The differences in usage of have/has to + V and must + V constructions in written English of the 1960s. A distinctive collexeme analysis

Table of content

Introduction

Theoretical background
Collostructional analysis
Distinctive collexeme analysis

Method
Database and CQL search
Collexeme analysis of have/has to + V construction
Collexeme analysis of must + V construction
Distinctive collexeme analysis of have/has to + V and must + V
constructions

Conclusion

References

Introduction

According to Gries and Stefanowitsch, traditional linguistic studies view the lexicon and the grammar of a language as two completely diferent phenomena (Gries & Stefanowitsch, 2003, 209). In their cited publication, they state that “various expression types that fall somewhere in between lexicon and grammar have been recognized but largely ignored by mainstream syntactic theories” (209). Over the past decades, however, a variety of linguists, such as Lewis Hunston, Sinclair, Gries and Stefanowitsch themselves, have published a variety of theories, which approach a diferent view towards the connections between grammar and lexicon. These theories assume that “grammar and lexicon are not fundamentally diferent, and that the long-ignored multi-word expressions serve as an important link between them” (Gries & Stefanowitsch, 2003, 210). Following these theories, pairs of semantically more-or-less corresponding expressions and constructions, as the ones shown in (1) to (3), have occupied the attention of linguists all over the world (Gries & Stefanowitsch, 2004, 97):

(1) Peter gave Mary the ball / Peter gave the ball to Mary
(2) Frank is going to be happy / Frank will be happy
(3) Peter must give Mary the ball / Peter has to give Mary the ball

So, as another example: Do I have to write this term paper or must I? Is there even a semantic diference between the two constructions have/has to + V and must + V ? Epistemically, there is little diference between the two constructions, but a deontic approach shows that the construction must + V obliges the subject of the sentence to do something, while the construction have/has to + V does not play a strong deontic role. Nevertheless, native speakers of the English language use both constructions more or less indiscriminately.

This term paper rejects the common approach that the two constructions are semantically identical, since certain verbs are more likely to occur in the have/has to + V -construction, than in the must + V -construction.

The attraction between the two mentioned constructions and accompanied verbs will be tested through a distinctive collexeme analysis. Distinctive collexemes are verbs, which are most distinctive for a construction (Gries & Stefanowitsch, 2003, 215).

The term paper is laid out in the following way: The frst section will present some theoretical background on both the collostructional analysis and the distinctive collexeme analysis. Afterwards, the second section, which represents the bulk of this term paper, will deal with the corpus-based study itself. Multiple sub-sections will outline the data and source used for the analysis, the analytical tools and required calculations. Finally, the last section will analyse the results of the It study and hopefully lead up to further research.

Theoretical background

Collostructional analysis

In 2003, collostructional analysis was developed by the two German linguists Stefanowitsch and Gries. It is a set of highly time-consuming and mathematically sophisticated corpus-based methods for the measurement of mutual attraction between lexemes and constructions. Its precision and transparency lead to more accurate results than previous methods simply based on raw frequencies. According to Hilpert, “rather than asking which elements occur most often, collostructional analysis uses relative frequency counts to determine which elements occur more frequently in a construction that would be expected by chance” (Hilpert, 2014, 392).

Distinctive collexeme analysis

Distinctive collexeme analysis, as introduced by Gries & Stefanowitsch in 2003, is one out of three related methods within collostructional analysis. It compares multiple constructions with regard to certain lexical items which occur with them. This form of collostructional analysis is highly relevant, when it comes to the comparison of constructions that are almost synonymous, such as will vs. be going to, or must vs. have/has to (Gries & Stefanowitsch, 2003). This term paper will analyse the relationship between the two mentioned constructions must + V and have/has to + V which semantic meanings are almost identical: While must + V is being used whenever a certain action appears to be necessary, have/has to + V refers to an action that is necessary due to laws or other regulations.

As mentioned before, distinctive collexeme analysis “contrasts constructions in their respective collocational preference {and is} particularly suited for the study of related constructions” (Hilpert, 2008, 39). Generally, this type of analysis is based on the assumption that the semantic meaning of a certain word or construction can adapt to its environment (Klimcikova, 2015, 6). The authors of this analysis, Stefanowitsch and Gries, explain their method with the following words:

“Collostructional analysis always starts with a particular construction and investigates which lexemes are strongly attracted or repelled by a particular slot in the construction. Lexemes that are attracted to a particular construction are referred to as collexemes of this construction; conversely, a construction associated with a particular lexeme may be referred to as a collostruct; the combination of a collexeme and a collostruct will be referred to as a collostruction” (2003, 215).

Method

Database and CQL search

In order to run a representative distinctive collexeme analysis, both constructions need to be extraced from the exact same corpus. This particular analysis is based on the Standard Corpus of Present-Day Edited American English, short the Brown Corpus. Compiled in the 1960s, it was the frst computer-readable general corpus of texts for linguistic research on contemporary English. With around 500 samples of over 2000 words each, it contains more than a million words of running text of edited English prose published in the United States during the year 1961. Although some linguists call the Brown Corpus outdated, it had been chosen for this term paper since its size is easier to manage and analyse than the one of bigger corpora, such as the British National Corpus (BNC), which contains over 100 million written and spoken words.

For the below analysis, the online tool SketchEngine had been, in order to extract the two analysed constructions and their related environments from the BNC. SketchEngine features multiple functions, such as frequency lists or listing by node forms.

With the help of a CQL search query, both the types and the tokens of the two constructions could be found within the Brown Corpus. Since there is no past form of the verb must, this term paper only observes the present tense results for both constructions. The following two CQL queries were being used: for have/has to + V constructions and [word = "must"][tag = "VV"] for must + V constructions.

Collexeme Analysis of Have/Has to + V construction

This subsection leads to a list of lexical verbs which are strongly related to the construction have/has to + V. It needs to be mentioned again that the following collexeme analysis is only based on the Brown Corpus, which does not include spoken English and its results are based on written American English of the 1960s.

The following CQL query had been used in order to create a spreadsheet of complete concordance for the featured construction: [word="have|has"][word="to"][tag="VV"]. The query lead to 234 tokens, distributed across 129 diferent types within the Brown Corpus. Despite the token frequency of a certain construction and the raw frequencies of all collexemes, two more frequencies are needed, in order to complete a collexeme analysis: The token frequencies of particular collexemes within the Brown Corpus and the overall token frequency of all constructions that can be found within the Brown Corpus.

Therefore, tag-sensitive CQL searches had to be done, in order to flter out the corpus frequencies of all verbs, which occurred with the observed construction.

[...]