“OH I JUST TALK NORMAL LIKE”: A CORPUS

Transcription

“OH I JUST TALK NORMAL LIKE”: A CORPUS
 “OH I JUST TALK NORMAL LIKE”: A CORPUS-BASED, LONGITUDINAL STUDY
OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
JOANNE BARTLETT*
(Newcastle University)
Abstract
This paper investigates the Constituent-Final LIKE form in Tyneside English. Its results are
generated from a real-time, corpus-based comparison of Tyneside speakers recorded in the
1960s and 1990s. From this data, it would appear that the use of Constituent-Final LIKE in the
North-East of England has declined over time. This paper studies the sociolinguistic (sex and
age) patterning of the tokens and finds that Constituent-Final LIKE is more so a feature of the
speech of males and younger informants. It also studies the pragmatic nature of the tokens
and finds that Constituent-Final LIKE fulfils a number of roles: helping to structure opinions,
explanations and interrogatives in order to “save face” between interlocutors. Following this,
the paper unites these two fields and focuses on how different social groups use the
Constituent-Final LIKE tokens for different functions.
1.
Introduction*
The volume of scholarly work concerning LIKE as a discourse-pragmatic marker is
vast, but very few studies have focused upon its use in constituent-final position, as
demonstrated in (1). Instead, Constituent-Final LIKE (henceforth CFL) has been overshadowed
by its focuser and quotative cousins ⎯ exemplified in (2) and (3) respectively ⎯ despite the
form itself having a ‘clear interactional function’ in discourse (Kerswill & Williams 2002:
103).
That’s the only time I was away LIKE
I thought she would be LIKE
Did you go into Keswick LIKE?
[TLS/13]1
[TLS/36]
[PVC/09B]
(2)
If you live LIKE in the city and all that
[PVC/10B]
(3)
So I was LIKE, “Oh right!”
[PVC/12A]
(1)
a.
b.
c.
can occur within an utterance or attach at its end (Kallen 2006: 12), and is argued
to have both ‘wide-scope’ (D’Arcy 2005: 69) and ‘backward scope’ (D’Arcy 2005: 71) over
CFL
*
I would like to take this opportunity to thank my supervisors Karen Corrigan and Cathleen Waters for their
support and advice throughout this project. Their help has been, and continues to be, much appreciated. This
paper has been presented at the 7th Newcastle upon Tyne Postgraduate Conference in Linguistics, DiPVaC1 and
Sociolinguistics Summer School 4. I would like to thank those who have made suggestions regarding its further
development. 1
Note that [TLS/13] is a speaker code: ‘TLS’ refers to the particular corpus, while ‘13’ refers to the individual
informant identifier. Newcastle Working Papers in Linguistics 19.1, 2013
Selected Papers from Sociolinguistics Summer School 4
Roberts, N. S. and Childs C. (eds.)
JOANNE BARTLETT
2
the constituent which precedes it. Its status as an adjunct, or optional inclusion (D’Arcy 2005:
69), means that the existence of CFL in discourse is argued to be meaningless (Oxford English
Dictionary 2007: 1601). I dispute this view and claim that, whilst constructions remain
undeniably grammatical without it, CFL instead performs a number of pragmatic functions in
conversation that its absence does not.
Due to its prominent attestation in Ireland (such as Kallen 2006; Columbus 2009;
Luckmann 2009; Siemund et al. 2009; Schweinberger in press) and observation in Scotland
(Miller & Weinert 1995), CFL is thought to predominantly be a feature of Northern Englishes
(Andersen 2001: 222). Despite this, CFL remains unstudied within the dialects of Northern
England, meaning that little is known as to its distribution, or indeed its existence, in these
varieties. This dearth is addressed here during the quantitative investigation of CFL
occurrence in Tyneside English, an urban variety which occupies the ‘extreme North-Eastern
corner of England’ (Beal 1993: 187) and exhibits a number of idiosyncratic phonetic, lexical
and morpho-syntactic features (cf. Beal 1993; Beal & Corrigan 2009: 231).
Of the studies which make reference to CFL, most are pragmatic in nature and so
describe the functions that this feature performs in discourse. In contrast, very few have
investigated CFL within a variationist sociolinguistic perspective. This study amalgamates
these two approaches by investigating the sociolinguistic distribution of CFL function within
the Tyneside community. Taking a corpus-based approach, the variable distribution is
analysed across two distinct time periods ⎯ the 1960s and the 1990s ⎯ in order to examine
whether diachronic changes in the pragmatic use of CFL have occurred. Additionally, I
comment qualitatively upon the feature’s pragmatic functions and analyse more generally the
distribution of the CFL tokens in Tyneside, in relation to a number of extralinguistic variables
(sex and age). These, when compared to the other, albeit few, variationist studies already
conducted, will provide a local perspective on a global linguistic trend (cf. Buchstaller &
D’Arcy 2009).
2.
Literature review
The research concerning CFL is not extensive. This section will introduce CFL as a
discourse-pragmatic marker, before contextualising the form amidst the variationist and
pragmatic literature which has been published thus far.
2.1.
CFL as a discourse-pragmatic marker
This paper refers to CFL as a discourse-pragmatic marker. The study of these features
is complex, not least because they mean slightly different things to different researchers (cf.
Fraser 1999: 932) which, in turn, instigates the use of ‘competing terms […] with partially
overlapping reference’ (Schourup 1999: 228). Thus, simply defining the feature under
investigation becomes problematic when the author is presented with conflicting opinions on
the matter. Indeed, CFL has already been referred to by several different terms, despite its
infrequent discussion in the literature. Kallen (2006), Levey (2006), Miller (2009) and
Schweinberger (in press), for example, refer to CFL as a discourse marker, whilst others
describe it as a discourse particle (Corrigan 2010), invariant tag (Columbus 2009) and
sentence adverbial (D’Arcy 2005). These terms are, for the most part, adequate. However,
this study adds another, namely discourse-pragmatic marker, on the basis that CFL
‘structure[s] discourse’ in the same manner as the other LIKE discourse-pragmatic markers
(Corrigan 2010: 101) and has pragmatic functions. I dispute the view that discourse markers
can only occur sentence-initially (such as D’Arcy 2005: 69), taking heed instead from the
notions that these features ‘bracket units of talk’ (Schiffrin 1987: 31) and can occur in ‘final
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
3
position’ (Fraser 1999: 938). Indeed, CFL conforms to the specification of a discoursepragmatic marker given its status as an adjunct, its lack of semantic meaning and it boasting a
degree of multifunctionality (Pichler & Levey 2010: 17). Therefore, it seems only logical (or
acceptable at least) for CFL to be studied here as a discourse-pragmatic marker.
2.2.
History and geography
In contrast to the popular focuser and quotative LIKEs of today, whose observation
extends back only to the 1980s (such as Butters 1982; Schourup 1985; Underhill 1988), CFL
has been recorded in written discourse since the late 18th Century (OED 2007: 1601). Despite
this, the development of CFL is not well-documented. Romaine & Lange (1991), for example,
acknowledge its existence but fail to incorporate this form when detailing the
grammaticalisation of LIKE over time. CFL is included in the trajectory modified by D’Arcy
(2005), who explains that the form represents a medial point between the original use of LIKE
as a preposition and its newer, semantically-bleached and pragmatically-strengthened guise
(D’Arcy 2005: 72). D’Arcy (2005: 68f) contends that as CFL developed from a conjunction,
its function ‘shifted to the pragmatic domain’; an analysis which is especially important to
my labelling this form a discourse-pragmatic marker.
CFL was once a ‘fairly widespread’ feature in the UK (Kallen 2006: 14), but is now a
phenomenon more commonly associated with the North (Andersen 2001: 222). This notion is
supported by its ‘robust’ (Kallen 2006: 14) presence in Ireland; a geographical location in
which much of the CFL literature has focused (such as Harris 1993: 176; Kallen 2005, 2006;
Columbus 2009; Luckmann 2009; Siemund et al. 2009; Corrigan 2010; Schweinberger in
press). In addition, CFL has been attested in Scottish English (Miller 1993: 136f; Miller &
Weinert 1995). However, excluding an acknowledgement of existing in the North-East (Beal
2004: 135f), it seems that CFL has not been recorded, nor quantitatively studied, in the
dialects of Northern England. Therefore, its association with this area may be seldom true.
Indeed, CFL has been noted in Southern England too (Andersen 2001; Levey 2003, 2006;
Anderwald 2004: 193), though these frequencies were extremely low and fail to dispute the
claim that CFL is, in fact, a feature of the North. In addition to these UK observations, CFL has
been recorded in Australasia (Miller 2009), the USA (Schourup 1985; Kallen 2006) and in
the Englishes of India, East Africa and the Philippines (Siemund et al. 2009).
By virtue of its ‘robust’ (Kallen 2006: 14) nature in Ireland, it seems logical to
suggest that CFL may have been brought to Tyneside by the large number of Irish migrants
who settled in the community and impacted heavily upon the local linguistic system during
the 19th Century (Beal & Corrigan 2009: 231f). However, as the data in this study came from
speakers born after 1895, there is no way to determine whether this hypothesis is true because
the Irish (who potentially brought CFL with them) entered Tyneside prior to this date.
2.3.
Variationist sociolinguistics
CFL has seldom been investigated from a variationist sociolinguistic perspective. This
fact is, for several reasons, somewhat unsurprising. First, taking an argument put forth by
Kerswill & Williams (2002: 101) concerning focuser LIKE, CFL ‘is not a variable sensu
stricto’ in that it cannot be replaced consistently by another discourse marker. Moreover,
Dailey-O’Cain (2000: 64f) notes that the quantification of discourse variables involved in
such investigations is difficult due to the complexity, or even impossibility, in determining
exactly where adjuncts can and cannot occur. In addition to this, it may be that some studies
intended to implement quantitative analyses of CFL but were hindered by the low numbers
found in their data. Levey (2003, 2006), for example, found 1 CFL token in the speech of his
JOANNE BARTLETT
4
London pre-adolescents and so excluded this from the statistical analyses of his other LIKE
forms (Levey 2006: 431), whilst Miller & Weinert (1995) found only 16 examples in their
Scottish data. Concerned with the latter, Andersen (2001: 219) argues that this result
demonstrates that CFL is used ‘commonly’ in Scotland in comparison to its ‘virtual[…] nonexisten[ce]’ (ibid: 222) in Southern English speakers; a conclusion which should, in my
opinion, be treated with some trepidation since the token numbers are so low.
In fact, the only location in which studies note substantial CFL frequencies is Ireland,
which forces the reiteration that the feature is ‘robust’ (Kallen 2006: 14) in this part of the
world. Although these studies have each used the Irish component of the International
Corpus of English, they have returned different CFL totals, ranging from approximately 200
tokens (Columbus 2009) to almost 600 (Siemund et al. 2009). Of these, only Luckmann
(2009) and Schweinberger (in press) analyse their CFL tokens in correlation with social
factors such as sex and age, which makes the task of pinpointing or suggesting universallyheld CFL constraints difficult. Both studies report that females use CFL more frequently than
males, whilst the vernacular of the younger speakers is found to contain more tokens than that
of the older informants. However, it should be noted that CFL still remains a feature of older
speakers in Ireland (Corrigan 2010: 100). This is due, most probably, to it being a more
traditional form of discourse marking LIKE.
Evidently, there are few studies to draw upon regarding the investigation of CFL from
a comparative sociolinguistic perspective. The current paper addresses a much needed dearth
in the LIKE paradigm.
2.4.
Pragmatic function
LIKE,
as a discourse-pragmatic marker more generally, is said to loosen or enrich the
utterance within its scope (Andersen 2000: 17f). This attribute can be extended to the study of
CFL as this form too provides an additional, pragmatic interpretation to the semantic
proposition encoded in an utterance.
Corrigan (2010: 100) states that CFL ‘structure[s] discourse’ and marks the end of
information or the closing of a narrative. CFL likely signals to the hearer that their turn to take
the conversational floor is imminent and may therefore act as a turn-transition device
(Schiffrin 1987: 218) in a bid to organise the ongoing movement of discourse. In addition to
this, several more pragmatically-specific functions have also been attributed to CFL.
In keeping with the enrichment and loosening aforementioned, CFL is argued to hedge
or mitigate the force of its preceding assertion (Kallen 2005, 2006; Corrigan 2010). This
method allows opinions to be declared without the hostility which may otherwise alienate the
interlocutors. Others argue that CFL ‘anticipate[s] an objection’ (Miller 2009: 336) and may
counter potentially incorrect inferences before the hearer has a chance to protest what has
been said (Miller & Weinert 1995: 389). Thus, it appears that CFL plays a significant role in
both mitigating or resolving potential misunderstandings and differences in opinions. In
addition to this, CFL is said to reinforce the proposition uttered (Beal 2004: 135) and focus
attention on particular pieces of information (Kallen 2006: 12). These proposed functions
relate to Miller’s (2009: 336) contention that CFL encourages the hearer to accept the
explanation being provided, although the use of the form signals that objections are, as
before, anticipated by the speaker. Finally, it is well-acknowledged that CFL occurs in
interrogatives in order to elicit clarification (Miller & Weinert 1995: 390) and ask for further
explanations (Miller 2009: 329). It could be argued that this function too provides mitigation,
in that the CFL form may make the question appear less probing, intrusive or brazen.
Thus far, I have not addressed Columbus’ (2009) classification of CFL function. This
study transferred the pragmatic functions previously postulated for invariant tag questions to
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
5
the classification of the CFL tokens. Columbus (2009) accounts for CFL across 14 categories
which can be related to the functions already mentioned, such as it being used in narratives,
as a softener and to emphasise information. However, several of these categories accounted
for merely one or two of the CFL tokens. This is not a criticism of Columbus’ (2009)
categorisation per se, since the aims of that paper were fulfilled, but this is a method which
cannot be implemented here because such small cell numbers make quantitative analysis
problematic.
It appears, therefore, that this study requires the implementation of several, broad
functional categories which account for the pragmatic roles that CFL performs in discourse,
but also allow quantitative analyses to be conducted easily. To do this, I argue that the
aforementioned functions overlap in meaning, allowing larger pragmatic categories to be
created when these are collapsed under three distinct headings. This method will be presented
and outlined in more detail later.
3.
Methodology
The study of discourse-pragmatic markers within the variationist paradigm is difficult,
due largely to them having been ‘neglected’ in corpus-based research thus far (Pichler and
Levey 2010: 17). This section introduces the corpus used in this study, before detailing the
quantitative methodologies implemented and the problems overcome.
3.1.
NECTE
The data analysed in this study were obtained using the Newcastle Electronic Corpus
of Tyneside English (NECTE)2. This corpus amalgamates two sociolinguistic projects ⎯ the
1960s’ Tyneside Linguistic Survey (TLS) and the 1990s’ Phonological Variation and Change
in Contemporary Spoken English (PVC) scheme ⎯ which allow the Tyneside dialect to be
studied diachronically3.
NECTE contains spoken data from 73 Tyneside speakers. As Table 1 shows, 37
speakers comprise the TLS sample4 and 36 form the PVC. As they originally formed two
distinct projects, the TLS and PVC were, unsurprisingly, created using two distinct
methodologies. The TLS implemented a traditional one-on-one interview set-up, whilst the
PVC used dyads of personally-related speakers in order to reduce the fieldworker’s input
(Allen et al. 2007: 21). Further inconsistencies between the TLS and PVC are evidenced in the
demographic information they provide. Both record the sex and age-range of the speakers but
differ in their definition of socio-economic class. The TLS supplies the speakers’ educational
level and occupation in terms of whether their job is skilled, semi-skilled or unskilled. The
PVC, however, provides each informant with a ‘broadly defined socio-economic class’ (Watt
& Milroy 1999: 27). The PVC also records the informants’ educational status and occupations
only sporadically, which means that socio-economic class comparisons between (or even
within) the corpora cannot be executed easily. This problem was noted previously by
Barnfield & Buchstaller (2010: 259f), who were forced to carry out a random stratified
sample using unequal numbers of WC and MC speakers. However, after much deliberation
2
I would like to thank Adam Mearns for providing the “Informant-Only” files. The TLS and PVC have been combined with NECTE2 data collected since 2007, creating the Diachronic
Electronic Corpus of Tyneside English (DECTE). However, this investigation will focus upon the TLS and PVC
data only. 4
The TLS originally recorded at least 114 interviews (Allen et al. 2007: 20), of which only 37 survive with ‘all
corpus components’ intact. 3
JOANNE BARTLETT
6
and many attempts to find a theoretically-sound constant, I decided against coding for social
class in the present study.
Table 1: The distribution of speakers across NECTE
TLS - 1960s
TLS - 1960s
PVC - 1990s
PVC - 1990s
MALE
FEMALE
MALE
FEMALE
YOUNGER
8
11
9
11
OLDER (41+)
9
9
8
8
TLS = 37
PVC = 36
NECTE = 73
Despite the problems noted above, NECTE remains a valuable resource as it contains
over 300,000 spoken words. Although the use of data collected in earlier time periods can
evidently place ‘constraints on […] contemporary research’ (Barnfield & Buchstaller 2010:
259), NECTE nonetheless provides an excellent opportunity to investigate linguistic change
longitudinally. Moving on, I now detail the sample used and methods implemented in the
collection of the CFL tokens.
3.2.
The sample and CFL collection
As NECTE contains a small number of speakers who produce a relatively low number
of CFL tokens, I chose to analyse the data produced in its entirety. Therefore, the social matrix
of the speakers studied in the current analysis corresponds to those detailed in Table 1.
Although some would argue that a random stratified sample should be used due to the
slightly uneven speaker numbers per cell, this method would mean eradicating many of the
already infrequently-occurring tokens. This would then lead to the analysis of very small
numbers; a circumstance which would, in itself, be subject to criticism. Instead, I take heed
from previous studies which note that, whilst raw token numbers may misrepresent results
due to the analysis of different speaker numbers or conversation lengths (Miller & Weinert
1995: 387), procedures such as CFL Frequency per N words ‘[provide] a normalising measure
for accountable comparisons to be made’ (Levey 2006: 424). The word count for each of the
speakers in NECTE was calculated manually by editing the transcripts to remove the speaker
codes and transcription conventions which would otherwise distort the actual totals.
Although the search for function words such as LIKE can be a laborious task involving
an ‘unacceptable amount of time-consuming “manual” intervention’ (Blackwell 2000: 12),
NECTE contained only 3,508 instances of the feature in any of its roles. This number was
easily analysable, thus permitting the assessment of each LIKE token as to whether it
constituted an instance of CFL. By extension, this number allowed the entire corpus to be
sampled. Firstly, using Notepad++ (Ho 2011) as a concordance programme, I manually
extracted any instances of LIKE which may have qualified as a desired CFL token, excluding
those which obviously functioned as prepositions, conjunctions, quotatives and so forth.
However, some examples were ambiguous on paper as to whether they were instances of CFL
or Constituent-Initial LIKE (CIL), as in (4):
(4)
a.
b.
CFL: […she’s forward LIKE] [she’s a forward lass…]
CIL: […she’s forward] [LIKE she’s a forward lass…]
[PVC/01B]
[PVC/01B]
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
7
To resolve this ambiguity, I listened to each of the potential CFL tokens using the The
Talk of the Toon website (Corrigan et al. 2010-2012)5 and assigned them CFL status when
LIKE followed the constituent preceding it with no pause separating the two, nor was it
accompanied by the rise in intonation which would occur with CIL6. In several cases, potential
tokens had to be excluded from the analysis where the sound recordings were poor and
prosodic features were unavailable to confirm CFL status.
3.3.
Coding and statistical analysis
Although the results will be addressed in more detail later, it is worth noting that the
collection methods described above found 565 tokens of CFL across the entirety of NECTE.
These were, of course, distributed unevenly across the 73 speakers in the corpus, with some
producing many tokens and some producing none.
As Table 1 suggests, I coded each token according to the corpus from which it came,
as well as the speakers’ sex (male; female) and age (younger; older). Regarding the
classification of the latter, the speakers were stratified according to whether they were aged
40 and below, or 41+. This age range was chosen because previous studies regarding both
focuser and quotative LIKE have noted similar age distinctions between the use and non-use
of these discourse-pragmatic markers (Dailey-O’Cain 2000).
In addition to these extralinguistic variables, I also coded for the pragmatic function
of each token. Although discourse-pragmatic markers are known to be ‘notoriously
multifunctional’ (Levey 2006: 425), several broad categories were introduced to avoid having
so many different functions (cf. Columbus 2009) that quantitative analysis could not be
conducted upon, what would be, very small numbers. The process involved in choosing these
categories will be detailed in Section 4.2, but it is hoped that these will make statistical
analysis more manageable and, by extension, more effective.
As remarked earlier, CFL ‘is not a variable sensu stricto’ (Kerswill & Williams 2002:
101) as it does not have equivalent alternatives. Thus, it appears impossible to formulate a
situation for statistical analysis in which CFL can act as the dependent variable, which
requires, unsurprisingly, at least one variant for alternation. Consequently, I have statistically
analysed this data in two ways. Firstly, by calculating the relative CFL frequencies per 1,000
words, I have been able to investigate comparable diachronic CFL distributions within the
aforementioned social groups. Furthermore, as the distribution of a feature across its different
contexts of use is said to be most important in determining the status of that form
(Tagliamonte 2002: 732), I later enlist the pragmatic functions of the CFL tokens as the
dependent variable, whose use can be correlated with the extralinguistic variables.
This section has highlighted several difficulties in the quantitative, corpus-based study
of CFL. However, I have attempted to resolve or amend such problems in order to provide a
larger, and therefore more representative, data-set for the ensuing analysis.
5
This website supplies the NECTE and NECTE2 data to the general public. Ahead of the DECTE launch in
2012, I made use of the sound files already located here. 6
An exception to this is the use of CFL in an interrogative construction, as in (1c). Although the speaker has
clearly reached the end of their turn, their intonation does not fall because questions are, in technicality,
‘incomplete propositions’ (Schiffrin 1987: 291). JOANNE BARTLETT
8
4.
Results
This section presents the results pertaining to the qualitative and quantitative study of
the CFL tokens found in NECTE. First, the sociolinguistic distribution of this feature will be
studied across time. Following this, the pragmatic functions of the Tyneside CFL tokens will
be discussed before the two approaches are combined.
4.1.
Diachronic distribution in Tyneside English
contains 3,508 tokens of LIKE. Table 2 displays a breakdown of this number
and shows that, at 1,162 and 2,346 tokens in the TLS and PVC respectively, LIKE-use has more
than doubled in the time between which the two sub-corpora were recorded. This result is
unsurprising due to the emergence of, for example, the quotative BE LIKE form in the 1980s
(cf., for example, Butters 1982).
NECTE
Table 2: Distribution of LIKE and CFL in NECTE
CORPUS
NO. OF LIKE
TOKENS
NO. OF CFL
TOKENS
% OF LIKE = CFL
TLS - 1960s
1162
299
25.73
PVC - 1990s
2346
266
11.34
NECTE TOTAL
3508
565
16.11
Table 2 shows that 565 CFL tokens were identified in NECTE. There is some suggestion
that CFL-use has declined in Tyneside English because the raw totals in the TLS and PVC show
a slight decrease over time, although this fluctuation is minimal and conclusions cannot be
made easily. The proportionality of CFL tokens within the total LIKE numbers suggest a
similar result, having more than halved from a little over 25% in the TLS to only 11% in the
PVC.
Though these results are not shown, the CFL-N/1,000 word frequencies for the entirety
of both the TLS and PVC also support this hypothesis. In the TLS, CFL-use totalled at 2.68
occurrences per 1,000 words, whilst the PVC’s equalled only 1.26/1,000. Therefore, CFL-use
has seemingly more than halved between the 1960s and 1990s; a finding which suggests that
its productivity in Tyneside English has waned. It is difficult to pinpoint a reason for this
apparent decline, though it could be hypothesised that the presence of the focuser LIKE form
in the PVC files has contributed to this. This feature occurs ‘in cases where what the speaker is
saying is obviously imprecise’ (Schourup 1985: 38), so it could be argued that CFL is no
longer needed to ‘counter[…] objections and assumptions’ (Miller & Weinert 1995: 366). In
other words, if the hearer is able to realise that the use of focuser LIKE signals
‘nonequivalence’ (Schourup 1985: 42) between speech and meaning, they should not use CFL
to question nor protest what is being said because, by using focuser LIKE, the speaker has
admitted that their utterance is merely an approximation. Thus, the focuser LIKE discoursepragmatic marker may now be occupying one of the roles which CFL is said to play, although
more specific empirical investigation is needed before this hypothesis can be confirmed.
Table 3 displays the relative CFL frequencies per 1,000 words according to the social
matrix put forth in Table 1. Analysing the material in this manner is important as it accounts
for the speakers who do not produce any CFL tokens, given that ‘accurately measuring
frequency in a population requires the consideration of frequencies of zero’ (Waters 2008:
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
9
27). The results in Table 3 suggest that in Tyneside English, CFL is a feature which most
commonly occurs in the speech of both males and individuals aged below 41.
Table 3: Frequency of CFL/1,000 words as distributed across NECTE
TLS - 1960s TLS - 1960s PVC - 1990s PVC - 1990s MALE FEMALE MALE FEMALE YOUNGER 5.3 1.36 2.24 1.14 OLDER (41+) 4.34 0.43 0.92 0.71 Before investigating the effects of the extralinguistic variables upon CFL-use
separately, Table 3 deserves further comment. The relative frequencies displayed above show
that, in both the TLS and PVC data, the male speakers aged below 41 use CFL most frequently,
whilst the older female informants are the most infrequent users. Table 3 also appears to
show that sex has a greater influence than age upon CFL-use in the TLS data, as the male
speakers in both age ranges use the feature more frequently than their female counterparts.
Interestingly, this effect has seemingly changed over time. In the PVC data, age appears to be
a stronger stipulator of CFL-use because the younger speakers in both sexes use the form more
frequently, at 2.24/1,000 and 1.14/1,000 words, than the older informants do. CFL-use shows
a decrease over time in each corresponding cell of the matrix in Table 3, except for the older
females whose relative use has climbed from 0.43/1,000 words to 0.71/1,000 words. Despite
this, their use of the feature continues to remain incredibly infrequent and seems unlikely to
be reversing the apparent trend of CFL decline in Tyneside.
4.1.1. Sex
Figure 1 displays the CFL-N/1,000 words for both sexes for both the TLS and PVC.
Figure 1: CFL Frequency/1,000 words according to sex in the TLS and PVC
χ2 = 0.478; d.f. = 1, p > .05
Figure 1 shows that across time, male speakers use more CFL than females. Despite
this, such a difference between the sexes is much more pronounced in the TLS data. Here, the
male speakers use CFL more than five times as often as their female counterparts, whereas the
10
JOANNE BARTLETT
males use CFL only less than twice as frequently as the female interviewees. Indeed,
Figure 1 demonstrates that time has been witness to a large decline in CFL-use by male
speakers, from 4.81/1,000 words in the TLS to only 1.67/1,000 in the PVC. It could be argued
that this feature is recycled diachronically across the sexes (cf. Dubois & Horvath 2000;
Buchstaller 2006), so males in the 1990s simply represent a “dip” in CFL-use. On the other
hand, this may simply reflect a decline in CFL-use more generally ⎯ especially as a huge
majority of TLS tokens came from the male interviewees (see Table 3). Thus, a large decline
in its use by this group causes a decrease in the use of CFL overall.
PVC
4.1.2. Age
Figure 2 presents the comparable (N/1,000 word) frequencies for the
present in the TLS and PVC, according to the age of the speaker.
CFL
tokens
Figure 2: CFL Frequency/1,000 words according to Age in the TLS and PVC
χ2 = 0.022; d.f. = 1; p > .05
Figure 2 shows that CFL-use has decreased across both age ranges diachronically. This
result holds greater accuracy than focusing on raw token numbers because the averaged
frequencies account for the different speaker numbers per cell, as well as the inevitable
variation in interview length. The result itself adds weight to the suggestion that CFL presence
in Tyneside English is falling, especially as the speakers aged below 41 show a slightly
steeper decline in CFL-use and this age range is likely to have the greatest influence upon
ongoing linguistic change.
However, both the TLS and PVC demonstrate that CFL is a feature more commonly
found in the vernacular of younger speakers, at 3.22/1,000 words and 1.5/1,000 across time.
This result mirrors the age distinction of both focuser and quotative LIKE (such as Schourup
1985: 44; Romaine & Lange 1991: 251; Dailey-O’Cain 2000: 62). Therefore, whilst many
today criticise the incessant and meaningless production of LIKE by young people, it seems
that its use as a discourse-pragmatic marker has, for a long time, been a linguistic feature
typical of this age group. Despite this, Figure 2 shows that CFL is still present in the speech of
the older speakers ⎯ especially in the TLS. Indeed, the relative frequencies both within and
between the two corpora are actually so slight when segregated by the two age ranges that
they fail to gain statistical significance.
The most interesting finding here is that CFL-use is apparently not utilised at a
consistent level during the lives of these speakers. This analysis is possible because the
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
11
speakers aged below 41 in the TLS are all grouped in the older category at the time of the PVC
recordings. Therefore, as the relative CFL frequency for the younger speakers in the TLS
equals 3.22/1,000 words whilst that of the older speakers in the PVC equals 0.8/1,000, it
seems that the use of this feature has been lost from the linguistic repertoires of the speakers
over time. This results parallels with Schweinberger’s (in press) finding that, in Ireland, CFLuse declines as the speakers’ age increases, at least when his informants were not additionally
segregated by sex. Of course, the TLS and PVC used different informants so firm conclusions
can only be established when a longitudinal study using the same speakers is carried out.
Moreover, Cheshire (2007: 162) suggests that such assumptions cannot be made in this way
‘unless [one] can be sure that the data from different age groups is comparable in both
content and context’. However, as sociolinguists already make ‘inferences about what is
happening in the community as a whole’ (Labov 1994: 4) using small samples, the hypothesis
that CFL-use declines as the speakers increase in age can be upheld if the TLS and the PVC are
indeed representative of their respective Tyneside communities.
This section has detailed the sociolinguistic distribution of the CFL form in Tyneside
English diachronically. This method has, in isolation, produced some interesting results.
However, reliance upon such frequencies alone is said to be inadequate for studying change
as the numbers may differ as the result of non-linguistic factors (Tagliamonte 2002: 732;
Cheshire 2007: 161f). Therefore, the analysis which follows takes pragmatic function as its
dependent variable in order to observe whether the intralinguistic roles which CFL performs
have changed diachronically, both alone and in comparison with the aforementioned
extralinguistic variables.
4.2.
Pragmatics in Tyneside
The investigation of CFL as a discourse-pragmatic marker cannot be achieved without
also discussing the pragmatic roles that it performs. It was noted earlier that CFL is said to
have a variety of functions, though it seems that many of these have overlapping reference.
This study, therefore, condenses these roles into three broad pragmatic categories, arguing
that the main roles of CFL are to (i) accompany opinions in the anticipation of objections, (ii)
provide explanations and (iii) to ask for explanations or further information. These are
addressed as OPINION, EXPLAINING and INTERROGATIVE respectively. These categories come
from work by Miller (2009), who describes the functions of CFL most succinctly. Indeed, his
classification has easily accounted for the entirety of the CFL data found in NECTE. These
categories are addressed individually below and provide examples as to how these pragmatic
roles function in Tyneside English.
4.2.1. OPINION
Miller’s (2009: 336) contention that CFL anticipates objections from the hearer relates
to others noting that it mitigates confrontation (Kallen 2005, 2006; Corrigan 2010) and acts
as a softener (Columbus 2009: 19). Therefore, these roles have been grouped together under
the OPINION category. This accounts for the fact that CFL is used to tone down the personal
opinion of the speaker which, in turn, reduces the likelihood of the interlocutor protesting
what has been said, as below:
(5)
a.
b.
Oh it’s not a very nice place the eh this end LIKE
I wouldn’t say that LIKE
[TLS/18]
[PVC/06B]
In (5a), the speaker is providing a personal opinion regarding the local area. Rather
JOANNE BARTLETT
12
than hedging the validity of his opinion per se, CFL in this instance signals to the interviewer
that the speaker has reasons for this view. This prevents his opinion from being contradicted
by the hearer. Therefore, the use of CFL signals that the speaker has more to say and may
even be inviting the interviewer to ask what his reasons are to hold such an opinion.
Therefore, in addition to mitigating potential disagreement and confrontation, CFL has
additionally acted here as a device to propel the conversation forward. This relates to the
aforementioned suggestion that CFL may act as a turn-transition device (Schiffrin 1987: 218)
in discourse. It could be argued that instead of CFL acting as a ‘focusing device’ (Miller &
Weinert 1995: 366), its use in this construction aims to actually remove emphasis from the
negative view which he is expressing. The use of CFL in (5b) helps to clear up an assumption
that the hearer has made by suggesting the speaker holds an opinion that, in reality, they do
not. However, in clarifying this mistake, [PVC/06B] may fear offending or humiliating the
hearer: a situation which can be avoided by the use of CFL. This feature softens what could be
a rather hostile correction.
4.2.2. EXPLAINING
Kallen (2006: 12) notes that, as with focuser LIKE, CFL can focus attention on
particular pieces of information. This relates to Miller (2009: 336) arguing that CFL
encourages the hearer to accept the explanation being provided, though signals that the
speaker (as before) anticipates objections. These have been collapsed here as an EXPLAINING
category, in that CFL aids in the providing of information or the closing of narratives
(Corrigan 2010: 100). Thus, CFL may signal the end of the speaker’s turn, suggesting to the
hearer that the speaker has no more to add to their explanation and it must therefore be
accepted as it is. NECTE provides many examples of CFL being used in this way, as (6)
demonstrates:
(6)
Eh well it’s machine engraving I do LIKE
[TLS/12]
This example shows that speaker [TLS/12] has provided an answer to a question
posed by the interviewer. However, the inclusion of CFL suggests that a misunderstanding has
occurred, which the speaker is aiming to correct in a manner that will not appear as
condescending. CFL contributes to this by anticipating an objection from the hearer, to whom
it signals that an answer has been provided fully and accurately. Thus, it reinforces the
proposition which has been uttered (Beal 2004: 135). Indeed, it could be argued that the use
of CFL in this manner indicates to the hearer that another question must be asked if additional
information is desired.
4.2.3. INTERROGATIVE
has been noted to occur in questions in order to request further explanations
(Miller 2009: 339; Columbus 2009: 17) and clarification (Schourup 1985: 47; Miller &
Weinert 1995: 390). CFL occurs in this manner in Tyneside English too, as (7a-b)
demonstrate, resulting in the implementation of this INTERROGATIVE category.
CFL
(7)
a.
b.
I was wondering how you got the address LIKE?
Where’s she playing at LIKE?
[TLS/27]
[PVC/01B]
Unsurprisingly, both (7a) and (7b) request further information with or without the
inclusion of the CFL feature. (7a) follows Miller’s (2009: 339) observation that the use of CFL
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
13
in interrogatives signals the need for explanations to be provided, but has the additional role
of mitigation, as described for the other pragmatic categories too. This example could be
produced in a rather confrontational manner, but is softened by the inclusion of CFL. Here,
CFL signals that the speaker is not accusing the hearer of, for example, stealing the address or
gaining it in some untoward fashion; CFL makes the interrogative merely appear somewhat
light-hearted. Of course, prosodic features, such as the speaker’s tone of voice, are also
important to such analysis and should be incorporated into future pragmatic studies on CFL.
(7b) signals that the interlocutor of [PVC/01B] has failed to provide enough information on
the music concert being discussed. CFL signals this fact to the hearer without belittling them
for not having provided an optimally relevant utterance (Wilson & Sperber 2004).
The collapse of the functions ⎯ however similar ⎯ into three pragmatic categories
may induce criticism from some scholars. Cheshire (2007: 158), for example, contends that
by stripping discourse-pragmatic markers of their multifunctionality, the ‘most salient
characteristic’ of these forms is lost. This statement is difficult to dispute, nor is that my
intention, but the quantitative study of CFL (or indeed discourse-pragmatic markers in
general) requires compromise as too many functions restricts statistical analysis from being
implemented, as previously noted.
4.3.
Pragmatic function in the variationist paradigm
In order to both represent CFL function in Tyneside and create an optimal situation for
quantitative study, this section implements the three broad pragmatic categories described
above. Taking Miller’s (2009) categorisation, I argue that the main uses of CFL are to (i)
accompany opinions in the anticipation of objections, (ii) provide explanations and (iii) to ask
for explanations or clarification. Here, these functions are labelled OPINION, EXPLAINING and
INTERROGATIVE respectively. This section excludes 27 of the tokens analysed previously due
to their functions being UNCLEAR7. Until now, the pragmatic function of CFL was not under
investigation so these tokens did not need to be excluded since they remained an instance of
CFL, regardless of their role in discourse.
The bar graphs below represent the CFL tokens as their percentage distributions across
the OPINION, EXPLAINING and INTERROGATIVE pragmatic categories. The raw token numbers
which create these percentages are also provided within each segment. These numbers alone
are not directly comparable because the TLS and PVC present different total frequencies of CFL
tokens, but have been included to provide a complete representation of the data.
4.3.1. CFL function in TLS and PVC
The percentage distribution of the functions that the CFL tokens perform in the TLS
and PVC are shown in Figure 3. Interestingly, a slight divergence in its pragmatic roles is
apparent. This finding suggests that the function, or functions, for which a linguistic feature is
used does not remain stable over time. Indeed, this also suggests that CFL (which is an older
form within the LIKE paradigm) is continuing to evolve as an individual over time.
Before discussing the changes, it is worth noting that CFL-use within the OPINION
category ⎯ that is, to hedge or validate the speakers’ opinions so as not to alienate the hearer
⎯ remained relatively stable over time. The use of CFL in such constructions accounted for
38.2% (N=108) of its total occurrence in the TLS, a number which is only slightly higher, at
7
This UNCLEAR category was implemented for instances where poor sound recordings meant that the
indistinguishable utterance preceding CFL prohibited its function from being determined. 14
JOANNE BARTLETT
40.4% (N=103), in the PVC. This stability across time suggests that the
represents one pragmatic role in which CFL-use is robust.
OPINION
category
Figure 3: Percent (%) distribution of CFL function in the TLS and PVC
χ2 = 36.513; d.f. = 2; p < .001
However, CFL-use in the EXPLAINING and INTERROGATIVE pragmatic categories do
display change over time. In both the TLS and PVC, CFL-use with an EXPLAINING function is
most frequent, accounting for 60.1% (N=170) of the CFL tokens in the TLS and 44.3%
(N=113) of those in the PVC. Despite this, there has been a decline in the use of the feature in
this role over time ⎯ leading to its proportional use in the PVC being only slightly higher than
its use in the OPINION category. CFL-use in the INTERROGATIVE category contrasts this by
displaying a significant rise over time from only 1.8% (N=5) of CFL occurrences in the TLS to
15.3% (N=39) in the PVC. However, its use in this construction type remains fairly infrequent
overall.
It is difficult to suggest an intralinguistic reason for these changes, but an
extralinguistic one regarding the elicitation of the data can be proposed. The TLS and PVC
were recorded using different sociolinguistic techniques: the former implemented one-on-one
interviews with one interviewee, whilst the latter asked two friends or family members to talk
with minimal interaction from the interviewer. Therefore, it seems acceptable to suggest that,
in the TLS interviews, the interviewees would ask fewer questions because they were not in a
position to dictate the progression of the interview. The PVC interviewees, however, could
question one another freely. Consequently, it can be argued that CFL-use may not have
actually increased within the INTERROGATIVE context, but that this result is merely
representative of an increased use of questions in the PVC recordings. Indeed, the use of CFL
in the INTERROGATIVE context may have been observed more frequently had the speech of the
interviewers been investigated too. Of course, further investigation probing the number of
interrogative forms in the corpora generally is needed before this suggestion can be
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
15
confirmed.
This change in interview set-up may also explain the diachronic decline in CFL-use
within the EXPLAINING category. The TLS interviews, as suggested, consisted simply of
questions from the interviewer and answers from the interviewee, meaning that these
informants were constantly providing explanations. However, as the PVC was recorded using
pairs of closely-related individuals, explanations ⎯ or at least explanations which require CFL
to eliminate objections ⎯ were no doubt used less frequently due to the contextual
knowledge present between the interlocutors. Therefore, it seems that, in at least some cases,
the diachronic change in CFL function may be less concerned with an internal pragmatic shift
in the CFL form and more to do with the interview techniques used to elicit the tokens. This
situation is noted by Macaulay (2002: 749), who contends that the comparison of data
‘collected under very different circumstances’ is difficult. However, he also notes that such
difficulties are unimportant if ‘the aim is to explore the effect of different situations’
(Macaulay 2002: 749). Therefore, this study can arguably be used to demonstrate that
linguistic features fulfil different functions when collected in different situations. This is a
factor, therefore, which should always be considered by those carrying out corpus-based
studies of language or when recording data under different circumstances.
The change in CFL function between the TLS and PVC corpora gains statistical
significance (p < .001). However, there was the potential that this significance may have been
caused by the large rise in CFL-use in the INTERROGATIVE context. This increase which may
not have been purely linguistic. However, a chi-square test eliminating the 44
INTERROGATIVE occurrences resulted in continued significance, albeit only marginally
(p=.049). However, as the level of significance decreased from p < .001 to p = .049 when the
INTERROGATIVE tokens were removed, it is clear that the original significance level presented
in Figure 3 was due in some part to the number of tokens in this category diverging radically
across time.
4.3.2. CFL function: sex
Moving on, the result depicted above was collapsed according to the sex of the TLS
and PVC informants. Figure 4 displays both the raw token frequencies and comparable
percentage distributions created, demonstrating that the sex-based division of CFL function
has somewhat changed over time.
Taking the TLS data in the two leftmost columns, Figure 4 shows that the CFL tokens
used by both sexes show an almost identical percentage use in each pragmatic category. This
finding is even more interesting when one considers that the raw numbers of CFL produced by
the two sexes are incredibly different. The OPINION category, for example, accounts for 38%
(N=89) of all CFL tokens uttered by the TLS male speakers and 38.8% (N=19) of the total
female use. Therefore, although the number of raw tokens that they produce differs,
proportionally, both male and female speakers in the TLS use CFL in an identical manner.
Unsurprisingly, this stability causes the chi-square analysis to generate a hugely insignificant
result (p = .981). However, given the current discussion, it would be incorrect to argue that
there is nothing of interest occurring in the data.
This stability is no longer apparent in the PVC data, where male and female speakers
use CFL to fulfil different pragmatic functions. The EXPLAINING category shows the greatest
divide in CFL-use, with these tokens constituting 50.9% (N=57) of all CFL occurrences in the
speech of the PVC women, but only 39.2% (N=56) in that of the PVC men. These men use CFL
in the OPINION category only marginally more frequently than the women, but proportionally
use CFL in the INTERROGATIVE category almost twice as often, at 18.9% (N=27) compared to
the 10.7% (N=12) for females. Although these numbers do not lead to a statistically
JOANNE BARTLETT
16
significant result, the significance value for the sex differences in the PVC is closer to being so
(p=.087) than the results in the TLS (p=.981). Hence, the PVC data suggests that the sexes have
started to diverge in their use of CFL to fulfil particular pragmatic functions.
Figure 4: Percent (%) distribution of CFL function in the TLS and PVC according to sex
Male vs. Female: χ2 = 0.062; d.f. = 2; p > .05
TLS Male vs. TLS Female: χ2 = 0.039; d.f. = 2; p > .05
PVC Male vs. PVC Female: χ2 = 4.888; d.f. = 2; p > .05
Figure 4 shows that, across the two corpora, the proportional use of CFL in the
context remains quite static since this hovers around 40% of the total CFL uses in the
social categories. Over time, males show a decline in the use of CFL in EXPLAINING contexts,
with this number falling from 60.3% (N=141) in the TLS to only 39.2% (N=56) in the PVC.
This fall contrasts the males’ increased use of CFL within the INTERROGATIVE categorisation.
OPINION
4.3.3. CFL function: age
In contrast to speaker sex, the age of the informants appears to differentiate the use of
in specific pragmatic roles more-readily. Figure 5 displays the percentage distributions
and raw token numbers of the three pragmatic categories for both the younger and older
speakers in the corpora.
Interestingly, Figure 5 shows that the use of CFL in the INTERROGATIVE contexts is,
overwhelmingly, a trait of younger speakers, with only one instance occurring in the speech
of the older informants. Figure 5 also demonstrates that across each of the age groups in the
TLS and PVC (bar the PVC’s younger speakers) CFL-use in the context of EXPLAINING is most
common. The younger informants in the PVC, however, favour CFL-use in the OPINION
category, which accounts for 46.8% (N=88) of their total CFL occurrences. This number
contrasts their older counterparts by more than double, who use CFL in this manner merely
CFL
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
17
22.4% (N=15) of the time. Therefore, as this result depicts a divergence from the relatively
similar OPINION and EXPLAINING percentage uses by the two age groups in the TLS, it appears
that CFL function is becoming specialised in its use by speakers of different ages.
Figure 5: Percent (%) distribution of CFL function in the TLS and PVC according to age
Younger vs. Older: χ2 = 39.910; d.f. = 2; p < .001
TLS Younger vs. TLS Older: χ2 = 4.795; d.f. = 2; p > .05
PVC Younger vs. PVC Older: χ2 = 39.357; df.. = 2; p < .001
Despite this, CFL-use in the OPINION category is favoured by younger speakers in both
the TLS and the PVC. This result may indicate that this age group hold greater levels of
linguistic insecurity and anticipate objections more so than the older speakers do. In contrast,
the older speakers favour the use of CFL in the EXPLAINING contexts across time. This
suggests that CFL does indeed play different roles in the speech of different age groups. The
validity of this hypothesis is strengthened when one considers the observation that the TLS’
younger speakers moved into the older age category in the PVC. The columns which straddle
Figure 5 show that the use of CFL in the OPINION category has declined from 40.1% (N=65) of
the total number of tokens used by the TLS’ younger informants to 22.4% (N=15) of the PVC’s
older speakers. This result combines with a rise in the use of CFL in the EXPLAINING category,
from 56.8% (N= 92) to 76.1% (N=51), and suggests that the functional use of CFL changes
over the lifetime of an individual. Indeed, it appears that the use of CFL to hedge opinions
declines as Tyneside speakers age and it is instead used to provide explanations.
This section has suggested that the functions which discourse-pragmatic markers
perform can change over time. Indeed, CFL function appears to be evolving and may, in at
least some cases, be used by different social groups for different reasons.
JOANNE BARTLETT
18
5.
Conclusion
In this paper, the fields of pragmatics and comparative sociolinguistics were
amalgamated in order to investigate the CFL discourse-pragmatic marker in Tyneside English.
The number of tokens located in the data were rather small, but they do compare to the
similar numbers found in previous work using ICE-Ireland (such as Kallen 2006; Columbus
2009; Luckmann 2009; Siemund et al. 2009; Schweinberger in press). Thus, as CFL is thought
to be a ‘robust’ (Kallen 2006: 14) feature of Irish English, it can be argued that this
conclusion should be extended to its existence in Tyneside English too.
This investigation has suggested that the frequency of the CFL form in Tyneside has
declined over time and has shown CFL to be a feature more commonly associated with male
speakers. The latter result differs to Luckmann’s (2009) investigation of CFL in Ireland, so it
can be concluded that the social factors determining CFL-use are not universal, but specific to
the area under investigation (cf. Buchstaller & D’Arcy 2009: 318).
CFL has been shown to play pragmatic functions in discourse ⎯ roles that cannot be
performed by its absence. Thus, CFL adds additional meanings to an utterance, demonstrating
that it is not a meaningless linguistic expression. It has been illustrated that these functions
can be collapsed into three distinct pragmatic categories which encompass the CFL tokens
present in NECTE both easily and entirely. Therefore, in contrast to the social factors
aforementioned, it appears that the internal pragmatic roles that this feature holds may be
universal, especially since the three categories used in this study correspond to those
proposed by Miller (2009) to describe his Australasian CFL tokens. Furthermore, variationist
investigation of these function categories has shown that, whilst each of the three pragmatic
roles were present in both the TLS and PVC, their proportional distributions have changed over
time. Therefore, it can be argued that CFL is continuously evolving in order to meet the needs
of its current users.
Finally, this paper has highlighted several difficulties involved in the quantitative
investigation of discourse-pragmatic markers. However, efforts were made to amend the
problems, or at least consider them when discussing the results. Barnfield and Buchstaller
(2010: 261) contend that such an analysis still yields ‘interesting and diachronically reliable
results’ when such awareness is apparent.
Further investigation can extend and add to the findings discussed in this paper by
studying CFL in other dialects. It would be interesting to trace the historical development of
the CFL form in order to observe whether its existence in Tyneside English has come from
Ireland, as is proposed. Furthermore, a longitudinal investigation using the same group of
speakers would be especially exciting, for this would generate the most conclusive and
reliable results as to how CFL patterns across the lifetime of Tyneside individuals.
References
Allen, W., Beal, J. C., Corrigan, K. P., Maguire, W. & Moisl, H. L. (2007). A Linguistic
“Time Capsule”: The Newcastle Electronic Corpus of Tyneside English. In Beal, J.
C., Corrigan, K. P. & Moisl, H. L. (eds.), Creating and Digitizing Language Corpora
⎯ Volume 2: Diachronic Databases, 16−48. Houndmills: Palgrave Macmillan.
Andersen, G. (2000). The role of the pragmatic marker like in utterance interpretation. In
Andersen, G. & Frothier, T. (eds.), Pragmatic Markers and Propositional Attitude,
17−38. Amsterdam: John Benjamins.
Andersen, G. (2001). Pragmatic Markers and Sociolinguistic Variation. Amsterdam: John
Benjamins.
Anderwald, L. (2004). English in the Southeast of England: morphology and syntax. In
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
19
Kortmann, B., Burridge, K., Mesthrie, R., Schneider, E. W. & Upton, C. (eds.), 175−
195.
Barnfield, K. & Buchstaller, I. (2010). Intensifiers on Tyneside: Longitudinal developments
and new trends. English World-Wide 31, 252−287.
Beal, J. C. (1993). The grammar of Tyneside and Northumbrian English. In Milroy, J. &
Milroy, L. (eds.), 187−213.
Beal, J. C. (2004). English dialects in the North of England: morphology and syntax. In
Kortmann, B., Burridge, K., Mesthrie, R., Schneider, E. W. & Upton, C. (eds.), 114−
141.
Beal, J. C. & Corrigan, K. P. (2009). The impact of Nineteenth-Century Irish-English
migrations on contemporary Northern Englishes: Tyneside and Sheffield compared.
In Penttilä, E. & Paulasto, H. (eds.), 231−258.
Blackwell, S. (2000). Looking up look: Discourse markers in the Bank of English. In Kirk, J.
M. (ed.), Corpora Galore: Analyses and Techniques in describing English, 3−16.
Amsterdam: Rodopi.
Buchstaller, I. (2006). Diagnostics of age-graded linguistic behaviour: The case of the
quotative system. Journal of Sociolinguistics 10, 3−30.
Buchstaller, I. & D’Arcy, A. (2009). Localized globalization: A multi-local, multivariate
investigation of quotative be like. Journal of Sociolinguistics 13, 291−331.
Butters, R. (1982). Editor’s note [on be like ‘think’]. American Speech 57, 149.
Cheshire, J. (2007). Discourse variation, grammaticalisation and stuff like that. Journal of
Sociolinguistics 11, 155−193.
Columbus, G. (2009). Irish like as an invariant tag: evidence from ICE-Ireland. Paper
presented at American Association for Corpus Linguistics (AACL) 2009, Edmonton,
Alberta, Canada.
<http://www.ualberta.ca/~aacl2009/PDFs/Columbus2009AACL.pdf> (Last accessed
October 2011).
Corrigan, K. P. (2010). Irish English: Volume 1 ⎯ Northern Ireland. Edinburgh: Edinburgh
University Press.
Corrigan, K. P., Buchstaller, I., Mearns, A. & Moisl, H. (2010-2012). The Talk of the Toon.
Newcastle University. <http://research.ncl.ac.uk/decte/toon/> (Last accessed February
2012).
D’Arcy, A. (2005). LIKE: Syntax and Development. Unpublished Ph.D. Thesis, University of
Toronto.
Dailey-O’Cain, J. (2000). The sociolinguistic distribution of and attitudes toward focuser like
and quotative like. Journal of Sociolinguistics 4, 60−80.
Dubois, S. & Horvath, B. (2000). When the music changes, you change too: sex and
Language Change in Cajun English. Language Variation and Change 11, 287−313.
Fraser, B. (1999). What are discourse markers? Journal of Pragmatics 31, 931−952.
Harris, J. (1993). The grammar of Irish English. In Milroy, J. & Milroy, L. (eds.), 139−186.
Ho, D. (2011). Notepad++. Version 5.9.6.2. <http://notepad-plus-plus.org/> (Last accessed
December 2011).
Kallen, J. (2005). Politeness in Ireland: “In Ireland, it’s done without being said”. In Hickey,
L. & Stewart, M. (eds.), 130−144. Politeness in Europe. Clevedon: Multilingual
Matters.
Kallen, J. (2006). Arrah, like, you know: The dynamics of discourse marking in ICE-Ireland.
Presented at Sociolinguistics Symposium 16, University of Limerick, 6th−8th July
2006. Paper available online at <http://hdl.handle.net/2262/50586> (Last accessed
October 2011).
20
JOANNE BARTLETT
Kerswill, P. & Williams, A. (2002). “Salience” as an explanatory factor in language change:
evidence from dialect levelling in urban England. In Jones, M. C. & Esch, E. (eds.),
Language Change: The Interplay of Internal, External and Extra-Linguistic Factors,
81−110. Berlin: Mouton de Gruyter.
Kortmann, B., Burridge, K., Mesthrie, R., Schneider, E. W. & Upton, C. (eds.) (2004). A
Handbook of Varieties of English ⎯Volume 2: Morphology and Syntax. Berlin:
Mouton de Gruyter.
Labov, W. (1994). Principles of Linguistic Change: Social Factors. Oxford: Blackwell.
Levey, S. (2003). He’s like “Do it now!” and I’m like “No!”. English Today 19, 24−32.
Levey, S. (2006). The sociolinguistic distribution of discourse marker like in preadolescent
speech. Multilingua 25, 413−441.
Luckmann, K. (2009). The pragmatic marker like in clause-final position: Its functional and
social distribution in Irish English. Unpublished MA Thesis, University of DuisburgEssen.
Macaulay, R. (2002). You know, it depends. Journal of Pragmatics 34, 749−767.
Miller, J. (1993). The grammar of Scottish English. In Milroy, J. & Milroy, L. (eds.),
99−138.
Miller, J. (2009). Like and other discourse markers. In Peters, P., Collins, P. & Smith, A.
(eds.), Comparative Studies in Australian and New Zealand English: Grammar and
Beyond, 317−337. Amsterdam: John Benjamins.
Miller, J. & Weinert, R. (1995). The function of LIKE in dialogue. Journal of Pragmatics 23,
365−393.
Milroy, J. & Milroy, L. (eds.) (1993). Real English: The Grammar of English Dialects in the
British Isles. New York: Longman Publishing.
NECTE (Newcastle Electronic Corpus of Tyneside English). Newcastle University.
<http://research.ncl.ac.uk/necte/> (Last accessed February 2012).
Oxford English Dictionary. (2007). United States of America: Oxford University Press.
Penttilä, E. & Paulasto, H. (eds) (2009). Language Contacts meets English Dialects: Studies
in honour of Markuu Filppula. Newcastle upon Tyne: Cambridge Scholars
Publishing.
Pichler, H. & Levey, S. (2010). Variability in the Co-occurrence of Discourse Features.
University of Reading Language Studies Working Papers 2, 17−27.
Romaine, S. & Lange, D. (1991). The Use of like as a Marker of Reported Speech and
Thought: A Case of Grammaticalization in Progress. American Speech 66, 227−279.
Schiffrin, D. (1987). Discourse markers. Cambridge: Cambridge University Press.
Schourup, L. (1985). Common Discourse Particles in English Conversation. New York:
Garland Publishing.
Schourup, L. (1999). Discourse markers. Lingua 107, 227−265.
Schweinberger, M. (in press). The discourse marker LIKE in Irish English. In Migge, B. & Ní
Chíosáin, M. (eds.), New Perspectives on Irish English. Amsterdam: John Benjamins.
Siemund, P., Maier, G. & Schweinberger, M. (2009). Towards a More Fine-grained Analysis
of the Areal Distributions of Non-standard Features of English. In Penttilä, E. &
Paulasto, H. (eds.), 19−45.
Tagliamonte, S. (2002). Comparative Sociolinguistics. In Chambers, J. K., Trudgill, P. &
Schilling-Estes, N (eds.), The Handbook of Language Variation and Change,
729−763. Oxford: Blackwell.
Underhill, R. (1988). Like is, like, focus. American Speech 63, 234−246.
Waters, C. (2008). Actually, it’s more than pragmatics, it’s really grammaticalization.
Generals Papers: Toronto Working Papers in Linguistics, University of Toronto.
A LONGITUDINAL STUDY OF CONSTITUENT-FINAL LIKE IN TYNESIDE ENGLISH
21
Watt, D. & Milroy, L. (1999). Patterns of variation and change in three Newcastle vowels: is
this dialect levelling? In Foulkes, P. & Docherty, G. (eds.), Urban Voices: Accent
Studies in the British Isles, 25−46. London: Arnold.
Wilson, D. & Sperber, D. (2004). Relevance Theory. In Horn, L. R. & Ward, G. (eds.), The
Handbook of Pragmatics, 607−632. Oxford: Blackwell.
Joanne Bartlett
School of English Literature, Language and Linguistics
Newcastle University
Newcastle upon Tyne
United Kingdom
NE1 7RU
[email protected]