An Overview of Corpus-based Studies of Semantic Prosody

The study of semantic prosody (SP) is a brand new area in linguistic field. It has found great potential applications in dictionary compiling, translation, and second language acquisition etc. This paper reviews SP studies both in and out of China. Some related issues, such as definitions of SP, its origin, development and classification are discussed. Finally, its applications are reviewed.


Introduction
The concept of semantic prosody (SP) was first introduced to the public by Bill Louw in 1993.From then on it has become one of the important concepts in corpus linguistics and has claimed more and more attention from corpus linguists.So far, it has found great potential applications in dictionary compiling, translation, and second language acquisition etc.This paper reviews SP studies both in and out of China.Some related issues, such as definition of SP, its origin, development, application and research methodology, are discussed.Finally, its applications are reviewed.

Defining schemes of semantic prosody
Since Sinclair (1987) first noticed the phenomenon of SP in the collocational behaviour of words, there have been a great variety of defining schemes of SP.In this section, the most popular five are in turn reviewed and the strengths and weaknesses of each defining scheme will be examined critically.

Louw's studies of SP
"Prosody" in the term "semantic prosody" is borrowed from Firth (1957), who used it to refer to phonological colouring which spreads beyond semantic boundaries.To give an example, the word animal has so strong a nasal prosody that the vowel sound of the letter a is endowed with a nasal quality through assimilation, simply because a is closely adjacent to the nasal sound of n.In the same way, lexical items share this particular phenomenon of "prosody" in lexical patterning.Enlightened by Firthian sense of a "prosody", Bill Louw coins the term "semantic prosody" and endows it with its first definition, a "consistent aura of meaning with which a form is imbued by its collocates" (Louw, 1993: 157).
Louw illustrates SP with several examples such as the adverbs utterly, the phrase bent on and the expression symptomatic of, which simultaneously carry negative SP.These three words are followed by expressions which refer to undesirable things, such as destroying, ruining, clinical, depression, multitude of sins, etc.Although Louw's definition, the most frequently quoted one till now, is in line with the linguistic position that form and meaning are inseparable, it has not shed any light on the functions of SP or its pragmatic meaning.Semantic meaning and pragmatic meaning are the two extremes in meaning system, for semantic meaning can be seen as the meaning which arises only from linguistic factors in a piece of communication, while pragmatic meaning is that meaning imposed by the non-linguistic elements which has an impact on communication.

Sinclair's studies of SP
John Sinclair, the father of SP study, defines semantic prosody as "The initial choice of SP is the functional choice which links meaning to purpose; all subsequent choices within the lexical item relate back to the prosody" (Sinclair, 1996: 86).This definition points out three defining features of SP.First, functionality.When a speaker/writer is choosing lexical items to make sensible sentences, in addition to the lexical and grammatical rules which govern the grammaticality of the sentence, what he also needs to take into consideration includes the semantic preference and SP which point to the functions.Sinclair insists that the initial meaning choice will actually be at the functional level of the SP (Sinclair, 1996: 87).
Second, linguistic choice.The combination of every collocation is not in the least arbitrary.But all words are in a mutually selectional relation.
Third, communicative purpose.According to Sinclair, semantic prosodies are "attitudinal and on the pragmatic side of the semantics-pragmatics continuum" (Sinclair, 1996: 87).The right SP is bound to express the attitudes of speakers/writers and their purpose with harmony and explicitness.
Sinclair's definition hits the nail on the head, pointing out straightly that the study of SP should lay particular emphasis on the pragmatic function.However, not a clear clarification of what purposes or functions semantic prosodies play in language use has been provided.

Stubbs' studies of SP
Michael Stubbs defines SP as "a particular collocational phenomenon" and collocation as "the habitual co-occurrence of two or more words" (Stubbs, 1996: 176).Moreover, he classifies SP into three categorizations: negative prosody, positive prosody and neutral prosody.In Stubbs' framework of SP, he highlights the relationship between collocation and SP and emphasizes that they both are subject to the co-selection of lexis.For instance, provide is liable to collocate with words (e.g.assistance, funds, opportunities, relief, etc) in the semantic fields of care, food, help, money, to build up a favorable prosody.
Nevertheless, SP is by no means a simple collocation.Collocation is realized more on the lexico-grammatical level, while SP arises from the pragmatic level.Thus this definition only depicts a partial picture of SP, insufficient to provide the entire synopsis of both semantic meanings and pragmatic meanings.

Partington's studies of SP
Alan Partington (1998: 68) offers a new explanation for the term "semantic prosody".He defines it as "the spreading of connotational coloring beyond single word boundaries".To take the adjective impressive for instance, in English it collocates with items such as achievement, talent and dignity.In this manner, it is considered to have a positive SP (1998: 77).By contrast, rife tends to co-occur with words such as crime, misery and disease; hence its interpretation has a negative SP (1998: 67).
Partington emphasizes on the spreading of connotation of single words through word boundaries, since semantic prosodies sometimes are interpretable in terms of connotations.If the node words and their collocates tend to co-occur frequently, they will acquire the same connotaitonal features, and meaning and form merge.
On the other hand, similar to Louw's definition, Partington is also incomplete because SP is more on the pragmatic side of the semantics-pragmatics continuum.It is not only a route to convey similar connotaitonal meanings, but a strategy in verbal communications.

Hunston and Francis' studies of SP
In Hunston and Francis' framework, a word may be said to have a particular SP if it can be shown to co-occur typically with other words that belong to a particular semantic set (2000: 137).Hunston and Francis regards SP as a phenomenon of node word typically co-occurring with lexical items of the same semantic field.However, this definition does not offer any new elements compared with the previous four definitions.In a sense, the expression semantic set cannot be said to be precise enough.Semantic field is a ubiquitous phenomenon in language.Therefore, this definition is incomplete to give a comprehensive illustration to the properties of SP.
The above five defining schemes approaching the concept of SP can roughly be thought of falling into two major lines of thinking.One addresses the phenomenon from the functional perspective, and the other from the collocative meaning perspective.Sinclair's defining scheme obeys the golden rule of functionality of SP and whereupon belongs to the first line.The rest four defining schemes emphasize the SP from the perspective of collocation and belong to the second line.Each definition of the two lines well serves the research aim of different corpus linguists trying to gain access to further understanding of SP.However they are not able to provide a full yet comprehensive description of SP.

Classification of semantic prosody
Michael Stubbs (1996) proposes that some words have a predominantly negative prosody, a few have a positive prosody, and many words are neutral in this respect.If the collocates that a node word attracts are mostly of strong negative semantic characteristics, the node word bears a strong negative prosody.If the collocates are mainly positive words, then the node word is endowed with a positive prosody.If both positive and negative collocates exist in the context, the node word can be said to bears a neutral or mixed prosody.Here we need to reiterate that semantic prosody differs from the phenomena of commendatory words and derogatory words treated in traditional lexicology.The commendatory and derogatory words or phrases themselves have either a positive or a negative sense.But in a semantic prosody, there is nothing explicitly positive or negative for the node word.It is its characteristic collocates that have a similar particular semantic association.Following this categorization, Sinclair observed that set in has a negative connotation, because its main collocates include rot, decay, malaise, ill-will, decadence, , inflection, prejudice, vicious, rigour, mortis, numbness, bitterness, mannerism, anti-climax, anarchy, etc.Just opposite, career mostly collocates with words like satisfying, successful, good, potential, academic, teaching, etc, so is dominated by positive prosody.Louw (1993) discusses build up and show that as a transitive verb the semantic prosody is favourable (e.g.build up confidence), but as an intransitive verb the prosody tends to be negative (e.g.resistance builds up).He called this kind of phenomenon mixed prosody instead of neutral prosody, although both of the two labels indicate that the whole profile of the node word presents a rather intricate picture.Partington's (2004) classifies SP into favourable, neutral, and unfavourable prosodies.A pleasant or favourable affective meaning was labelled as positive while an unpleasant or unfavourable affective meaning was judged as negative.When what was happening was completely neutral, or the context provided no evidence of any semantic prosody, the instance was labeled as neutral.

Relevant studies of SP: a summary
As is mentioned above, while most English words collocate with negative words as well as positive ones, only a small number of words in English are found to bear strong negative or positive semantic prosody.While most previous significant studies of semantic prosody have been described while reviewing definitions of semantic prosody, here a table adapted from Xiao and McEnery (2006) is given as a complete summary.As can be seen from Table 1, more than 25 lexical items in English have been proved by corpus linguists abroad to bear apparently negative or positive prosody.And it is noteworthy that more items with negative prosody than those with positive prosody have been identified.
In contrast with the considerable explorations of the phenomenon of SP abroad, there is an apparent shortage of this topic in China.Apart from the several significant studies made by Wei Naixing, only a few Chinese researchers have conducted in-depth studies of SP.Nevertheless, the few studies concerning SP are summarized here and their significance or/and weakness are pointed out.
Wei Naixing from Shanghai Jiao Tong University is a pioneer in conducting the study of SP in China.He has published three important essays concerning this topic in domestic magazines.Wei (2002 a) introduces three approaches to the study of semantic prosody: data-based approach, data-driven approach and integrated approach.The significance of this study lies in providing effective approaches for the further and deeper exploration of semantic prosody.Wei (2002 b) examines the semantic prosodies in the specialized texts of JDEST corpus, by observing behaviours of lexical items as CAUSE, incur, utterly, probability and careers.The study has found the word CAUSE has a stronger negative prosody than the case in general English texts whereas the word career has a weaker or less obvious positive prosody than the case in general English texts.This study has set an example for domestic researchers to undertake a study of semantic prosody.However, given the corpus this research is based upon includes only one genre of texts, the result might be applied to English for special purpose but not to general English.Wei (2006) is a significant contrastive study of the semantic prosodic features of the lexical combinations in the Chinese learners' English writings.The study shows that there exist three major types of word combinations in learner's interlanguage, including typical collocations, interlanguage collocations and unusual collocations.The paper argues that collocational acceptability is no less important than grammatical acceptability in second language learning and.Despite the significance and originality of this study, there still exists a weakness.When we take a verb as the subject of a study of collocation or semantic prosody, we should study all the word-forms of the verb, the original form, the third personal form, the past tense, the past participle and the present participle.Only by doing so can we get a whole profile of collocation and semantic prosody of a verb.However, when Wei (2006) studies the semantic prosody of the verbs cause and commit, he merely concordance their original form and neglects the rest forms.Pan and Feng (2003) also introduce the concept of SP, illustrate the exploitation of semantic prosody and discuss its various applications.But this study is far from a systematic and in-depth exploration.Wang and Wang (2005) is a comparative study of English writing made by native speakers and Chinese learners.The study examines the SP of the lemma CAUSE on the basis of CLEC and part of SBNC.The study shows that great differences exist in the semantic prosody of CAUSE between Chinese learners of English and English native speakers.Chinese learners of English underuse the typical negative SP and at the same time overuse the atypical positive SP.However, the study is only confined to one lexical item, which might lack generality and typicality.Besides, it is confined to the SP of CAUSE without adequate attention to its collocation patterns used in English writing of Chinese learners.Xiao and McEnery (2006) explore the SP of near synonyms from a cross-linguistic perspective and compare the phenomena of semantic prosodies existing in Chinese and English.And the study shows that semantic prosody and semantic preference are as observable in Chinese as they are in English.Remarkably, it is the first attempt to study the semantic prosodic behaviour existing in the Chinese language and one of cross-linguistic investigation of SP.However, the study doesn't provide enough statistical data to support his argument.

Implication
The comprehensive review concerning SP in this study can provide the following implications.
First of all, it's suggested that the knowledge for SP should be transferred to second language learners.Due to its nature of subtleness, SP is often hidden from human intuition and so can only be explored by the powerful means of corpus linguistics.Therefore, a DDL (data-driven learning) approach to learn or teach lexis in context should be proposed so that the learners could be aware of the semantic prosodic features of a word.Then there will be fewer occurrences of inappropriate SP and unusual collocation in their English.
Moreover, the knowledge of SP can also provide insight into the teaching of vocabulary, especially near synonyms.In essence, near synonyms are identical or similar in denotational meaning but usually differ in their collocational behaviours and semantic prosodies.
In addition, the incorporation of SP and collocation information into English dictionaries should be proposed.Most contemporary learner dictionaries have not given enough attention to SP and collocation information.The availability of large corpora and concordance software has made it possible to provide adequate information of SP and typical collocation for such dictionaries.

Conclusion
This paper reviews SP studies both in and out of China.Some related issues, such as definitions of SP, its origin, development and classification are discussed.Finally, its applications are briefly reviewed.Semantic prosody is one of the important concepts in corpus linguistics and has claimed more and more attention from corpus linguists.However, the current research concerning SP is still in its preliminary stage.While lots of explorations have been made on the characteristic patterning of SP and its application in language use and second language acquisition, there are still not sufficient systematic and in-depth explorations.Therefore, in the future research we would need to observe more lexical items, look in larger corpora and make a more insightful analysis before we could be reasonably confident of our conclusions.