A corpus study on words similar in meaning

Table of content

1. Introduction

2. Hoey’s Lexical Priming Approach
2.1. Collocation
2.2. Colligation
2.3. Semantic Preference
2.4. Semantic Prosody

3. Corpus Linguistics
3.1 Neo-Firthian Approach to Corpus Linguistics
3.2. Sketch Engine for Language Learning
3.3. Limitations of the Study

4. Corpus analysis
4.1 Analysis of to beg
4.2 Analysis of to beseech
4.3 Analysis of to supplicate
4.4 Analysis of to importune
4.5. Analysis of to entreat
4.6. Analysis of to implore

5. Comparison

6. Conclusion

7. Reference
7.1 Electronic reference
7.2 Secondary literature

8. Appendix

1. Introduction

‘You shall know a word by the company it keeps.’ (Firth 1957: 11)

This Firthian principle will guide the following corpus-based study which investigates how the words to beg, to beseech, to supplicate, to importune, to entreat, and to implore, which are similar in meaning, differ in semantic patterns.

Analysing and comparing the words in terms of their linguistic environment will reveal how the specific word is used naturally. Its idiomatic use might be surprising as language is always changing and developing, so one might find out new and unexpected aspects of a words use.

Chapter 2 explains the Lexical Priming approach by Michael Hoey and its significance regarding this study. Furthermore, the four categories collocation, colligation, semantic preference and semantic prosody will be defined in order to lay the groundwork for the analysis in chapter 4. In the following 3rd chapter, corpora will be defined as well as the Neo-Firthian tradition of using corpora for lexical analysis. and information about the source corpus for this study, Sketch Engine for Language Learning, will be provided. The chapter then concludes explaining the limitations of this study. Chapter 4 will state a few conditions and limitations of the analysis and will then continue to present the qualitative analysis the target words modelled after Sinclair’s model of extended lexical meaning. The 5th chapter will compare the results of chapter 4 to give an efficient overview of the semantic similarities and differences of the words. Lastly, chapter 6 will provide a concluding statement of whether they differ from each other at all or if they can substitute each other seamlessly. A final outlook will give additional ideas on how to expand this study.

To quote from the beginning of this chapter entails that the target words will most likely be influenced heavily by the context in which they occur, but there must be a few similarities otherwise they would not be classified as similar words.

2. Hoey’s Lexical Priming Approach

This second chapter will start with explaining Michael Hoey’s theory of lexical priming and will continue with defining the categories collocation, colligation, semantic preference, and semantic prosody, which belong to John Sinclair’s model of extended lexical units (Stubbs 2009: 22). Sinclair was not the only linguist who coined this term, so this chapter will use a variety of definition and well known examples. They account to the varying semantic or pragmatic environments and the combinations of grammatical choices in which words can occur and will be applied in the analysis in chapter 4 (Sinclair 1991:112).

The theory of lexical priming essentially reverses the role of lexis and grammar, offering a new approach which states that grammar is just an outcome of systematically structured lexis. (Hoey 2005: 1). Before scholars such as Firth focused their studies on lexis, linguists usually disregarded and underestimated lexis because the perceived it as a ‘marginal part attached to grammar’ (Jonhansson 2011: 19). Hoey combines corpus linguistics with the psychological concept of repetition to explain word association. Earlier theories describe only possibilities of language, for instance the Lexicalized Grammar by Hunston and Francis or Sinclair’s Idiom Principle. The Lexical Grammar differentiates semantic meanings by different patterns, and while these patterns are mutually dependent with lexis, they are entirely unpredictable. Sinclair’s Idiom Principle states that grammar is what a person uses when collocational and other patterns are not used (Hoey 2009: 41-46). While both of these approaches disregard the naturalness of an expression, they still contribute to the understanding of the relationship between grammar and lexis (Hoey 2005: 2; Hoey 2009: 46).

Hoey’s approach delves deeper into the psychological notions of language and looks at how language is used naturally. Every word a person encounters brings up associations, belonging to, for instance, a certain conversational situation or the knowledge of different contexts. When hearing of the train station Kings Cross in London, one person might remember recent journeys, while the other will immediately think of Platform 9¾. Although there are certain idiomatic expressions, for example it’s raining cats and dogs, which have equal meaning across the world, every person has individual associations which are impossible to understand from mere sentence examples (Sinclair 2003: 51).

There are various combinations of individual collocations. These individual patterns arose from previous experiences with a particular word. Overall, experiences with language shape the way words will be used in future discourses. Co-occurrences of a word with certain grammatical or semantic choices will be reproduced and eventually become natural in use.

This particular approach is important regarding this study, as the notion of lexical association is explained with the psychological concept of repetition. Consequently, lexical units, which are commonly used, are not random, but they were shaped by human interaction with language. In order to understand and correctly use a language, one must understand their idiomatic use, which is why the concordance lists, this study uses, consist of descriptive examples of the target words.

2.1. Collocation

Hoey defines collocation as a ‘psychological association between words [...] up to four words apart and is evidenced by their occurrence together [...] more often than is explicable than is explicable in terms of random distribution’ (Hoey 2005: 5). His definition of the term differs from previous ones in that aspect that he draws attention to the fact that collocations are individually primed. This suggests that the same word can have a range of varying collocations, depending on the language user.

Phrases, such as to sign a contract, are frequently and intentionally used in the same word sequences by native speakers. A well-known example by Michael Halliday demonstrates that the substitution of the collocate strong in the collocation strong tea with a similar word such as powerful would be perceived as awkward and unnatural by native speakers (Butler: 2003: 177).

The target word is called the node which primes its collocates. Collocates are the co-occurring words connected to the node with a range from a single word up to a complex phrase (Geeraerts 2010: 170). In Halliday’s example, tea would be the node and strong would be its collocate.

2.2. Colligation

Colligations are the co-occurring grammatical patterns of the node and the collocate. Evidently, this means that not only does a node prime other words or phrases, but it also primes relational processes. For example, the phrase one of the most usually appears connected to an adjective in a sentence initial position. Colligations do not only refer to grammatical constructions, but also to the position of the sentence, paragraph or text in which the collocation occurs which is called textual colligation (Hoey 2005: 40-43).

2.3. Semantic Preference

Collocations are able to shift from their immediate environment to a higher level of semantic preference as the co-occurrence of some words is not always explainable by collocation. Hoey describes this phenomenon with the phrase two-hour bus ride. As both of the collocates of the node bus are random and not necessarily connected in their meaning, we must assume that collocations are used in a wider sense of meaning, connoting a certain semantic set. This specific semantic set creates the notion of a journey (Hoey 2005:17,18).

Some semantic sets are primed locally for instance, but in a corpus, these collocations would appear to be random. How these semantic associations were connected to the node is not traceable through corpus study as these associations rely on undocumented and unique experiences with language. (Hoey 2005:19)

2.4. Semantic Prosody

Semantic prosody ‘expresses [the] attitudinal und pragmatic meaning’ of a phrase (Sinclair 2004: 174). The collocates of a target word colour it either negatively or positively (Stewart 2010: 8). When evaluating the semantic set there always additional, pragmatic meaning to be found, which connotes the speakers attitude (Stewart 2010: 20). The target words usually display a preference for either negative or positive connotation, though both can occur.

An example would be the verb to cause, which has been analysed by Michael Stubbs. The verb shows a clear preference for negative consequences, for example accidents, while it would be never associated with a positive emotion or event (Stewart 2010: 28).

3. Corpus Linguistics

As the previous chapter discussed Hoey’s lexical priming, the theoretical basis of this study, this chapter will define corpora, the tool for investigating lexical behaviour. After that, the Sketch Engine for Language Learning corpus which was used for this particular study will be inspected and limitations of the study will be explained.

Corpora are electronic collections of texts. The quantity of words which are accessible through online corpora permit to analyse lexis in detail and discover new theories of language (Kennedy 1998: 3,7). In direct discourse, observing linguistic behaviour is next to impossible due to the lack of a perfect memory. Electronic corpora have solved this problem. The amount of real recorded data, written or spoken, improved the accuracy of defining and observing languages. Before, linguists had to intuitively guess about their work and were not able to prove their ideas immediately with sufficient data (Sinclair 2003: iv).

Some corpora are serving a special purpose, for instance collecting the words most frequently used in a certain time period or supporting language learning. Essentially corpora are used to identify frequent structural patterns and the usage of the particular language system (Kennedy 1998: 4).

Corpora serve merely a representative function and do not contain countless works, but only a sufficient amount of data to represent a certain language (Kennedy 1998: 4).

With the invention of computers, collecting and evaluating corpora became more efficient and faster. There is a vast collection of online corpora available, serving as a reference for the different variations of English. An example would be the British National Corpus which was created in the 1990’s as a balanced source of varying genres, designed for educational, academic and commercial purposes (Kennedy 1998: 5).

Although it was more difficult and time-consuming, Linguists without access to online corpora were still able to inspect language, using for instance the Bible as a source (Kennedy 1998: 13).

3.1 Neo-Firthian Approach to Corpus Linguistics

The Neo-Firthian approach to Corpus Linguistics is based on J.R. Firth’s theory of contextualized meaning. Firth first developed the notion of contextualized meaning after he was influenced by Malinowski, a fellow linguist, who stated that ‘you can make nothing of words in isolation’ (Monaghan 1979 :24). His following studies were concerned with the reality on which language was dependent because ‘any text can be regarded as a constituent of a context in situation’ (Firth 1957: 7). The context he studied, is defined by Firth as follows:

[a] group of related categories at a different level from grammatical categories but rather at the same abstract nature. A context of situation for linguistic work brings into relation the following categories:

A. The relevant features of participants: person, personalities.

(i) The verbal action of the participants.
(ii) The non-verbal actions of the participants.

B. The relevant objects.

C. The effect of the verbal actions. (Monaghan 1979: 32).

J.R. Firth’s principles have been developed and applied since 1957 by neo-Firthian scholars which include Halliday, Sinclair, Stubbs, Hunston and Hoey (Monaghan 1979: 184, Flowerdew 2012: 53). Neo-Firthians analyse whole texts from open-ended corpora to show tendencies of phraseology (Flowerdew 2012: 54). Open-ended corpora are significant because as Hoey stated, language is continually developing and open-ended corpora provide the option to add new phrases and/or texts to sufficiently define all the layers of a language (2005: 10). As a student of Firth, Sinclair prepared empirical ground to investigate Firth’s notions in corpus-based studies and developed the model of extended lexical meaning which is used in this study (Oakey 2009: 141).

3.2. Sketch Engine for Language Learning

Sketch Engine describes itself as an ‘state-of-the-art’ tool useful for linguists, historians, translators, students and teachers. The website provides 400 accessible corpora which each contain up to 20 billion words. (available at https://www.sketchengine.co.uk/#blue)

The Sketch Engine for Language Learning (SkELL) which was used for this paper is a simpler version of this website and was designed particularly for English language learning. Teachers and students can use it to examine how a particular phrase or word is used correctly and how it works in context. This version of Sketch Engine does not require a membership and is freely accessible to everyone. (available at https://www.sketchengine.co.uk/skell/)

SkELL provides not only concordances lists, but also collocates and synonyms of the target words. The concordance list offers 40 sentence examples of the target word to give insight how it behaves in its semantic and syntactic environment. The search engine is case-insensitive and will give results derived from the original word, such as mice for mouse. Furthermore, the user does not have to specify what part of speech he is searching for. With one billion words, SKELL covers the English sufficiently in regard to every, formal, and professional genres. Their sources include the English Wikipedia, Project Gutenberg, enTenTen14, the WebBootCat corpus, as well as the whole British National Corpus (Baisa 2014: 63-68).

3.3. Limitations of the Study

As Sketch Engine only provides a restricted selection of examples for each word, the results will be only depicting the meanings of the target words at based on the data available until this point in time. One must be aware that in the future their meaning can either change or connote additional semantic layers. SkELL provides 40 sentence examples from various sources in its concordance lists and these are regarded as the only 40 examples existent of the nodes. As the quantitative analysis in regard to frequency and representation was already conducted by SkELL, this next chapter will present the qualitative analysis of the target words. Due to the word limitation, this study can only be seen as exploratory.

4. Corpus analysis

Before this following chapter will look at the qualitative analysis of the target words – beg, beseech, supplicate, importune, entreat and implore – there are a few conditions and limitations that must be stated which are due to the limited word count of this study and the source material. This analysis will only look at the most representative and prominent collocations, colligations, semantic preferences, and semantic prosodies because as this is a limited study that investigates primarily differences in semantics but also because meaning is conveyed rather though context than grammar, which was established in the previous chapters.

Each subchapter will start with the table of analysis modelled after Sinclair’s table of the analysis of true feelings (Sinclair 2003: 151). The collocations, colligations and semantic preference of the six target words will be analysed in the table and further explained in the texts following the tables in order to investigate their prosodic meaning.

Not every sentence can and will be used in this analysis as some lack greater context to analyse them thoroughly. It is possible that these sentences, which seem to lack greater context, are individual or local primings associated with the particular word. Some of the examples show exceptions to regular orthography such as non-usage of hyphens or commas.

All of the examples not incorporated in the table will be annotated for reference in the appendix. The frequency of patterns will be stated in percentages.

4.1 Analysis of to beg

This image was removed for copyright reasons

Table 1. The BEG Profile

The Verb to beg appears in 25% of the instances with the noun question. This construction is well known as the idiom to beg the question, which means that an obvious fact is omitted from the particular discourse. But as this fact is important, one discourse partner is stating the obvious need for clarification. Other frequently found nouns that collocate with to beg are personal pronouns or (named) persons in 57,5% of the examples and words expressing that people are begging for money. The collocation of begging for money appears in 37,5% of the instances. Some of these examples depict a converse opinion on begging, which states that it is shameful and illegal.

A prominent colligation of to beg can be seen in the construction of beg to + lexical verb which occurs 20% in the examples. Through this constellation another meaning of to beg is displayed. Next to the already above mentioned meanings, one can beg their discourse partner for a certain action. In 10% of the examples to beg the question appears in an interrogative clause. 60% of the instances demonstrate that to beg favours to appear in the middle part of the sentence.

To beg displays two distinct meanings. With preferring semantic sets concerning money, a clear negative connotation is made, as money also collocates with instances of either legal or social disapproval. The idiom to beg the question displays a preference for appearing in interrogative sentences connected to a preceding discourse, thus it connotes clarification. A further, but less prominent meaning depicts that one can not only beg for money/physical items, but also for actions. These actions, though, would need further context to assign them a clear connotation.

[...]