PART ONE Phraseology and corpus linguistics

Chapter 1 What is phraseology？

1.1 Significance of phraseology

It is no exaggeration to say that phraseology occupies a central position in language use. This important role of phraseology has been recognised by many linguists（e. g. Sinclair 1991,2004a; Hoey 2005; Granger and Meunier 2008; R?mer and Schulze 2009; Stubbs 2009a; Moon 2010; O'Keeffe and McCarthy 2010; Hunston 2011; Granger et al. 2013）. Advances in technology and an increasing use of corpus-based methods have also made the research on different aspects of phraseology more feasible and easier, e.g. automatic extraction of certain phraseological items or studies on phraseology using large quantities of data. Various academic disciplines such as discourse analysis, psycholinguistics, sociolinguistics and multilingualism have taken phraseology as a starting point to investigate features of language, mind and society（see Deignan 2005; Hoey 2005; Mahlberg 2005; McEnery et al. 2006;Granger and Meunier 2008; Murphy 2010; O'Keeffe and McCarthy 2010; Hunston 2011; McEnery and Hardie 2012; Granger et al. 2013; John and Laso 2013）.

Phraseology is a pervasive phenomenon in language use（see Sinclair 1991, 2004a; Howarth 1996; Cowie 1998; Meunier and Granger 2008; McEnery and Hardie 2012）. For example, Altenberg（1998）has suggested that over 80% of the words in the London - Lund corpus of spoken English are involved in recurrent word combinations（cf. Pawley 2007:20）. Greaves and Warren（2010:221）further argue that if the“multiword units with constituency and or positional variation”（e.g. concgrams or collocational frameworks）are taken into consideration, the figure related to the phraseological nature of language could be“closer to 100 percent”. Similarly, the current corpus-based analysis of time and thing suggests that these two words are associated with a strong phraseological tendency, i.e. they tend to occur in fixed phrases or patterns（cf. Sinclair 1991,1996）. Approximately 96% of the concordances of time in a random corpus sample of 500 lines contain recurring phraseological items; and around 79% of the concordances of thing in a 500-line corpus sample involve the occurrences of multi-word phrases. Thus, it can be reasonably argued that language itself exhibits a strong phraseological nature（see Sinclair 1991,2004a; Cowie 1998;Wray 2002; Hoey 2005; Ellis 2008; R?mer and Schulze 2009; Moon 2010; Stubbs 2011; Granger et al. 2013）.

Phraseology can also reflect the linguistic features of a certain discourse. For instance, the phraseological items which occur frequently in conversational discourse may differ from those in academic writing（see Biber 2006; Biber and Conrad 2009;Charles et al. 2009; O'Keeffe and McCarthy 2010; John and Laso 2013; Sindoni 2013）. Similarly, the phraseological features of lexical items in business-related discourse are somewhat different from those in newspaper articles. In other words, this relation between phraseology and genre makes it possible to analyse features associated with a certain discourse or compare two different genres based on the use of phraseology in texts（see Aijmer and Stenstr?m 2004; Biber et al. 2004; Sinclair 2004a; Hoey 2005; Biber 2006; Biber and Conrad 2009; R?mer and Schulze 2010;John and Laso 2013）. Furthermore, a comparison of the use of phraseology can also be carried out across different languages, e.g. by using parallel corpora to explore the preference of phraseology in two languages（cf. Granger and Meunier 2008; Ji 2010;McEnery and Hardie 2012）. In short, it is undeniable that phraseology is important in language use and that further research on phraseology would be beneficial, which is the main reason why the current investigation focuses on phraseology or the phraseological behaviour of lexical items.

1.2 Scope of phraseology

Increasingly, it has been widely recognised that language has a strong phraseological nature（see Sinclair 1991,2004a; Hoey 2005; Granger and Meunier 2008;R?mer and Schulze 2009; Stubbs 2009a; Moon 2010; O'Keeffe and McCarthy 2010；Hunston 2011; Granger et al. 2013; John and Laso 2013）. Sinclair（1991）in particular proposes the idiom principle which accounts for how language normally works and suggests that phraseology is central to language use. This principle states that“a language user has available to him or her a large number of semi-preconstructed phrases that constitute single choices, even though they might appear to be analysable into segments”（Sinclair 1991:110）. It suggests to some extent how the lexical or lexicogrammatical co-occurrence may be the default mode of language organisation and provides the theoretical basis for the current study.

However, there is still no clear consensus to date about either the definition or scope of phraseology（cf. Anderson 2006; Granger and Meunier 2008）. In other words, the definition of this term varies from researcher to researcher. For instance, some traditional studies of phraseology have been restricted to idioms and fixed expressions（cf. Howarth 1996; Cowie 1998; Granger and Meunier 2008）, while later studies（e. g. Sinclair 1991,2004a; Hoey 2005; Meunier and Granger 2008; O'Keeffe and McCarthy 2010; Hunston 2011; McEnery and Hardie 2012; Hanks 2013）adopt a broader view and define phraseology as a cover term for all sorts of multi-word phrases.

In this book, I argue that the term“phraseology”refers to two aspects of multiword sequences:1）the analysis of the form, i.e. recurring multi-word phrases as well as sequences with paradigmatic choices（such as patterns and collocational frameworks）;2）the analysis of the use of these phraseological items（phraseological behaviour）, e.g. the semantic and pragmatic uses of a phraseological item and the relation between this phraseological item and its co-texts. In the following sections, I will provide the rationale for my definition of phraseology, discuss the criteria for defining and describing a phraseological item, and introduce relevant terminology which is used in this book.

1.2.1 Phraseology and a phraseological item

A phraseological item involves two main types of lexical（or lexicogrammatical）co-occurrence:continuous or discontinuous multi-word phrases which are constructed by specific words, and a type of ‘sequence'which involves paradigmatic choices of words or phrases（e.g. pattern and frame）.

The first type of phraseological item, “recurring（dis）continuous multi-word phrases”, is discussed under different labels, e. g. “lexicalised stems”, “lexical phrases”, “formulaic sequences”, “lexical bundles”, “n-grams”, “lexical items”,“multi-word expressions”（MWEs）, and“multi-word units”（MWUs）（see Wray 2000:465; Biber et al.2004:372; Meunier and Granger 2008; O'Keeffe and McCarthy 2010; McEnery and Hardie 2012）. However, these labels for multi-word phrases are not used entirely synonymously by researchers, except for the more ‘general' terms such as“multi-word sequences”, “multi-word units”（MWUs）and“multiword expressions”（MWEs）.

For instance, Pawley and Syder（1983:191）use the term“lexicalised stem”to refer to“a unit of clause length or longer whose grammatical form and lexical content is wholly or largely fixed”, which covers longer sequences and highlights the criterion of“institutionalisation”and“lexicalisation”for a phraseological item. Nattinger and DeCarrico（1992）adopt the term“lexical phrases”to mean“prefabricated language chunks”or“conventionalised form/function composites that occur more frequently and have more idiomatically determined meaning than language that is put together each time”（Nattinger and DeCarrico 1992:1）, which emphasises the criteria of frequency and semantic non-compositionality for a phraseological item. Nattinger and DeCarrico's（1992）study is also normally associated with the application of multi-word phrases in the field of language teaching.

Wray（2000,2002）, writing from the perspective of psycholinguistics or clinicallinguistics, prefers the term“formulaic sequences”for the description of a wide range of multi-word strings which are“stored and retrieved whole from memory at the time of use, rather than being subject to generation or analysis by the language grammar”（Wray 2000:465）. Additionally, one positive aspect that has emerged from Wray's studies is the inclusion of discontinous sequences into the scope of formulaic language, which supports my definition of phraseology, i.e. a phraseological item can be a continuous multi-word sequence or a discontinous sequence. Biber et al.（2004）and Biber（2006）, on the other hand, use the label“lexical bundles”to investigate recurrent sequences of words with a corpus-based approach, paying more attention to the use of phraseology in specific genres or across genres.

The term“lexical items”among the group of labels for recurring multi-word sequences is also worthy of more explanation. In this book, this term is used to refer to both single words and multi-word phrases as long as the word or phrase is used as an individual unit of meaning in context. This definition of“lexical items”follows Sinclair et al.（2004:9）who suggest that“a lexical item”may not always be associated with“an orthographic word”; among many forms a lexical item can refer to（e.g.“morpheme”）, the term“lexical item”also covers“a pair or group of words associated syntagmatically”such as a multi-word sequence. In other words, a lexical item can refer to either a word（e.g. time or thing）or a phrase（e.g. at the same time and from time to time）.

The second type of phraseological item, as discussed earlier, involves a type of‘sequence'which is not entirely constructed by specific words and is associated with paradigmatic choices of lexical items, e. g. “pattern”（Hunston and Francis 2000）and“frame”（Renouf and Sinclair 1991）. The main reason that I include this type of sequence within the scope of phraseology is that patterns and frames are also associated with lexical or lexicogrammatical co-occurrence. Generally speaking, patterns（e. g.‘it v-link time to-inf.'and ‘the ADJ thing to do'）describe the co-selection of lexical items and grammatical categories; and frames（e.g. ‘a（n）+? +of'）concern the co-occurrence of a fixed part of lexical items and a variable part of lexical or grammatical items. The above argument that patterns and frames can be included within the scope of phraseology is also supported by other researchers. For instance, Hunston（2011）has illustrated the relation between phraseology and evaluation using language patterns（e.g. ‘it v-link ADJ to-inf.'and‘a time of N'）as well as multi-word phrases. This suggests that it is possible to consider phraseology as a broad term to include patterns. Similarly, in Granger and Meunier（2008）, frames（e.g. ‘a（n）+? + of'and‘be +? +to'）and patterns（e.g. ‘ADJ N'and‘as ADJ as'）are examined under the scope of phraseology（cf. Granger and Paquot 2008:39; Martin 2008:51-66; Arnaud et al. 2008:111-126; Wikberg 2008:127-142）. Furthermore, the categorisation of patterns and frames as types of phraseological item extends the scope of phraseology, which could be viewed as a better recognition of the significant role of phraseology in language use.

In addition to being an umbrella term for all types of multi-word sequences or sequences with paradigmatic choices, phraseology can also refer to various aspects of a phraseological item:e.g. the syntagmatic features of this item, the semantic and pragmatic uses of this item, and the textual and social functions of this item. Cowie（1994:3168）, for instance, defines phraseology as“the study of the structure，meaning and use of word combinations”. His definition of phraseology includes at least three aspects of the study of word combinations:the form, semantic features and pragmatic use. Hunston also suggests that phraseology refers to more than just a collection of phrases or sequences:“it encompasses all aspects of preferred sequencing”（2002a:138）. Similarly, Thompson and Hunston argue that：

The term ‘phraseology'can be used to describe not just the fact that many words frequently occur in phrases, but the more abstract tendency for words to co-occur non-randomly and for the selection of particular lexical items to alter the probability of other lexical and grammatical choices.（Thompson and Hunston 2006:10）

In otherwords, Thompson and Hunston regard the features from the cooccurrence of words, e.g. the collocational or colligational behaviour of a lexical item, as a part of the study of phraseology.

To summarise the above discussion, it is argued in this book that the term“a phraseological item”refers to recurring multi-word phrases and sequences with paradigmatic choices（e.g. patterns and frames）. The term“phraseology”, on the other hand, involves the study of both the form and the use of a phraseological item. More specifically, the investigation of phraseology in this book concerns the syntagmatic features of a phraseological item, the semantic and pragmatic features of this item, and the discourse functions of this item.

1.2.2 Criteria for identifying a phraseological item

ccording to previous studies on phraseology, there are three generally-accepted criteria fordefining and describing a phraseological item:frequency, syntagmatic fixedness and semantic non-compositionality（see Sinclair 1991; Howarth 1996; Cowie 1998; Moon 1998; Hoey 2005; Anderson 2006; Granger and Meunier 2008; Gries 2008; Meunier and Granger 2008; Bolly 2009; R?mer and Schulze 2009; Herbst et al. 2011; Maienborn et al. 2011）. Before discussing these three criteria in more detail, it is important to note that each criterion should be perceived more as a dimension because each is not a binary feature but involves a continuum of features with different degrees. For instance, the criterion semantic non-compositionality describes not only the lexical items which are compositional or non-compositional, but also the items which are partially compositional or partially non-compositional（a further discussion of this criterion is provided below）. These three criteria are represented in Figure 1.1.

The first dimension, frequency, is considered to be the statistical criterion for identifying a phraseological item（see Sinclair 1991; Hunston 2002a; McEnery et al. 2006; Gries 2008）. Phraseological items can range on a continuum based on their occurrences in a corpus. Some of these can occur very frequently in a corpus, e.g. for the first time（110 times per million in the BoE）and at the same time（57 times per million in the BoE）; while others may be less frequent, e.g. time after time（0.85 times per million in the BoE）and time flies（0.27 times per million in the BoE）. It is thus up to the researcher to decide how many times a sequence needs to occur in the corpus to be counted as a phraseological item, but generally, if a sequence occurs more frequently in the corpus, it is more likely that this sequence will be regarded as a phraseological item. In these cases, the normalised frequency（e. g. the number of occurrences per million words）is often used instead of raw frequency so that the occurrences of a phraseological item can be compared across corpora. Other statistical measures, such as the t-score, can be used as supplementary criteria if the multiword sequence occurs with a relatively low frequency. It has been suggested from the statistical point of view that the combinations with a t-score above 2.4（or sometimes even 3）can be regarded with confidence as strong co-occurrences（see Barnbrook 1996:97; Hunston 2002a:72; Hoover et al. 2014:154）.

The dimension of fixedness relates to the degree of syntagmatic variability or flexibility（see Burger 2007:910; Granger and Meunier 2008）. Several factors may contribute to different degrees of syntagmatic fixedness, e. g. the insertion of words into the combinations, the omission of components in the combinations, the substitution of components in the combinations and other alterations of components in the combinations（see Howarth 1996; Granger and Meunier 2008; R?mer and Schulze 2009; Herbst et al. 2011）. For instance, the phrase for the first time which occurs very frequently in the BoE can allow the insertion of additional words without interrupting the ‘core'structure of the phrase, such as for the very first time. In the phrase time and time again, the third word time can be omitted（as in time and again）and the meaning（or use）will remain more or less the same. The above two phrases, for the first time and time and time again, would be cases of lexical items with a relatively lower degree of syntagmatic flexibility. One example with a higher degree of syntagmatic flexibility is the verb phrase spend time, as its variants can be spent time, time is spent, spend some time and spend a lot of time. In fact, it is argued that the majority of phraseological items are associated with some sort of syntagmatic flexibility, i.e. a lexical item may involve one canonical form and one or more variants（see Sinclair 1991）. Examples which are fully syntagmatically fixed are rare. The phrases from time to time and once upon a time may be cases of this sort.

The third criterion which is often used for identifying a phraseological item is semantic non-compositionality which refers to the degree to which the meaning of a multi-word sequence cannot be derived based on the sum of its constituent words（see Moon 1998; Anderson 2006; Gries 2008; Meunier and Granger 2008; R?mer and Schulze 2009; Aijmer and Altenberg 2013）. Semantic non-compositionality is also sometimes referred to as semantic opaqueness or“non-predictability”（Gries 2008:4）. One type of phraseological item at the non-compositional end could be idioms because the meaning of idioms（especially‘pure'or‘typical'idioms）is usually hard to interpret on a word-for-word basis（cf. Howarth 1996; Cowie 1998; Moon 1998;Anderson 2006; Skandera 2007; Granger and Meunier 2008）. Examples which are semantically non-compositional in this study could be from time to time, big time and in time whose meaning is associated with the entire unit rather than separate components of the unit. There are also examples which are partially compositional（or transparent）. For instance, it could be argued that in the cases of full time and part time， the meaning of at least one part of the lexical item can be interpreted. The examples at the compositional end may be phrases like a waste of time, for the first time and‘the time has come for n.'.

To illustrate the interactive relationship between the three criteria for the description of a phraseological item, two phrases for the first time and from time to time are used as examples. As can be seen from Figure 1.2, the phrase for the first time occurs frequently in the BoE; it is relatively fixed and semantically compositional. The phrase from time to time, as represented in Figure 1.3, occurs less frequently than for the first time, but this phrase is syntagmatically more fixed than for the first time and it is relatively non-compositional.

In other words, each phraseological item is associated with the three features to a different degree. In thecase of from time to time, it involves a high degree of syntagmatic invariability and semantic non-compositionality; it however occurs relatively less frequently（see Figure 1.3）. On the other hand, a sequence or a combination which is non-frequent, syntagmatically free and semantically compositional is not likely to be regarded as a phraseological item in this book.

Even though the three criteria or dimensions are represented as being parallel to each other in these figures（e.g. Figure 1.3）, it is argued in this book that there is an order in which these three criteria play a part in the determination of whether a sequence should be considered as a phraseological item. The criterion of frequency is prioritised before the other two criteria in this study since the frequent occurrences of an item in the corpus may highlight its importance in language use（see Sinclair 1991;McEnery et al.2006; Gries 2008; Herbst et al.2011）. A sequence which occurs only twice in the BoE, for instance, may not be important enough to teach to students in language classrooms（see Sinclair 2004b; Meunier and Granger 2008; Aijmer 2009;Reppen 2010）. Additionally, the criterion of frequency is less likely to involve subjective interpretations than the other two criteria（e.g. semantic non-compositionality）. The corpus-based method is also more suitable for showing the frequency data of any multi-word sequence in text. After frequency, syntagmatic fixedness is considered and then semantic non-compositionality. The reason that this book places less emphasis on semantic non-compositionality than the other two criteria is that it recognises the significance of frequently occurring phraseological items which are compositional. Although traditional research on phraseology has mainly focused on the study of more fixed and opaque multi-word units, increased attention has been paid recently to a much wider range of lexical units which are associated with a higher degree of semantic compositionality（and syntagmatic variability）because it is believed that these semantically more compositional combinations（e.g. the majority of n-grams analysed by Biber 2006）are equally important to the studies of phraseology（cf. Wray 2002; Schmitt 2004; Biber and Barbieri 2007; Granger and Meunier 2008）.

1.3 Relevant terminology

In the investigation of phraseology in previous studies（e. g. Howarth 1996;Cowie 1998; Lewis 2000; Hunston and Francis 2000; Hunston 2002a; Barnbrook et al. 2013）, other terms are often mentioned. The following sections introduce four commonly discussed terms, namely, collocation, colligation, pattern, and frame.

1.3.1 Collocation and colligation

Among the various terms used in the study of phraseology, “collocation”is probably one of the most controversial and slippery terms because different scholars have used it to refer to slightly different groups of phraseological items（cf. Sinclair 1991;Howarth 1996; Cowie 1998; Lewis 2000; Sinclair et al. 2004; Hoey 2005；Nesselhauf 2005; Barnbrook et al. 2013）. According to Handl（2008:50）, the use of the term“collocation”can“stretch on the continuum between free word combinations and fully fixed idioms or compounds”, which indicates the ‘looseness'of“a collocation”.

Despite the lack of consensus about this term, most researchers have agreed on a few of the features which are exhibited by a collocation, e.g. its frequent occurrences and a certain degree of syntagmatic variability and semantic compositionality（cf. Howarth 1996; Handl 2008; Barnbrook et al. 2013）. In other words, in this book a collocation is regarded as a phraseological item and it is used broadly to refer to frequently occurring combinations which may be compositional or non-compositional.

It should be noted that the term“collocation”, however, is sometimes referred to differently from“a collocation”. A collocation usually refers to a sequence of two or more words, e.g. spend time and part time. On the other hand, collocation can be used to describe the phenomenon of co-occurrence of two（or more）words（see Sinclair 1991; Sinclair et al. 2004; Hoey 2005; Lindquist 2009）. For instance, when the word time exhibits the sense of ‘occasion', it can co-occur frequently with words such as first, second, next and last. This co-occurrence of the word time with words like first and second is relevant to the phenomenon of collocation. In this case, time is regarded as the node word and the words which co-occur with time（e.g. first and second）are referred to as collocates of time. It is also possible to consider the above features exhibited by the word time as collocational features or“collocational behaviour”（see Sinclair et al. 2004; Hoey 2005; Walker 2011; Barnbrook et al. 2013）.

Additionally, collocation in this study is not limited to the co-occurrence of words. Instead, it is argued that the phenomenon of collocation can be extended to the co-occurring features of lexical items, i.e. collocation can refer to the co-occurrence of two lexical items（cf. Sinclair 1991; Sinclair et al. 2004; Hoey 2005; Walker 2011; Barnbrook et al. 2013）. The analysis of that sort of thing, for instance, shows that this lexical item co-occurs frequently with the word and, as in and that sort of thing. To some extent, it is argued that this co-occurrence of that sort of thing with the word and may also be regarded as collocation（or collocational features of that sort of thing）. Another example is the sequence the big time. The analysis of this sequence shows that it tends to co-occur with hit and make, as in hit the big time and make the big time. This tendency of co-occurrence of the big time with hit（or make）, similar ly, is considered in this book as a type of collocational feature of the big time.

Related to collocation in this study is the term“colligation”, or, in the case of collocational behaviour, “colligational behaviour”（see Sinclair 1991,1996,2004a;Stubbs 2009a）. Collocation or collocational behaviour describes the co-occurrence of features at the lexical level; while colligation or colligational behaviour refers to the co-occurrence of features at the syntactic（or lexical-syntactical）level（see Sinclair 1991; R?mer 2005; Lindquist 2009; Stubbs 2009a）. In other words, collocation is related to the frequent co-occurrence of one lexical item and another lexical item whereas colligation involves the frequent co-occurrence of a lexical item and grammatical categories（cf. Hoey [2005:43]for a broader definition of colligation）. In the case of the sequence and that sort of thing, for example, the current study shows that it tends to co-occur with noun phrases and verb phrases. This feature may be referred to as the colligational behaviour of and that sort of thing.

1.3.2 Pattern

As mentionedearlier, “a pattern”is regarded as a phraseological item which describes the co-selection of lexis and grammar（see Hunston and Francis 2000; Hunston 2002a,2002b; Hunston 2011; McEnery and Hardie 2012）. More specifically,“a pattern”in the current study refers to a sequence which contains specific lexical item（s）and at least one‘slot'which involves paradigmatic choices of lexical items. What is more, the lexical items which fit the ‘slot'in a pattern are often grammatically and/or semantically related（see Hunston and Francis 2000）.

For instance, the sequence‘spend time v-ing'is considered as a pattern in this study because, firstly, it is associated with a type of lexicogrammatical cooccurrence, i.e. the co-occurrence of specific lexical item（s）and a group of“word types or clause types”（Hunston 2002b:169）; and, secondly, this sequence contains one ‘slot'which involves paradigmatic choices of related clause type（‘ving'）. In this case, the words or phrases which fit the slot ‘v-ing'in the pattern are at least grammatically similar（e. g. working, reading and watching movies）. Equally, the sequence ‘it v-link time to-inf.'is referred to as a pattern because it involves paradigmatic choices of word types or clause types. In this pattern, two‘slots' are involved:‘v-link'which refers to link verbs such as is, was and may be; and‘to-inf.'which refers to to-infinitive clauses such as to change, to act and to take a closer look. Another example, ‘＜place＞ + time', is also regarded as a pattern in this study because the lexical items which occur at the slot‘＜place＞'in this sequence are mainly words which express similar meanings:describing a place, e.g. Brisbane, New York, London and British. In other words, the lexical items which fit the‘slot'in the pattern ‘＜place＞ + time'are at least semantically related.

This definition of“pattern”in this book is more or less consistent with Hunston and Francis（2000）or Hunston（2002a,2002b）. In these studies, “a pattern”is similarly viewed as a language phenomenon which relates to both lexis and grammar（or breaks the sharp division between lexis and grammar）and involves paradigmatic choices of lexical items. For example, Hunston（2002b:169）defines a pattern as“a sequence of grammar words, word types or clause types which co-occur with a given lexical item”, which is somewhat similar to the definition of pattern in this book. One difference, however, is that the current study uses the term“pattern”slightly more broadly than Hunston and Francis and includes more sequences which describe lexicogrammatical co-occurrence. For instance, the sequences such as ‘＜place＞ + time',‘the ADJ thing to do'and ‘at the time of N'were not analysed in Hunston and Francis's（2000）study of patterns, but these sequences are regarded as patterns in the current study because they fit the definition as discussed earlier. By adopting a broader definition of pattern, the current study also contends that patterns are more common and important in language than previously believed.

In this study a pattern can also be described by the three criteria associated with a phraseological item. For instance, Gries（2008:7）suggests that patterns involve several features:1）they are“lexically partially filled”;2）they“require the insertion of additional lexical material”;3）they“allow for syntactic variation”. In other words, the two criteria, semantic non-compositionality and syntagmatic fixedness, are regarded as features of a pattern. The other criterion, frequency, is not mentioned by Gries（2008）; but this criterion is equally important for the definition of a pattern. It is argued in this book that a pattern can only be identified if the sequences which fit the pattern occur relatively frequently（see Hunston and Francis 2000; Hunston 2002b; Mahlberg 2006）.

1.3.3 Frame

The term“frames”in this study is also viewed as a type of phraseological item which is“an alternative”to the combinations constructed by specific individual words in a language（cf. Hunston 2002a:50）. To be more specific, a frame is defined as a discontinuous sequence which contains a paradigmatic choice of lexical items（cf.“collocational frameworks”in Renouf and Sinclair 1991）. For instance, the sequence‘the time is +? + minutes past/before the hour'is regarded as a frame because it involves a variable lexical‘slot'where words such as seven, nineteen and twenty can be inserted.

This definition of“frame”in the current study mainly follows Renouf and Sinclair（1991）who have suggested that a frame or a“collocational framework”is composed of a fixed part of lexical items and a variable‘slot'which can be filled by a group of words（cf. Butler 1998; Marco 2000; Vincent 2013）. However, the definition in this book concerns frames which include at least three words（e.g. ‘the time is +? + minutes past/before the hour'）. Traditionally, researchers have investigated the frames composed of generally three or four words, e.g. ‘a（n）+? +of'which includes the fixed part of two grammatical words（the article a（n）and the preposition of）and a variable part where words such as couple, lot, number and indication can be inserted（see Renouf and Sinclair 1991; Francis 1993; Stubbs 2007b; Granger and Meunier 2008）. The current study argues that the term“frame”could cover a broader scope of sequences, e.g. longer sequences such as ‘the time is +? + minutes past/before the hour'.

Another point about the term“frame”which is worth mentioning is that even though patterns and frames are similar in the way that they both involve paradigmatic choices of lexical items, a frame is different from a pattern with regard to the type of lexical items which fill the variable‘slot'. The lexical items which are associated with a‘slot'in a pattern as defined in the current study can involve phrases and clauses（e.g. ‘it v-link time to-inf.'）while it is mainly individual words which fill the‘slot'in a frame（e.g. ‘a（n）+? + of'）. Additionally, the lexical items which fit a pattern are considered to be either grammatically and/or semantically related; on the other hand, the group of lexical items which fill the‘slot'in a frame may be less closely related, e.g. couple, lot, number and indication which fit the frame‘a（n）+?+ of'. Similarly, regarding the above-mentioned example‘the time is+? +minutes past/before the hour', although the items which fit this frame appear to be largely numbers（e. g. seven, nineteen and twenty）, they may not be categorised into the same grammatical or semantic group as those for a pattern like ‘it v-link time to-inf.'.

To sum up the discussion on the terminology of phraseology, Table 1.1 again presents the definitions and examples for the above-mentioned terms. It is also important to emphasise that the above description in Table 1.1 is based on my definitions of“a phraseological item”and“phraseology”. To reiterate, “a phraseological item”in this book is used as an umbrella term which refers to different types of recurring multi-word phrases and discontinuous sequences which contain paradigmatic choices of lexical items（including“patterns”and“frames”）. The identification and description of a phraseological item are usually associated with three criteria:frequency, syntagmatic fixedness and semantic non-compositionality. The term“phraseology”, on the other hand, encompasses the study of both the form of phraseological items（e.g. various types of multi-word sequences）and their use（i. e. the syntagmatic, semantic, pragmatic and textual features of a phraseological item）.