to response latencies. However, cumulative frequency only had an effect for compounds with high whole-word frequency, while there was no such effect for words with low whole-word frequency. Thus, it seems that low-frequency compounds are preferentially accessed as wholes and/or that con-stituents in such compounds become activated later. The role of constituent frequency in visual lexical access in Mandarin was further demonstrated by Taft et al. (1994) , who found that words with two high-frequency characters were verified faster than words containing one low-frequency and one high-frequency character. Surprisingly, words with two low-frequency characters were verified as fast as those

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



with two high-frequency characters. However, since this last group of words included binding words, this likely mirrors Kuperman and colleagues’ (2009) finding that initial constituents found in a small number of words decrease the need for the recogni-tion of the second constituent; as both characters in a binding word are found in only one word, the rec-ognition of the first character is all that is required for the entire word to be recognized.

To sum up, both whole-word and constituent fre-quency have a demonstrable effect on visual lexical access in Mandarin. Overall, whole-word effects are more pervasive. The role of constituent frequency is more limited; particularly in the case of non-initial constituents. The above findings on lexical process-ing in the visual modality largely agree with results from the auditory modality, where whole-word fre-quency effects are also more robust than constituent frequency effects, with the additional finding that the frequency of the first syllable slows down responses to auditorily presented Mandarin com-pounds ( Zhou and Marslen-Wilson, 1994 ). This is in agreement with the well-known principles of the Cohort Model of spoken word recognition ( Marslen-Wilson, 1987 , 1990 ), as words with shared onsets are cohort competitors.

3.3 Semantic transparency While compounds may appear to be ideal candi-dates for compositional processing, their meanings are extremely vulnerable to semantic drift ( Libben, 1998 ). Not only does this greatly increase the theo-retical need to postulate whole-word storage, but it also raises the question of how various semantic properties of compounds influence their storage and processing. Libben (1998) proposes a taxonomy of compound semantic transparency that takes into account two dimensions, which he terms ‘trans-parency’ and ‘componentiality’ (I only discuss the former here). He defines transparency as the relation-ship between a compound's meaning and the indi-vidual meanings of its components. Thus, there is a potential four-way classification of bimorphemic compounds: (1) transparent–transparent (TT) com-pounds, such as ‘blueberry’, in which the independent meanings of both ‘blue’ and ‘berry’ are obviously

related to the meaning of the entire word (‘a kind of berry that is blue’); (2) opaque–transparent (OT) compounds, such as ‘strawberry’ (it is a berry but the connection with ‘straw’ is only transparent to those who know a bit about farming); (3) transparent–opaque (TO) compounds, such as ‘jailbird’, in which only the first constituent has a meaning related to the meaning of the whole compound; and (4) opaque–opaque (OO) compounds, such as ‘buttercup’ (it is not a cup and it is not used for storing butter). Unfortunately, not all studies investigating the role of semantic transparency in lexical processing use a fine-grained taxonomy such as this one (although, admittedly, operationalizations of transparency are becoming increasingly sophisticated).

An oft-cited investigation of the role of trans-parency in compound processing is Sandra (1990) . In this study, transparent Dutch compounds were verified faster when preceded by primes which were semantically related to one of their constituents ( peer ‘pear’ preceding winterappel ‘winter apple’) while opaque compounds ( peer ‘pear’ preced-ing aardappel ‘earth + apple = potato’) or pseudo-compounds were not. However, as Libben (1998) remarks, Sandra's compounds were not uniformly opaque (e.g. aardappel is OT in Libben's taxonomy). In a reversal of the above design, Zwitserlood (1994) used compounds as primes and had participants make lexical decisions to semantic associates of the compounds’ constituents. Zwitserlood's opaque compounds were subdivided into a fully opaque ( klokhuis ‘clock + house = core of an apple’) and par-tially opaque ( drankorgel ‘drink + organ = a drunk’) group. In this experiment, both transparent and par-tially opaque compounds primed their targets, but fully opaque compounds and pseudocompounds did not. In contrast, studies using a lexical decision task with constituents (rather than semantic associates) used as primes and compounds as targets have found priming effects for all types of compounds in Libben's taxonomy for English and French and for all types except OO for Bulgarian ( Libben et al. , 1997 ; Jarema et al. , 1999 ). This indicates that, in the visual modality, some constituent activation occurs regardless of compound transparency, at least at the level of form ( Jarema et al. , 1999 ).

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


Moving on to Mandarin, another finding of the aforementioned study by Peng et al. (1999) was that when primes and targets shared the ini-tial character (identical pronunciation, but repre-senting different morphemes, e.g. ānníng 安宁 ‘peaceful + peaceful = peaceful’ and ānzhuāng 安装 ‘install + install = install’), only transparent masked compound primes could speed up lexical decision times to semantically unrelated compounds. The observed effect was probably not caused by mere overlap in form, as the primes were masked and the fonts used for prime and target presentation were different. Moreover, as Myers (2007) points out, a pure form-based effect would also have been pro-duced by the opaque primes. Therefore, it would appear that the transparent compounds in this exper-iment were decomposed and that activation then spread from the representation of the first morpheme in the prime to that of the first morpheme in the target, most likely mediated by the shared ortho-graphic and/or phonological representations.

Liu and Peng (1997) , in an additional experi-ment, used semantically unrelated compounds as both primes and targets with an SOA of 86 ms. Either the first or the second constituent of the prime and target were semantically related (but were dif-ferent morphemes, written with different charac-ters, and pronounced differently). Only transparent primes speeded up lexical decisions to targets. Taken together, the findings of Liu and Peng's three experi-ments suggest that, when reading Chinese charac-ters, constituent meanings become available faster in transparent than in opaque compounds. In the case of opaque compounds, it is possible that the meaning of the entire compound, which is activated first, com-petes with constituent meanings, thereby delaying their availability. Moreover, Liu and Peng compare their results to Zwitserlood's (1994) , which showed no activation of constituent meanings at an SOA of 300 ms, and conclude that constituent meanings become available earlier in Mandarin than in Dutch, which is likely a consequence of the visual saliency of morphemes in Mandarin text.

In a visual lexical decision task, Myers et al. (2004) presented [N + N] compounds consisting of free roots in separate blocks of fully transparent (TT) or fully opaque (OO) compounds. When a TT block

preceded an OO block, there was no difference in reaction times. However, when the order of blocks was reversed, reaction times for TT compounds dropped off. Myers and colleagues remark that this may mean that decomposition may not be very useful in lexical access (it can slow down access to opaque compounds while not helping transparent ones too much, since whole-word representations are avail-able for these anyway). Therefore, they conclude that prelexical decomposition is optional and that it can be switched on or off depending on the input. Such a conclusion agrees with what is known about the effects that different proportions of certain types of stimuli can exert on morphological decomposition in the visual presentation of English prefixed and pseudoprefixed words ( Rubin et al. , 1979 ).

In a second experiment, Myers and colleagues (2004) used a visual lexical decision task with what they termed ‘target-internal priming’. This con-sisted of flashing one of the black characters in a bimorphemic compound red either early (0 ms) or late (200 ms) into stimulus presentation. All four of Libben's transparency types (1998) were used. With early flashing of the first character, process-ing was slowed down for OO and OT compounds and speeded up for transparent TT compounds. With late flashing of the second character, processing was speeded up only if the first character was transpar-ent. The authors interpret this pattern of results as a sign of left-to-right processing in reading, as the effects of flashing the first character occurred before those of flashing the second character and since the effects of the transparency of the first charac-ter were more important. These results also agree with Kuperman and colleagues’ findings (2009) discussed above.

To sum up, in reading, the morphological trans-parency of compounds appears to be more important when it comes to the activation of constituent mean-ings and less so when it comes to form. Furthermore, semantic transparency mediates access to homo-phonic/homographic Mandarin morphemes. It also appears to be able to switch morphological decom-position on or off during lexical access. Finally, we saw evidence of transparency interacting with other properties of compounds ( Myers et al. , 2004 ). This is further elaborated on below.

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



3.4 Other factors: position in the string, head status, free versus bound status, syntactic category of constituents, and relations between constituents Apart from the above (relatively well-investigated) factors, a number of additional variables are known to influence compound processing, such as a mor-pheme's position in the string, head status, free versus bound status, the syntactic category of a compound's constituents, and relations between con-stituents ( Gagné and Spalding, 2007 ; Jarema, 2007 ; Myers, 2007 ). Research on these factors and their interactions is slowly accumulating. For instance, recall that de Jong and colleagues (2002) found that family frequency constrained by position in the string was a better predictor of speed of access to Dutch and English compounds than constituent fam-ily size. (In fact, position family frequency was the only such predictor for English compounds written with a space.) Investigating French and Bulgar-ian compounds using constituent priming, Jarema and colleagues (1999) found that position in the string, headedness, and transparency interacted in complex ways. For example, in left-headed French compounds (regardless of transparency), position in the string and headedness combined to yield a stronger constituent priming effect for the first con-stituent relative to the second, whereas there was no such effect for right-headed OT French compounds. Libben and colleagues ( 1997 in Jarema et al ., 1999 ) is another study illustrating the complexity of the interaction between transparency, position in the string, and headedness. See also Kehayia and colleagues (1999) , a study with Greek and Polish speakers, which found an advantage for the first con-stituent despite the fact that the head was always in second position.

As pointed out earlier, Taft et al . (1994) found that Mandarin words with two low-frequency char-acters were responded to equally fast as those with two high-frequency characters, while both were faster than words with one high and one low-frequency character. Recall that this effect is likely attributable to the binding words among their LL stimuli. Subsequently, Taft and Zhu (1995) found

that first-position characters in binding words were named faster than second-position characters. Cru-cially, no such effect was found with bound (mor-phemic) characters which only appear in first or second position. Also, the same study found that position in the string does not seem to play a role in a character decision task. Thus, it would appear that the formal properties of bound morphemic characters are available even when their position across compounds is fixed, while for binding char-acters only the orthographic (but not phonological) representation appears to be available independently of position. Phonological representations of binding characters, however, appear to be accessible only through whole-word representations. Taft (2003) found that binding characters were easiest to reject as potential words, followed by bound characters, and that while error rates were high, free charac-ters produced fewer errors than bound ones. Taft remarks that this is probably because both free and bound morphemes have lemma representations in Mandarin, but that representations of bound mor-phemes are less precise, resulting in more errors. Paralleling these results, in a study of morphologi-cal awareness in Mandarin-speaking preschoolers, Hao et al. (2009) found that awareness of free mor-phemes was easier for young Chinese children to attain than that of bound ones.

Zhou and colleagues (1999) conducted a set of lexical decision experiments with masked and visual repetition priming using two-character com-pounds as primes and targets. The experimental manipulations and pattern of results were quite intri-cate (and, notably, featured consistent and robust morphological priming), but there are two findings which are particularly relevant here: (1) with masked primes, morphological priming was stronger if the shared morpheme occupied the same position in both compounds; and (2) although homophonous mor-phemes written with the same character consistently primed one another, they inhibited each other at a long SOA when both in initial position. The results of Myers and colleagues’ (2004) second experi-ment, outlined above, constitute another example of the complex interaction of position in the string with other variables. Position in the string is also a very important factor in the processing of Mandarin

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


compounds in the auditory modality, even more so than in the visual modality ( Zhou and Marslen-Wilson, 1995 ), due to the nature of auditory lexical access ( Marslen-Wilson, 1987 , 1990 ).

To complete this brief survey of the less well-investigated factors influencing the representation and processing of Mandarin compounds in the visual modality, one last variable needs to be considered: the syntactic category of compound constituents. Hsu et al. (1998) had participants judge the syntactic category of visually presented disyllabic compounds (noun or verb). They found that the speed and accu-racy of the judgements varied based on the syntactic category the constituents. For compound nouns, the judgements were the fastest and most accurate for [N + N] N compounds, followed, in declining order, by [V + N] N, [N + V] N , and [V + V] N . For verbs, [V + V] V were judged the fastest and most accurately, fol-lowed by [V + N] V and [N + V] V . Thus, the syntactic category of a compound's constituents is an example of another theoretical linguistic property which hap-pens to have psychological reality. At any rate, all of the less commonly studied variables discussed in this section have received insufficient attention from psycholinguists. Further cross-linguistic inves-tigation of these factors’ role in lexical representation and processing (as well as their intricate interactions) is required.

4 The Chinese Script and Mandarin Compound Processing

Bearing in mind the focus of Writing Systems Research , this article cannot conclude without explicitly addressing the issue of the extent to which the Chinese script influences compound processing in Mandarin. Due to the modality-neutral nature of the mental lexicon, it would be unrealistic to expect these effects to be so drastic as to completely alter its structure or the way it operates. After all, human language evolved in the auditory modality, and speech clearly takes precedence over writing in lan-guage processing. On the other hand, the well-known sweeping changes which literacy brings to cognition (phonological awareness, morphological awareness,

an increase in vocabulary size, and the use of more complex sentence structure, to name but a few) can-not be ignored. However, the focus of the present article is lexical access to compound words and their constituents during normal language use. How, then, might the Chinese script influence these (arguably core) aspects of lexical processing?

First, we have seen that despite the salience of morphemes and the obscurity of word boundaries in Chinese text, words are psychologically real in Man-darin. However, the visual salience of morphemes is certainly an important factor in Mandarin speakers’ awareness of morphemes. While the English alpha-bet makes words salient and morphemes obscure, the reverse is true of the Chinese script. It is, there-fore, not surprising that the demonstrable two-way relationship between morphological awareness and success in learning to read (see Kuo and Ander-son, 2006 , for a recent comprehensive review with a cross-linguistic focus) is particularly pronounced in Mandarin and that it even eclipses the universally important phonological awareness. However, both an English and a Mandarin speaker will essentially still end up having very similar internal lexicons in which, among other things, both words and mor-phemes will be relevant processing units. Second, the Chinese script can be a useful crutch to Manda-rin speakers when ambiguities arise in conversation (due to homophony), as they can refer to a char-acter, often by drawing it in the air. Still, however useful, this disambiguating strategy is not essential. Otherwise, oral communication would be impossible in Mandarin! Third, there are some indications of a certain degree of influence of characters on the pro-cessing of spoken Mandarin. For example, Myers and Gong (2002 in Myers, 2007 ) found a positive effect of word frequency and a negative effect of charac-ter frequency on lexical decision times to auditorily presented disyllabic Mandarin compounds. Coupled with the slight discrepancy between the patterning of the effects of homophonic/homographic ver-sus homophonic/heterographic primes attested in Experiment 3C in Zhou and Marslen-Wilson (1995) , one is led to conclude that the potential impact of Chinese orthography on spoken word processing in literate speakers certainly merits further inves-tigation. Fourth, as seen in Liu and Peng (1997) ,

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



compound constituent meanings in Mandarin seem to become available earlier in the course of visual lexical access than in languages written in alpha-betic scripts, such as Dutch. There may well be other important effects of the Chinese script on real-time lexical processing still waiting to be discovered.

5 Conclusion

In summary, this article outlined several notable linguistic properties of Mandarin compounds and presented an overview of important psycholinguis-tic factors which influence their processing in the visual modality. We have seen that properties of compounds identified by theoretical linguists often have a bearing on lexical processing. (Note, though, that such parallels should never be automatically assumed to exist without experimental verification.) It was argued that the basic unit of the Mandarin mental lexicon is the word and not, as we may be led to believe based on the morphosyllabic nature of the Chinese script, the morpheme. As is the case with other languages, a number of statistical properties of the lexicon were shown to influence compound processing in Mandarin, most notably whole-word frequency. Semantic transparency was also shown to be of relevance, particularly with regard to access to meaning. Finally, it was pointed out that a number of other factors, such as the position of a morpheme in the compound, head status, free versus bound sta-tus, the syntactic category of the constituents, and relations between them, are either known or can be expected to play a role in the processing of Mandarin compounds. These ‘other’ factors interact with one another and with transparency in complex ways.

Further research is needed in both Mandarin and other languages to help untangle these factors and demonstrate in a still more systematic fashion how each of them operates and is modulated by other variables. With its lack of strong preference for compound head position, Mandarin certainly offers a fertile testing ground for these purposes. Kuperman and colleagues’ (2009) study under-scores the ability of eye-tracking, due to its excellent temporal resolution, to answer questions about the interaction of various factors in visual access to

compounds and their precise role at different points in time. Eye-tracking research will certainly also be of great use in revealing precisely at what point and under what circumstances compound constitu-ents (as opposed to whole-word representations) are accessed. A particularly important question is whether the fact that Chinese compounds are visu-ally much more compact than the long Dutch com-pounds used by Kuperman and colleagues influences the time course of constituent activation and how.

Another issue that will require further exami-nation is the role played by the syntactic category of a compound's constituents and the entire com-pound on representation and processing. In light of Ceccagno and Basciano's (2007) classification, a further question which requires attention is whether there are differences in processing depending on a compound's subordinate, attributive, or coordinate as well as endocentric versus exocentric structure. The meaning relations between a compound's con-stituents ( Gagné and Spalding, 2007 ) also require further investigation. Furthermore, as evidenced by the reviewed studies, most research on the process-ing of Mandarin compounds in the visual modality is done with bimorphemic compounds. While this is justified by the preponderance of such compounds in the language, more systematic investigation of com-pounds containing larger numbers of morphemes is necessary. Trimorphemic (or longer) compounds introduce another level of complexity: the fact that compounds are hierarchically structured and not ‘flat’ ( Fabb, 2005 ). For instance, future research will need to investigate whether right-branching trisyllabic (or longer) Mandarin compounds (e.g. xiǎotíqín 小提琴 ‘[small + [lift + instrument]] = violin’) are processed differently than left-branching ones (e.g. diànhuàtíng 电话亭 ‘[[electricity + speech] + pavilion] = telephone booth’). Finally, with a view to further enhancing the cross-talk between theoretical linguistics and psycholinguis-tics, a question worth investigating more systemati-cally is whether the theoretical distinction between derivatives and compounds in Mandarin has any bearing on processing. If it does, this could serve to support the case for maintaining this theoretical distinction in formal morphological descriptions of Mandarin.

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


By way of a conclusion, it is worth noting that much of the research to date has failed to system-atically control all variables which evidently influ-ence compound processing (not least the effect of experimental stimuli on type of processing demon-strated by Myers and colleagues, 2004 ). All this has likely led to confounds and would explain many of the occasional ambiguous or conflicting findings. A further factor bound to influence the quality of data in studies of Mandarin lexical processing is the fact that Mandarin is a second language for a large proportion of Chinese, which opens the possibility that many participants in the studies reviewed here may have been native speakers of another Sinitic language who had only acquired Mandarin in school. Bearing in mind the substantial differences in lan-guage processing between first and second-language speakers (e.g. Abutalebi et al. , 2005 ; Paradis, 2004 , 2009 ), this is by no means a trivial detail and needs to be taken into serious consideration. One of the rea-sons that this issue has largely gone unnoticed might be that there is a widespread perception that Chinese is a single, unified language. Finally, while the exist-ing research bias towards visual stimuli has served the purpose of the present article well, more research in the auditory and a combination of auditory and visual modalities (e.g. cross-modal priming) will be necessary in the future. Not only will this advance our understanding of lexical processing in Mandarin, but it will also help zero in on those aspects of pro-cessing which are specific to the visual modality in general and the Chinese script in particular.

Acknowledgement I thank Xi Chen, Benedetta Bassetti, and three anonymous reviewers for their insightful comments on earlier drafts of this article.

References Abutalebi , J. , Cappa , S. F. , and Perani , D. ( 2005 ). What can

functional neuroimaging tell us about the bilingual brain? In Kroll , J. F. and de Groot , A. M. B. (eds), Handbook of Bilingualism: Psycholinguistic Approaches . New York : Oxford University Press , pp. 497 – 515 .

Arcodia , G. F. ( 2007 ). Chinese: a language of compound words? In Montermini , F. , Boyé , G. , and Hathout , N. (eds), Selected proceedings of the 5th Décembrettes: Morphology in Toulouse . Somerville, MA : Cascadilla Proceedings Project , pp. 79 – 90 .

Aronoff , M. ( 1976 ). Word Formation in Generative Grammar . Cambridge, MA : MIT Press .

Baayen , R. H. ( 1992 ). Quantitative aspects of morphological productivity . In Booij , G. and van Marle , J. (eds), Yearbook of Morphology 1991 . Dordrecht : Kluwer , pp. 109 – 49 .

Baayen , R. H. , Dijkstra , T. , and Schreuder , R. ( 1997 ). Singulars and plurals in Dutch: evidence for a parallel dual route model . Journal of Memory and Language , 36 : 94 – 117 .

Bassetti , B. ( 2005 ). Effects of writing systems on second language awareness: word awareness in English learners of Chinese as a Foreign Language . In Cook , V. and Bassetti , B. (eds), Second Language Writing Systems . Clevedon, UK : Multilingual Matters , pp. 335 – 56 .

Bauer , L. ( 1983 ). English Word-Formation . Cambridge : Cambridge University Press .

Bauer , L. ( 2006 ). Compound . In Brown , K. (ed.), Encyclopedia of Language and Linguistics . Oxford : Elsevier , pp. 719 – 26 .

Bauer , L. and Renouf , A. ( 2001 ). A corpus-based study of compounding in English . Journal of English Linguistics , 29 : 101 – 23 .

Bisetto , A. and Scalise , S. ( 2005 ). Classification of compounds . Lingue e Linguaggio , 2 : 319 – 32 .

Booij , G. E. ( 1977 ). Dutch Morphology: A Study of Word Formation in Generative Grammar . Dordrecht : Foris .

Branner , D. P. ( 2001 ). Chinese languages: classical Chinese . In Garry , J. and Galvez Rubino , C. R. (eds), Facts about the World's Languages . New York : H. W. Wilson , pp. 134 – 8 .

Butterworth , B. ( 1983 ). Lexical representation . In Butterworth , B. (ed.), Language Production , vol. 2 . San Diego, CA : Academic Press , pp. 257 – 94 .

Carstairs-McCarthy , A. ( 2002 ). An Introduction to English Morphology: Words and their Structure . Edinburgh : Edinburgh University Press .

Ceccagno , A. and Basciano , B. ( 2007 ). Compound headedness in Chinese: an analysis of neologisms . Morphology , 17 : 207 – 31 .

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



Chao , Y. R. ( 1968 ). A Grammar of Spoken Chinese . Berkeley : University of California Press .

Clahsen , H. ( 2006 ). Linguistic perspectives on morphological processing . In Wunderlich , D. (ed.), Advances in the Theory of the Lexicon . Berlin : Mouton de Gruyter , pp. 355 – 88 .

De Jong , N. H. , Feldman , L. B. , Schreuder , R. , Pastizzo , M. , and Baayen , R. H. ( 2002 ). The processing and representation of Dutch and English compounds: peripheral morphological and central orthographic effects . Brain and Language , 81 : 555 – 67 .

Dressler , W. U. ( 2007 ). Compound types . In Libben , G. and Jarema , G. (eds), The Representation and Processing of Compound Words , 1st paperback edn . Oxford : Oxford University Press , pp. 23 – 44 .

Fabb , N. ( 2005 ). Compounding . In Spencer , A. and Zwicky , A. M. (eds), The Handbook of Morphology , 3rd edn . Malden, MA : Blackwell , pp. 66 – 83 .

Foss , D. J. ( 1969 ). Decision processes during sentence comprehension: effects of lexical item difficulty and position upon decision times . Journal of Verbal Learning and Verbal Behavior , 8 : 457 – 62 .

Gagné , C. L. , and Spalding , T. L. ( 2007 ). Conceptual combination: Implications for the mental lexicon . In Libben , G. and Jarema , G. (eds), The Representation and Processing of Compound Words . Oxford : Oxford University Press , pp. 145 – 68 .

Hanley , J. R. , Tzeng , O. , and Huang , H. S. ( 1999 ). Learning to read Chinese . In Harris , M. and Hatano , G. (eds), Learning to Read and Write: A Cross-Linguistic Perspective . Cambridge : Cambridge University Press , pp. 173 – 95 .

Hao , M. L. , Dronjic , V. , Chen , X. , and Shu , H. ( 2009 ). The Development of Morphological Awareness in Chinese Children: The Role of Semantic Transparency and Morpheme Type . Paper presented at the Georgetown University Round Table 2009, Washington, DC .

Hoosain , R. ( 1992 ). Psychological reality of the word in Chinese . In Chen , H-C. and Tzeng , O. J. L. (eds), Language Processing in Chinese . Amsterdam : North Holland and Elsevier , pp. 111 – 30 .

Hsu , C-F. , Tzeng , O. J-L. , Hung , D. L. , and Tai , J. H-Y. ( 1998 ). Syntactic Effects at Sublexical level of Word Recognition in Chinese . Paper presented at the First International Workshop on Written Language Processing, New South Wales University, Sydney, Australia .

Huang , S. ( 1998 ). Chinese as a headless language . In Packard , J. (ed.), Compounding Morphology: New Approaches to Chinese Word-Formation . Berlin : Mouton de Gruyter , pp. 261 – 83 .

Institute of Language Teaching and Research . ( 1986 ). 现代汉语频率词典 [ Modern Chinese frequency dictionary ] . 北京语言学院出版社 [Beijing: Beijing Language Institute Press] .

Janssen , N. , Bi , Y. and Caramazza , A. ( 2008 ). A tale of two frequencies: determining the speed of lexical access in Mandarin Chinese and English compounds . Language and Cognitive Processes , 23 : 1191 – 223 .

Jarema , G. ( 2007 ). Compound representation and processing . In Libben , G. and Jarema , G. (eds), The Representation and Processing of Compound Words , 1st paperback edn . Oxford : Oxford University Press , pp. 45 – 70 .

Jarema , G. , Busson , C. , Nikolova , R. , Tsapkini , K. , and Libben , G. ( 1999 ). Processing compounds: a cross-linguistic study . Brain and Language , 68 : 362 – 9 .

Katamba , F. and Stonham , J. ( 2006 ). Morphology , 2nd edn . New York : Palgrave Macmillan .

Kehayia , E. , Jarema , G. , Tsapkini , K. , Perlak , D. , Ralli , A. , and Kadzielawa , D. ( 1999 ). The role of morphological structure in the processing of compounds: the interface between linguistics and psycholinguistics . Brain and Language , 68 : 370 – 7 .

Kuo , L-J. and Anderson , R. C. ( 2006 ). Morphological awareness and learning to read: a cross-language perspective . Educational Psychologist , 41 : 161 – 80 .

Kuperman , V. , Schreuder , R. , Bertram , R. , and Baayen , R. H. ( 2009 ). Reading of polymorphemic Dutch compounds: towards a multiple route model of lexical processing . Journal of Experimental Psychology: Human Perception and Performance , 35 : 876 – 95 .

Levelt , W. J. M. ( 1989 ). Speaking: From Intention to Articulation . Cambridge, MA : MIT Press .

Li , E. S. ( 2007 ). A Systemic Functional Grammar of Chinese: A Text-Based Analysis . London : Continuum .

Li , C. N. and Thompson , S. ( 1981 ). Mandarin Chinese: A Functional Reference Grammar , Berkeley, CA : University of California Press .

Libben , G. ( 1998 ). Semantic transparency in the processing of compounds: consequences for representation, processing, and impairment . Brain and Language , 61 : 30 – 44 .

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

V. Dronjic


Libben , G. ( 2007 ). Why study compound processing? An overview of the issues . In Libben , G. and Jarema , G. (eds), The Representation and Processing of Compound Words . Oxford : Oxford University Press , pp. 1 – 22 .

Libben , G. , Derwing , B. L. , and de Almeida , R. G. ( 1999 ). Ambiguous novel compounds and models of morphological parsing . Brain and Language , 68 : 378 – 86 .

Libben , G. , Gibson , M. , Yoon , Y. B. , and Sandra , D. ( 1997 ). Semantic transparency and compound fracture . CLASNET Working Papers , 9 : 1 – 13 .

Lin , H. ( 2001 ). A Grammar of Mandarin Chinese . Munich : Lincom Europa .

Liu , I-M. ( 1988 ). Context effects on word/character naming: alphabetic versus logographic languages . In Liu , I-M. , Chen , H-C. , and Chen , M-J. Cognitive Aspects of the Chinese Language , Vol. I . Hong Kong : Asian Research Service , pp. 81 – 91 .

Liu , Y. and Peng , D-L. ( 1997 ). Meaning access of Chinese compounds and its time course . In Chen , H-C. (ed.), Cognitive Processing of Chinese and Related Asian Languages . Hong Kong : The Chinese University Press , pp. 219 – 32 .

Marslen-Wilson , W. ( 1987 ). Functional parallelism in spoken word-recognition . Cognition , 25 : 71 – 102 .

Marslen-Wilson , W. ( 1990 ). Activation, competition, and frequency in lexical access . In Altmann , G. T. M. (ed.), Cognitive Models of Speech Processing: Psycholinguistic and Computational Perspectives . Cambridge, MA : MIT Press , pp. 148 – 72 .

Mattingly , I. G. and Xu , Y. ( 1994 ). Word superiority in Chinese . In Chang , H-W. , Huang , J-T. , Hue , C-W. , and Tzeng , O. J. L. (eds), Advances in the Study of Chinese Language Processing, Vol I: Selected Writing from the Sixth International Symposium on Cognitive Aspects of the Chinese Language . Taipei : Department of Psychology, National Taiwan University , pp. 101 – 11 .

McNaughton , W. and Ying , L. ( 1999 ). Reading and Writing Chinese , 2nd edn . Boston, MA : Tuttle Publishing .

Myers , J. ( 2007 ). Processing Chinese compounds. A survey of the literature . In Libben , G. and Jarema , G. (eds), The Representation and Processing of Compound Words , 1st paperback edn . Oxford : Oxford University Press , pp. 169 – 96 .

Myers , J. , Libben , G. , and Derwing , B . ( 2004 ). The nature of transparency effects in Chinese compound processing . Poster presented at the Fourth International Conference on the Mental Lexicon, Windsor, Canada. Slides

available at http://www.ccunix.ccu.edu.tw/ ∼ lngproc/MyersLibbenDerwing-chintrans.pdf

Nishimoto , E. ( 2003 ). Measuring and comparing the productivity of Mandarin Chinese suffixes . Computational Linguistics and Chinese Language Processing , 8 : 49 – 76 .

Packard , J. L. ( 2000 ). The Morphology of Chinese: A Linguistic and Cognitive Approach . Cambridge : Cambridge University Press .

Paradis , M. ( 2004 ). A Neurolinguistic Theory of Bilingualism . Amsterdam : John Benjamins .

Paradis , M. ( 2009 ). Declarative and Procedural Determinants of Second Languages . Amsterdam : John Benjamins .

Peng , D. , Liu , Y. , and Wang , C. ( 1999 ). How is access representation organized? The relation of polymorphemic words and their morphemes in Chinese . In Wang , J. , Inhoff , A. W. , and Chen , H-C. (eds), Reading Chinese Script: A Cognitive Analysis . Mahwah, NJ : Lawrence Erlbaum Associates , pp. 65 – 89 .

Phillips , C. and Wagers , M. ( 2007 ). Relating structure and time in linguistics and psycholinguistics . In Gaskell , M. G. (ed.), The Oxford Handbook of Psycholinguistics . Oxford : Oxford University Press , pp. 739 – 56 .

Rayner , K. and Duffy , S. A. ( 1986 ). Lexical complexity and fixation times in reading: Effects of word frequency, verb complexity, and lexical ambiguity . Memory and Cognition , 14 : 191 – 201 .

Reicher , G. M. ( 1969 ). Perceptual recognition as a function of meaningfulness of stimulus material . Journal of Experimental Psychology , 81 : 275 – 80 .

Rubin , G. S. , Becker , C. A. , and Freeman , R. H. ( 1979 ). Morphological structure and its effect on visual word recognition . Journal of Verbal Learning and Verbal Behaviour , 18 : 757 – 67 .

Sandra , D. ( 1990 ). On the representation and processing of compound words: automatic access to constituent morphemes does not occur . The Quarterly Journal of Experimental Psychology , 42A : 529 – 67 .

Schreuder , R. and Baayen , R. H. ( 1997 ). How complex simplex words can be . Journal of Memory and Language , 37 : 118 – 39 .

Selkirk , E. ( 1982 ). The Syntax of Words . Cambridge, MA : MIT Press .

Siewierska , A , Xu , J. , and Xiao , R. ( 2010 ). Bang-le yi ge da mang (offered a big helping hand): A corpus study of the splittable compounds in spoken and written Chinese . Language Sciences , 32 : 464 – 87 .

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4



Taft , M. ( 1981 ). Prefix stripping revisited . Journal of Verbal Learning and Verbal Behaviour , 20 : 289 – 97 .

Taft , M. ( 2003 ). Morphological representation as a correlation between form and meaning . In Assink , E. and Sandra , D. (eds), Reading Complex Words . Amsterdam : Kluwer , pp. 113 – 38 .

Taft , M. and Forster , K. I. ( 1975 ). Lexical storage and retrieval of prefixed words . Journal of Verbal Learning and Verbal Behavior , 14 : 630 – 47 .

Taft , M. , Huang , J , and Zhu , X. ( 1994 ). The infl uence of character frequency on word recognition responses in Chinese . In Chang , H-W. , Huang , J. T. , Hue , C-W. , and Tzeng , O. (eds), Advances in the Study of Chinese Language Processing , vol. 1 . Taipei : Department of Psychology, National Taiwan University , pp. 59 – 73 .

Taft , M. and Zhu , X. ( 1995 ). The representation of bound morphemes in the lexicon: A Chinese study . In Feldman , L. B. (ed.), Morphological Aspects of Language Processing . Hillsdale, NJ : Erlbaum , pp. 293 – 316 .

Whaley , C. P. ( 1978 ). Word non-word classification time . Journal of Verbal Learning and Verbal Behavior , 17 : 143 – 54 .

Williams , E. ( 1981 ). On the notions of ‘lexically related’ and ‘head of a word’ . Linguistic Inquiry , 12 : 245 – 74 .

Xiao , R. Z. , Rayson , P. , and McEnery , A ( 2009 ). A Frequency Dictionary of Mandarin Chinese: Core Vocabulary for Learners . London : Routledge .

Xing , J. Z. ( 2006 ). Teaching and Learning Chinese as a Foreign Language: A Pedagogical Grammar . Hong Kong : Hong Kong University Press .

Yin , B. Y. ( 1984 ). 汉语语素的定量研究 , [Quantitative analysis of Chinese morphemes] . 《中国语文》 , 第 5 期 : 第338 – 347页 [Chinese Language, 5: 338–47] .

Zhang , B. and Peng , D. ( 1992 ). Decomposed storage in the Chinese lexicon . In Cheng , H-C. and Tzeng , O. J. L. (eds), Language Processing in Chinese . Amsterdam : North-Holland , pp. 131 – 49 .

Zhou , X. and Marslen-Wilson , W. D. ( 1994 ). Words, morphemes, and syllables in the Chinese mental lexicon . Language and Cognitive Processes , 9 : 393 – 422 .

Zhou , X. and Marslen-Wilson , W. D. ( 1995 ). Morphological structure in the Chinese mental lexicon . Language and Cognitive Processes , 10 : 545 – 600 .

Zhou , X. , Marslen-Wilson , W. , Taft , M. , and Shu , H. ( 1999 ). Morphology, orthography, and phonology in reading Chinese compound words . Language and Cognitive Processes , 14 : 525 – 65 .

Zwitserlood , P. ( 1994 ). The role of semantic transparency in the processing and representation of Dutch compounds . Language and Cognitive Processes , 9 : 341 – 68 .

Dow

nloa

ded

by [

Uni

vers

ity o

f T

oron

to L

ibra

ries

] at

16:

16 1

0 A

ugus

t 201

4

Mandarin Chinese compounds, their representation, and processing in the visual modality

Documents

Transcript of Mandarin Chinese compounds, their representation, and processing in the visual modality