Hedging in Political Discourse

An Analysis of Hedging in an American City Council

Ellen Player Pellby

2013

Bachelor Thesis, 15 Credits

English C
English in Depth: Undergraduate Course for General Purposes

Tutor: Kavita Thomas

Examinator: Marko Modiano
Abstract

Gävle University
Bachelor Thesis, 15 Credits, Spring Term 2013
C-Level, English
Title: Hedging in Political Discourse – An
Analysis of Hedging in an American City Council
Author: Ellen Player Pellby
Supervisor: Kavita Thomas
Key Words: The Androcentric Rule, Hedges,
Political Discourse.

This thesis seeks to investigate the usage of hedges in political discourse in the Tampa
City Council for the purpose of examining whether or not women hedge more than men
in this area. An analysis of the occurrence of hedges illustrated that women hedged
more than men for various purposes in this meeting. These occurrences mostly involved
the epistemic modal function and shields which indicate uncertainty about the utterance
and certainty about the utterance respectively. The results also illustrate how political
discourse is still an area dominated by men in the sense that men had significantly more
speech time than women during this meeting. However, the results also disprove
Lakoff’s claim that women hedge simply to signal uncertainty and tentativeness.
Table of Contents
1. Introduction 3
1.1 Aim and Scope 4
2. Previous Research 5
2.1 Definitions of Hedges 6
2.2 The Function of Hedges 8
2.3 Hedges in Other Domains 9
2.4 Hedges and the Androcentric Rule 11
2.5 Hedges in Political Speech 12
3. Methodology 15
3.1 Data 15
3.2 Annotation Procedure 15
3.3 Validity and Reliability 21
4. Results 23
5. Conclusion 29
References 31

2
 ”Men will be seen to behave linguistically in a way that fits the writer’s
view of what is desirable or admirable; women on the other hand will be
blamed for any linguistic state or development which is regarded by the
writer as negative or reprehensible” (Coates, 2004: p. 10).

1. Introduction
Gender research in the field of linguistics is often associated with the linguist
Robin Lakoff, who in 1973 published her study Language and Woman’s Place,
where her standpoint exemplifies what is referred to as “the androcentric rule”.
This rule states that what is regarded as the language norm is based exclusively
on men’s way of speaking, which implies that women’s use of language deviates
from this norm and is regarded as non-standard and negative. For instance,
several linguists agree that men’s speech is powerful while women’s speech is
tentative and uncertain (Coates, 2004). Lakoff writes about uncertainty that “…
it submerges a woman's personal identity, by denying her the means of
expressing herself strongly, on the one hand, and encouraging expressions that
suggest triviality in subject-matter and uncertainty about [subject-matter]…”
(Lakoff, 1973: p. 48). According to Lakoff, tentativeness is a linguistic feature
associated with women and expected of them in conversation.
However, as women today are much more included in society, the
androcentric rule is challenged and may be on the wane. Since the publication of
Language and Woman’s Place, much research has been carried out in the field
of linguistics and gender. Stereotypes associated with both genders have been
questioned and further investigated by means of formal studies by researchers
such as Jennifer Coates (2004) and James J. Bradac, Anthony Mulac and Sandra
A. Thompson (1995). Lakoff found that some linguistic features distinguish
women’s speech from men’s, such features being women’s use of super-polite
forms, indirect requests and hedges. Hedges are mitigating words or phrases
such as sort of, probably and I’m fairly certain and are used to diminish the
force of an utterance as well as save a speaker’s face. According to Lakoff, these
features are not used as much by men, as they signal uncertainty, which
contradicts the expectation that men speak powerfully. However, the study
carried out by Bradac, Mulac and Thompson examines how often the frequency
of hedges expressing uncertainty is communicated by men and women in
problem-solving interaction, and shows results which counter Lakoff’s claims,

3
 namely that men actually use more hedges than women do in problem-solving
interaction (1995).

Despite this research denying that women’s use of language is subordinate

to men, in fact, some linguists still agree with Lakoff that women’s use of language
is subordinate to men’s. However, since women today have a more active role in
society than earlier, it is natural that their influence and power has grown in several
areas, for example in the workplace. For instance, women appear in professions
nowadays which were dominated by men not so long ago. One such domain is
politics, where female politicians are nearly as common as male politicians today.
This paper will address Lakoff’s claim that women use hedging to signal uncertainty
considerably more than men. The question which will be investigated in this essay is
as follows: Is women’s usage of hedges in political meetings greater than men’s?
The study aims to examine the use of hedges as mitigating words among men and
women in an American district council meeting.

1.1 Aim and Scope

The aim of this essay is to investigate the language women and men use in political
speech; focusing on hedges as mitigating words. This study compares men’s and
women’s use of hedging in a council meeting in America. The final results hopefully
lend support against Lakoff’s claim about how women repeatedly signal uncertainty
when using hedges.
The essay will discuss previous research about hedges in different
domains, and then explain the methodology and data. The study aims to analyze men
and women’s use of hedges in political discourse. Finally the results and an analysis of
the collected data will conclude the essay.

4
2. Previous Research
This chapter begins with several researchers’ attempts to define the linguistic
construction hedging, followed by what previous research has claimed about the
function of hedges. Hedges in relation to the androcentric rule will be highlighted, that
is the claim about how men’s way of speaking is the desirable one. Finally hedges in
political speech will be discussed.
Robin Lakoff’s work claimed women use a large number of hedges
compared to men, and that in doing so they signaled a greater propensity for uncertainty
and tentativeness (1973). Furthermore, she claimed that men did not use hedging
devices at all in their speech since they are taught at an early age to speak powerfully.
However, the data to support these claims was collected by Lakoff herself by means of
informal conversations within her own social network, with no regard to social factors
or controlling her participants for equal numbers of men and women.
Much research has been carried out on gender in the field of linguistics
since Lakoff’s seminal research. Several later studies show different results regarding
stereotypes associated with men and women. For instance, women are said to use more
hedges and tag questions than men, but researchers have various views about whether
these express uncertainty or not (Holmes, 1990). Janet Holmes found in her study that
hedges are multifunctional depending on situation, context and intonation. They could
signal uncertainty in some cases, but they are also used to signal politeness and a
positive attitude towards the other speaker in conversation, for example: “He has got a
sort of bad attitude,” or “Maybe I could sort of ask you out sometime?” When speakers
ask others for a favor, Bradac, Mulac and Thompson found in their study that women
used hedges when they wanted to signal certainty about something, while men used
hedges to signal uncertainty (1995). Their study analyzed hedges in problem-solving
interaction, and found that men tended to hedge more than women did in this type of
interaction.
The Androcentric Rule discussed by Lakoff claims that children are
socialized early in life, so that boys are taught to act superior in that they are allowed to
be loud and use taboo words for instance. At the same time girls are taught by their
parents and teachers that silence is feminine and expected of them (1973). Lakoff
discusses linguistic features associated with femininity, such as tag questions, hedges
and politeness, which, according to her, all deviate from the Androcentric Rule in the
sense that: “… [the girls’ and women’s] speech is in many ways more polite than that of
boys or men, and the reason for this is that politeness involves an absence of strong

5
statements, and women’s speech is devised to prevent the expression of strong
statements” (Lakoff, 1973: p. 57). Other ways in which women’s language deviates
from the Androcentric Rule according to Lakoff are by employing silence, super-polite
forms and weaker vocabulary. It should be noted however that her study was based on
introspective studies among her own social networks and intuition, which makes the
results questionable. Accurate results require information about social factors such as
age, educational background and gender, none of which Lakoff included in her study.
Next follows a discussion about definitions of hedges among linguists.

2.1 Definitions of Hedges

There are several different definitions of hedges corresponding to different types, face-
saving are not the only ones. Verbal hedges could be words or phrases used to mitigate
the significance of an utterance in order to save a speaker’s face. Face-saving in this
sense refers to a speaker’s attempt to avoid having other people lose respect for you, as
well as being “the public self-image that every member wants to claim for himself”
(Wardhaugh, 2010: p. 292). Furthermore, hedges could be used to express ambiguity or
“to avoid giving a promise or direct answer” (Webster’s Dictionary). They are used
unconsciously in everyday speech, whereas for example in political debates, they could
be used consciously to diminish or empower someone or something. Coates defines
hedges as “… linguistic forms such as I think, I’m sure, you know, sort of and perhaps
which express the speaker’s certainty or uncertainty about the proposition under
discussion” (Coates, 2004: p. 88). By diminishing the force of an utterance, one also
diminishes the risk of offending the other speaker by coming on to strongly, and hedges
are devices used to do this.
Tag questions are expressions such as isn’t it? which seek confirmation
about the utterance from the addressee. Sometimes tag questions and hedges can
function very similar, as both can, for instance, seek confirmation. According to Holmes
it is not that simple to categorize expressions such as sort of and isn’t it? as either
hedges or tag questions (1990). The reason for this is that they may function as either
one depending on the situation and the relation between the speakers in question, as
well as speaker’s attitude and certainty about what they’re saying. Holmes proposes two
functions of hedges, the first one is the epistemic modal function, which expresses the
speaker’s approximation of something and uncertainty. Such hedges could be about and
fairly. The second function is the affective function, which is used by a speaker to
reduce social distances between the speakers and to create a more casual conversation.

6
Furthermore it signals politeness and a positive attitude towards the other speaker. Such
signals could be tags like you know or isn’t it? which are used when the speaker expects
a certain answer or just seeks confirmation about the utterance. Holmes lets the
following examples illustrate how sort of could reflect both an epistemic function
(Example 1) as well as an affective function (Example 2):

(1) Context: male student to male interviewer in description task.

he’s got a sort of a / skirty thing on I suppose you’d call it (Holmes, 1990:
p. 198).
(2) Context: one neighbour to another.
do you think I could sort of come and watch a programme on your TV
tonight/ it’s only short// (ibid, p. 198).

In other words, Holmes claims hedges are multifunctional and signal certainty as well
as uncertainty depending on intonation as well as the context in which they occur.
Falling intonation could indicate certainty about a proposal, while rising intonation
could signal uncertainty.
Prince, Brosk and Frader propose that there are two different sub-
categories of hedges, namely approximators and shields (1982: p. 85, cited in Reppen,
Fitzmaurice and Biber 2002: p. 26). Approximators are used to create fuzziness within
the utterance, for instance “He kind of screamed at her”. Shields are used to empower
the speaker’s commitment to the truth of an utterance, as in “I think he screamed at her.”
What Holmes defines as epistemic modals are similar in function to approximators in
the sense that they express the speaker’s uncertainty about the utterance. Shields,
however, do not have the same qualities as the affective function Holmes describes, as
they are used to empower the speaker rather than reduce social distances. Epistemic
modals are speaker-oriented and the affectives are addressee-oriented. Prince, Brosk and
Frader claim men use more approximators and women more shields as their
communication goals are different. Women use hedges to create a relationship with the
other speaker, while men use them to signal uncertainty. For instance, women could
hedge to show respect to the hearer by not being too forward about the utterance, while
men sometimes hedge because they feel inferior to the other.
A third definition of hedges has been provided by Gunnarson, who claims
they are forms of euphemisms: “To deceive someone or to avoid saying something
straight out, euphemisms may be used – one states something in a gentler way. It could

7
be a question of using metaphors, loan words, abstract words or abbreviations”
(Gunnarson, 2003: p. 103). In other words, Gunnarson claims hedges are used mainly to
save the other speaker’s face and avoid disagreeing by mitigating the force of an
utterance. This agrees with what Wardhaug claims about face-saving, but counters
Holmes’ of hedges as multifunctional. In other words, Gunnarson claims hedges could
be used to create confusion between speakers, but one could also create confusion by
using metaphors or loan words apart from hedges.

2.2 The Function of Hedges

The most famous definition of the function of hedges is probably the one put forth by
Robin Lakoff, where she claims hedges are supposed “to make things fuzzier” (Lakoff,
1973: p. 195). Lakoff’s claim about hedges and euphemisms in her study agrees with
Gunnarson’s claim that they are used to lessen the bad connotation of something.
Furthermore, Lakoff regarded hedges as a sign of a weak language since they may
signal the speaker’s lack of opinion on the matter, and claims this is “one [view] often
levelled at women” (ibid, p. 55). However, she cannot provide evidence to support this
claim.
Lakoff’s theory about hedges’ association with women has been accepted
by some researchers and questioned by others. Holmes’ analysis based on an equal
amount of female and male speech shows results which disagree with Lakoff’s
however; only in 33 of 89 cases in which women used hedges did they signal
uncertainty, whereas men expressed uncertainty 50 times of a total 87 (Coates, 2004: p.
89). Uncertainty was distinguished from certainty by means of falling and rising
intonation here. Furthermore, Holmes claims the same hedge may belong in different
categories depending on the situation in which it occurs. She illustrates this with the
following examples, where I think signals uncertainty in (3) and certainty in (4) due to
intonation:
(3) Context: child in classroom discussion.
it’s got some writing on it I think//
(4) Context: high status interviewee on TV.
I think that’s absolutely right// (Holmes, 1990: p 187).

In (3), I think is pronounced with a fall-rise intonation which signals

uncertainty, as Lakoff claims. However, in (4) I think is placed in initial position as well
as receiving level stress, “both [of which are] means of expressing emphasis and
8
confidence”, furthermore I think is pronounced with a falling intonation in (4) (ibid, p.
187). A hedge is categorized not only depending on the word itself, but on intonation
and syntactic position as well. This shows that a hedge placed in initial position signals
certainty more often than a hedge placed in final position. In other words, I think in (4)
should not be classified as a hedge, but rather seen as emphasizing the statement.
According to Holmes, hedges are multifunctional and used when the speaker wants to
convey uncertainty as well as certainty, depending on the situation and on what degree
of power the speaker has in relation to the other speakers. If the speaker in question
feels inferior to the addressee, frequent use of hedges is natural in order to avoid conflict
and respect the other speaker’s face, whereas if the speaker feels superior the use of
hedges is reduced. This distinction is vital to bear in mind while analyzing political
speech, as classifying I think incorrectly may lead to very inaccurate results.

2.3 Hedges in Other Domains

Hedges have been studied in various contexts apart from a gender perspective, one of
them being frequency and function in scientific research articles. Hyland found in his
study that:“…scientific writers make substantially more use of devices in the
speculative judgment category” (Hyland, 1998: p.126). In this category, words such as
suggest, propose and predict were included, however, indicate was the most frequently
used speculative hedging verb with a usage of 10.8 per 10,000 words. The purpose of
these hedges in the articles was to “gradually move the reader from the known through
the less certain, and finally to the more speculative discussion” (ibid, p. 191). Apart
from speculative hedging verbs in this article, evidential sensory verbs were taken into
account, i.e. verbs which indicate possibility. In comparison for evidential sensory
hedging verbs, appear was the most frequently used one, with 4.0 words per 10,000
words. It seems as though the purpose of hedging in this domain is to persuade the
reader of the writer’s claims and the accuracy of these. Previous research needs to be
taken into account, and the goal is to persuade in a way which is acceptable to the
academic community.
Furthermore, hedges are relevant in relation to the Gricean Maxims and
what Grice calls the cooperative principle (1995). The Cooperative Principle requires
speakers to adhere to four maxims in conversation: quantity, quality, relation and
manner. The maxim of quality says to not speak if you lack evidence or if you believe
something to be false, and as hedges could be used to express ambiguity and uncertainty
they can be seen to flout this maxim. According to Chilton the four maxims put forth by
9
Grice are presupposed by people, and by extension, the Cooperative Principle itself is
presupposed (2004). In other words, Chilton claims the Cooperative Principle is not
necessarily true. Chilton claims presuppositions are used in political speech as face-
saving devices, and that it takes courage to flout the Cooperative Principle: “It takes
effort to retrieve, formulate and challenge a presupposition – the effort being both
cognitive, and, since a face-threatening act is involved, also social. Speakers will
therefore have the option of using presuppositions strategically to avoid [being]
challenge[d] or reject[ed]” (Chilton, 2004: p. 64). In other words, adhering to the
Cooperative Principle in political speech is in favor of the speakers, as the risk of
challenge and rejection from the addressee decreases.
However, other researchers such as Blommaert disagree, and claim the
maxims can only be adhered to in certain situations, as “Co-operative[ness] is a variable
in dialogue, not a rule” (Blommaert, 2005: p. 44). Blommaert disagrees with Grice in
that speakers cooperate linguistically at all times, claiming one speaker may recognize
the maxims while the other does not, as in the following example:

Example (5)
Susan: Nobody cares for me… Nobody loves me… The whole world
hates me!
Rick: That is not true dear, the whole world does not hate you. Some
people do not even know you.

In this example, Susan flouts the maxim of quality as she is not telling the
truth, she cannot possibly know for sure that the whole world hates her. In other words,
all speakers have to adhere to the Cooperative Principle in order for it to exist. Based on
this claim, one could argue that the Cooperative Principle is often useless in political
discourse, as debates are often based on two or more speakers not cooperating
linguistically. A similar situation to Example (5) could present itself in a political
discussion, since the speakers flout maxims when they generalize or presuppose. It is
common in political debates that one of the speakers presupposes something while the
other does not. Research carried out by Holtgraves and Lasky on political speech
showed that “A speaker who uses powerless language will be perceived as less assertive
[or] competent… than a speaker who uses powerful language” (1999: p. 196, cited in
Hargie, 2004: p. 349). In conclusion, the Cooperative Principle should not be able to
function in political debates, as Gricean Maxims would be flouted due to the fact that

10
speakers in debate sometimes tend to mislead their opponent by obscuring information.
Furthermore, language expressing uncertainty makes the speaker in a debate look weak
and tentative.
Evidently, non-native speakers tend to use a smaller amount of hedges
than native speakers, especially in informal encounters regarding personal matters. Data
collected by Markkanen et al shows this (1997). The lack of hedges could lead to that
non-native speakers risk being perceived as unyielding and careless about the
addressee’s face. In comparison with speaker number two below, number one contains
considerably more hedges:

Example (6)
S1 but I mean I hate to be very sort of capitalist and and whatever but I do
believe up to a certain point that if I mean I don’t know…
S2 wish I could fight against becoming a teacher as long as I can (Markkanen,
1997: p. 196).

Speaker one is a native speaker of English, while number two is a

non-native speaker of English. This example illustrates how the non-native speaker
offers a very forward opinion by not using any hedges at all. This shows that
information about the speaker’s sociological background is relevant in linguistic studies,
since it could explain why some speakers hedge more than others. Another reason for
why hedges may or may not be used is of course topic. “When sensitive topics are under
discussion, then hedges become a valuable resource for speakers, because they mitigate
the force of what is said and thus protect both speaker’s and hearer’s face” (Coates,
2004: p. 90). In the next section hedges’ connection to the Androcentric rule is
addressed.

2.4 Hedges and the Androcentric Rule

The Androcentric Rule described earlier is the source of norms of language, a norm
which women are seen to deviate from since the norm is based on men’s use of
language. However, research carried out on problem-solving interaction by Bradac,
Mulac and Thompson shows that men do use features associated with women and vice
versa: “Women used an average of 1.1 (per 100 words) intensifiers, whereas the
comparable value for men is 0.8… By contrast, women used 0.3 hedges per 100 words,
whereas the value for men is 0.5, again a significant difference” (1995: pp. 104-105).

11
Hedges were distinguished by words in their study, words such as kind of and pretty
were classified as hedges at all times, regardless of intonation or syntactic position.
Powerlessness in their study is demonstrated by hedges which decrease the force of an
utterance and also the speaker’s reluctance to commit to an utterance. Regarding
context, women in this study used more hedges to signal certainty and power in mixed-
sex groups than men, furthermore men used significantly more hedges expressing
uncertainty. In other words, hedges are found to be bound more to powerlessness than
women’s use of language in this study.
O’Barr and Atkins claim that powerless speech is confused with women’s
speech in the sense that powerless language contains some of the features used by
women. The reason for this confusion, according to O’Barr and Atkins, is that women
tend to speak in a more powerless way than men due to the fact that women often hold a
weaker position in society. Furthermore, their study on male-female language patterns
in trial courtrooms showed that the witnesses, three men and three women, all used
features of women’s language more or less. For instance, all men used a larger number
of intensifiers than any of the women (24 compared to 16), as well as a larger number of
hedges, (7 compared to 24) (ibid: p. 100). In their study they did not include the
different functions of hedges proposed by Holmes, yet they included information such
as age, occupation, gender and relational status about the six witnesses. O’Barr and
Atkins drew the conclusion that sex is insufficient to explain linguistic features, and that
social factors had to be taken into account as well.

2.5 Hedges in Political Speech

Okulska analyzed Bush’s style of answering questions proposed by reporters, and the
findings were different from what she expected. Instead of using hedges to avoid
responsibility, hedges were found to be used in a non-typical, non-hedging way:

(7) I appreciate that kind of commitment, Mr. Prime Minister.

(8) Newly elected President, Tony Blair came over, and he reached out; he was gracious
– was able to converse in a way that – where our shared interests were the most
important aspect of the relationship (Okulska, 2010: p. 207).

Okulska adds, “…there were no cases of hedged performatives, very few cases of
hedged felicity conditions or hedged Gricean Maxims, and no cases of I suggest that, I
wonder if, or I supposed that, all of which might have been anticipated” (ibid: p. 207).
12
Hedges in Bush’s speech occurred in a non-hedging, neutral way, which means they had
no impact on the utterance. The typical, expected pattern would have been hedges
which diminish the force of the utterance or reduce the speaker’s commitment to the
truth. In other words, the expected rhetorical pattern, which is hedging because of lack
of evidence or commitment, did not occur in Bush’s speech. Instead, hedges occurred in
a non-hedging way, and show Holmes’ claim about how hedges are multifunctional
depending on the context. It is possible that hedges used to signal uncertainty and lack
of commitment were not necessary in this situation since Bush did not feel the need to
defend his claims. Neither the epistemic function nor affective function of hedges
proposed by Holmes occurred in Bush’s speech. One might expect that Bush employed
the affective function of hedges as they reduce social distances, something rhetoricians
tend to use to make themselves one with the crowd.
Wardhaugh provides a different angle on hedging and their occurrence; he
claims that hedges belong mainly in unplanned and informal speech which possesses
certain characteristics, such as “…repetitions, simple active sentences… use of
deictics… It may also be filled with equivocations (or hedges), i.e., words and
expressions such as well, like, maybe, but, sort of, you know, I guess etc.” (Wardhaugh,
2010: pp. 314-315). In other words, according to Wardhaugh, hedges should not occur
in formal political debates such as in district councils, as the majority of speech is
planned in such a situation.
Another reason for why hedges did not occur as frequently as expected in
Bush’s speeches could be explained by Thomson (2006, cited in Precht 2008: p. 93). In
his study he found that men and women tend to use hedges quite similarly in contexts
regarding gender-neutral matters in public discussions, but in gender-stereotypical
matters women used a larger amount of hedges.
Precht carried out a study which examined men’s and women’s use of
stance, that is words associated with emotion, attitude and commitment (2008). Precht
recorded informal conversations in social contexts such as work environments, and
found that men had a significantly higher frequency for five hedges: about, basically,
like + adj/noun, something like and pretty, while women had a higher frequency for
almost, maybe and well. In this study, hedges had a frequency of 0.0008, which is 0.97
% of the corpus. This means that men and women use hedges very similarly, and only
in 8 of 27 cases did the occurrence of hedges show strong differences (Precht, 2008: p.
98-99). Compared to the study carried out by Okulska, hedges occurred much more

13
frequently in informal conversations than in formal appearances. As stated by Thomson,
this could depend on the situation and what matter is being discussed (2006).
As anticipated by Okulska, hedges used to relieve the speaker of
responsibility are expected in political debates. Hickey found results in his study of
political texts which agreed with Okulska’s, in that no anticipated hedges occurred
(1998). What Hickey found was what he refers to as evidentiality-hedges, that is, hedges
used by the politician to present a positive, trustworthy image of himself, such as as far
as I am concerned or I do believe. The purpose of such hedges is to employ a
“…’positive’ strategy of inspiring confidence by sounding fully committed to the truth
of their claims” (Hickey, 1998: p. 188). Hickey gives the following example to illustrate
his claim:
Example (9) And I honestly believe that if we had not changed,… we

could not change the country. (ibid: p. 187).

The example is taken from a speech given by Tony Blair, where the
function of the hedge I honestly believe is to reinforce the speaker and give credibility to
the statement. In other words, the main function of the hedge is to create an image of
Blair as a reliable person and in extension give him power in the form of people’s trust
and respect.
The different strands of research discussed are the function of hedges,
hedges in other domains but political discourse, hedges and the androcentric rule and
hedges in political speech. The next section will present the methodology, the data, the
annotation procedure and the validity and reliability for this study.

14
3. Methodology
The aim of this essay is to analyze men and women’ use of hedges in political
discourse. This will be done by analyzing hedges appearing in the minutes of a district
council meeting by quantifying data, in order to determine what is indicated by the
hedges and how the various types of hedges differ based on gender. The various types
of hedges have been annotated into different categories based on their syntactic
position, content and communicational function as well as context. This study is
quantitative, seeing as the attributes, (the hedges), are classified into a taxonomy (which
will be discussed in the next subsection). Then the hedges are counted in order to learn
about the distribution of them by communicative function (Bailey, 1994). What is
counted are the frequencies of hedges of different types in the taxonomy which will be
presented in the next subsection. An annotation procedure will be presented which
outlines how the different types of hedges can be distinguished. First, the data will be
presented, followed by a presentation of the annotation procedure. Finally, a section
discussing validity and reliability of the study will conclude this chapter.

3.1 Data
The data selected for this study is a transcript from a meeting at the Tampa City Council
in Florida. This particular transcript was chosen since it alone was long enough to
provide the amount of hedges needed for this study. Furthermore, there was a wider
selection of American than British transcripts available online. Analyzing one whole
meeting seemed more advantageous than analyzing only parts of several meetings, since
this would provide an overview of one particular meeting, and it is unclear how the
distribution of hedges relates to when they occur in a council meeting. In this meeting
there are seven council members, four men and three women. Apart from these seven,
20 more people attending this public hearing hold the floor, 18 men and two women.
During the meeting, there are 12 topics which are discussed.

3.2 Annotation Procedure

A taxonomy of hedges was constructed to capture the communicational content and the
syntactic features of the hedges. In order to do this, an annotation procedure was
developed, which allowed quantifying the hedges in the data in order to examine what
communication role they play. The taxonomy is based on the different functions of
hedges described in the literature. The goal of this study is to investigate whether
women use weak language as Lakoff claimed (1973). In the end, the hedges will lend
15
support for or against Lakoff’s claim about women’s weak speech habits. The findings
from the literature motivated four main categories for the analysis, each of which
correspond to a branch of the taxonomy, each category consists of a range of types.
Multiple categories can hold simultaneously for a hedge, but within a category, the
types are mutually exclusive.
The first such category is the epistemic modal function. This category is
based on Holmes’ study where she categorized hedges and tag questions into two broad
groups, one of them being the epistemic modal function (1990). This function can signal
imprecision as well as certainty about a proposal, for instance “I think that is absolutely
right” or “It has got some writing on it I think.” Here, the same hedge signals certainty
in sentence number one, but imprecision in sentence number two.
The second category is the affective function: hedges which apply here are
sort of and kind of, among others. This is the second broad category created by Holmes
apart from the epistemic modal function (1990). The affective function concerns the
relationship between the speakers, as opposed to the epistemic modal function which
concerns the force of the utterances. The affective function looks at face-saving and
considers politeness between the speakers, for example: “I was sort of late”.
The third category motivated is hedges which seek confirmation: this is
another category proposed by Holmes (1990). She discovered in her study that you
know, for instance, is used to receive feedback from the hearer in the sense that the
hearer understands what the speaker is saying. The same hedge could also be used for
other purposes, such as letting the addressee know you are expecting a certain response
if you know is placed in initial position.
The last category is termed shields; This category is proposed by Prince,
Brosk and Frader who argued that hedges can be used not only to signal lack of
commitment, but also full commitment to a claim, as in: ”I think you are absolutely
wrong” (1982). Like in the epistemic modal function, the same hedge in this category
can be multifunctional depending on its syntactic position.
Types of hedges are grouped according to which category they belong to,
and some are mutually exclusive within a category, but multiple types belonging to
different categories can hold simultaneously for a given hedge. There are nine types into
which the hedges can be categorized.
For the epistemic modal function category, which can express uncertainty, the first
type indicates the speaker’s uncertainty about something, for example: “She seemed
sort of unfocused.” In this case, as for all the other types, the hedges given are only

16
examples and not the only ones for which these types can apply, similar expressions to
the ones mentioned here can occur as well for all types. Type 1.5 also expresses
uncertainty but also mitigates the force of the utterance, but only when I think or similar
expressions appear parenthetically in the utterance, such as: “There should be some
other paintings, I think, not just these.” Type two handles hedges which express the
speaker’s approximation of something. About, fairly and similar words belong to this
type, for example: “It happened about a week ago.” Type three involves whole phrases
expressing uncertainty, unlike type one which do not contain phrases, for example:” I
guess I’ll take the train.”
For the affective function category, which reduces social distances
between the speakers, the fourth type carries an affective function, in this case face-
saving. It saves negative face by mitigating the force of the utterance by using hedges
such as sort of and kind of, as in: “Could I sort of ask you for a favor?” The risk of
offending the addressee by sounding too forward is reduced along with the force of the
utterance by inserting sort of. Type five also saves the hearer’s negative face, but only
when the hedge you know is placed in middle position of the utterance; “I thought it
was, you know, quite boring.” You know mitigates the negative tone of the utterance and
means the speakers share an understanding. Type six is another face-saver, but for the
speaker, not the hearer. Hedges such as kind of and similar expressions are used to avoid
having other people lose respect for the speaker, as in: “I sort of lost him.” Sort of
mitigates the negative tone of the utterance. Type 6.5 is the second type of face-saver
for the speaker, but unlike type 6, it must involve expressions, such as I believe or
similar. This is to avoid giving full commitment to the utterance by showing the speaker
is not sure about it, for example: “I believe I read that somewhere.”
The third category is confirmation seeking. Certain hedges are used to
signal certainty, as in type seven: “You know she is going to find out.” Here the speaker
indicates that he or she is expecting a certain response from the hearer. Type eight is
similar to type seven in that both indicate certainty about the expected response.
However in type eight, the speaker also seeks confirmation about the utterance, as in:
“She is really nice, you know.”
The fourth category, shields, signals the speaker’s full commitment to the
utterance. Type nine is the only type in this category, and I think or similar expressions
belong here and they must be placed in initial position, for example: “I think that is what
we have to do.”

17
 Finally, there are multiple hedges which can co-occur, as long as they
belong to different categories. Such an example is type 1.5 in combination with type
6.5: “You had a representative here, I believe you said, and now you have revoked the
power of attorney.” In this utterance, I believe signals the speaker’s uncertainty about
the truth of the claim. At the same time the risk of other people losing respect for the
speaker decreases since the speaker is not fully committed to his or her own claim.
Table 1 below illustrates which categories express uncertainty and which
indicate certainty:

Table 1. Distribution of Uncertainty and Certainty among the Categories

Epistemic Affective Confirmation Shields
Modal Function Function Seeking
Uncertainty Yes Yes
Certainty Yes Yes

Among the four categories there are two which signal uncertainty, the epistemic modal
function and the affective function. The remaining two, hedges which seek confirmation
and shields, indicate certainty. The epistemic modal function expresses uncertainty by
using various words or phrases, whereas the affective function does the same but in
combination with face-saving. Hedges which seek confirmation signal certainty about
the expected response from the addressee by using you know, similar to what shields do
with I think to empower the utterance.
An outline of the taxonomy will now be presented, which illustrates how the
categories and types fit together.
 Category A: Epistemic modal function: Signals the speaker’s imprecision or
uncertainty about the utterance.
Type 1: Indicates the speaker’s uncertainty by using sort of, kind of or
other similar types of hedges.
1. If the hedge is placed in any syntactic position
2. And expresses the speaker’s lack of commitment,
3. Then if it signals uncertainty it belongs to the epistemic modal function,
for example: “He has got a sort of bad attitude.” “He kind of screamed at
her.”
Type 1.5: Indicates the speaker’s uncertainty by using I think.

18
 1. If I think is placed anywhere and in parenthetical position of the
utterance,
2. Or placed in final position of the utterance and does not follow a
relative pronoun,
3. Or appears before a relative clause,
3. Signaling the speaker’s doubt or imprecision,
4. Then if it indicates uncertainty it belongs to the epistemic modal
function: “But to tell her in person, I think, is not the right thing to do.”
Type 2: Indicates the speaker’s approximation about the utterance.
1. If about or fairly is a part of an estimation or approximation
2. Occurring in any syntactic position of the sentence,
3. Then if it indicates imprecision it belongs to the epistemic modal
function, for instance: “There will be about one hundred guests.” “I am
fairly certain that is what he said.”
Type 3: Signals the speaker’s uncertainty by using I’m not sure and I
guess.
1. If the hedge is a main clause, which occurs parenthetically or at the end
of the sentence,
2. And indicates the speaker’s uncertainty,
3. Then it belongs to type three: “I’m not sure whether or not I agree with
you.”
 Category B: Affective function: A solidarity signal to reduce social distances
to the addressee, and to create a more casual conversation.
Type 4: Saves negative face by indicating uncertainty using sort of and
kind of.
1. If the hedge is a part of an interrogative utterance from the speaker
2. And occurs before the non-modal verb,
3. Then if it indicates uncertainty it saves negative face and belongs to the
affective function, such as: “Maybe I could sort of ask you out
sometime?”
Type 5: Signals the speaker’s uncertainty about the response to the
utterance from the addressee by saving the hearer’s negative face.
1. If you know is placed in middle position of the sentence
2. And is part of an opinion,

19
 3. Then if it indicates uncertainty it saves negative face and belongs to the
affective function: “The music was, you know, not very good.”
Type 6: Saves the speaker’s own face by using kind of and sort of.
1. If the hedge is a part of an opinion
2. Occurring in any syntactic position of the sentence,
3. And includes pronouns in nominative position of the clause,
4. Then if it indicates the speaker’s lack of commitment to the utterance it
saves the speaker’s own face and belongs to type six: “We were kind of
late.”
Type 6.5: Saves the speaker’s own face by using I believe, I imagine or
similar expressions.
1. If the hedge is part of an opinion
2. Occurring in any syntactic position of the sentence,
3. And includes 1st person singular in nominative position of the clause,
4. Then if it indicates uncertainty it saves the speaker’s own face and
belongs to type 6.5: “I believe there was a 30-day time frame to file a
formal appeal outside of that.”
 Category C: Hedges which seek confirmation: Seeks confirmation from
the addressee. These hedges differ from tag questions in the sense that
they cannot be placed in final position and are not part of a question.
 Type 7: Signals the speaker’s certainty about the expected response to the
utterance from the addressee.
1. If you know is part of the main clause
2. Then if it signals certainty it belongs to the category where the speaker
seeks confirmation: “You know he is not who he says he is.”
Type 8: A certain answer is expected from the addressee by using you
know.
1. If you know? is placed in parenthetical position,
2. Then if it indicates certainty you know belongs to the hedges which seek
confirmation: “He said that last week too, you know.”
 Category D: Shields: Empower the speaker’s commitment about the utterance.
Type 9: Signals the speaker’s full commitment to the utterance.
1. If I think is placed in initial position
2. And is part of an opinion,

20
 3. Then if it signals certainty it belongs to shields: “I think that is
absolutely right.”

3.3. Validity and Reliability

An analysis into the validity and reliability of this study will now be looked into.
Internal validity refers to the extent to which a conclusion drawn from a study is correct,
and also how likely or unlikely it is that errors are made in for instance the method used
when analyzing the data. For example, if there is lack of accuracy in the analysis
method, the chance of internal validity decreases since errors are likely to have been
made. External validity refers to which extent the results of the study can be applied to
other situations and people. For instance, if a study is conducted with only a sample of
people in a particular situation, the external validity is usually low since one cannot be
sure that the analysis of the sample is enough to apply to other groups.
There are two types of reliability which need to be considered. Internal
reliability refers to how consistent the annotation is. External reliability refers to the
extent to which the method can be applied to other studies. In this case it refers to
whether the annotation method can accurately be applied to other data. Even though the
annotation procedure was checked with the supervisor for this essay, there are some
factors which may lead to less internal reliability. One of these factors is the lack of
access to intonation in the meeting transcript, as falling or rising intonation often
indicates confirmation and requests respectively (Holmes, 1990). This affects the
internal validity in the sense that wrong conclusions can have been made about the
utterances in the meeting minutes. No test of inter-annotator reliability was performed,
however, the annotations were checked with the supervisor for this essay. This was a
modest attempt at improving internal reliability, in order to ascertain internal reliability
two new annotators would have had to annotate the data and had their scores compared
to see if they agreed more than would have been expected by chance.
Information about the power and social background of each council
member was too insufficient to take into account during the analysis, otherwise the
internal validity would have been more accurate. According to Lakoff women could be
expected to hedge more since they have less power than men, and accounting for power
gets around this and looks at gender without this added factor of imbalances between
speakers. Such information could also have explained other members’ use of hedges,
since if they are in a position of less power than their addressees the use of hedges could
increase (Coates, 2004).
21
 Furthermore, the use of descriptive statistics rather than inferential
statistics in this study also leads to certain consequences for external validity. For
instance the results might be skewed by chance and no universal truth about hedging in
political discourse can be achieved, since the sample of data is not based on probability.
Also, the lack of an inter-annotator test reduces the internal validity. Furthermore, the
data in this study was collected from one single meeting, which is not enough to
generalize the findings.
The next section will introduce and present the results of the analyzed
data.

22
4. Results
In this section, the results of the study will be presented. The annotation procedure in
the preceding section was applied to the council meeting minutes to identify and
classify the hedges found in the transcription of the minutes. The results quantify the
distribution of hedges in the taxonomy. An introduction to the numerical results of the
various types of hedges will be presented, in order to illustrate the distribution of hedges
in the political meeting of the Tampa City Council. Following this, an analysis of the
results will be performed, discussing the implications of the results for the hypothesis.
This essay hypothesizes that Lakoff’s claim about women’s tentative speech habits is
incorrect regarding political discourse.
There were altogether 76 instances of hedges in the minutes, 45 of which
were issued by men and 31 by women. There were altogether five women and 22 men
who held the floor during the meeting, including three women and four men in the
council. This gender imbalance points to the fact that this is still an area dominated by
men, which in turn means that men hold the majority of power. However, the fact that
half of the council members were women indicates a roughly even gender balance there.
The total number of words spoken in the meeting was 19, 821, of which 14, 774 were
by men and 5047 by women. Men spoke for 287 turns compared to women who had
only 110 turns. Table 2 indicates the occurrence of hedges by number and percentage in
relation to the total amount of words spoken by each sex. The first data column
indicates the total number of words spoken by gender, and the second shows the total
number of hedges spoken by gender. The third column shows the normalized percent,
which is calculated from the normalized frequency. The normalized frequency was
calculated in order to scale for differences in the amounts that men and women talk, as
this would otherwise skew the data. Normalized frequency is calculated by dividing the
total number of hedges spoken by men by the total amount of words spoken by men,
and likewise for women. Not accounting for this would skew the data in the way that
the amount of turns and speech time between men and women, which was very uneven
because men had a lot more speech time than women, and not accounting for this would
give inaccurate results. Normalized percent is calculated by multiplying the normalized
frequency value by 100.

23
Table 2. Normalized Frequency As a Percentage for Hedges Over Number of
Words
Total Total Normalized
Number Number of Percentage
of Words Hedges
Men
14, 174 45 0.32

Women 5647 31 0.8

To judge from Table 2, men did hedge more than women in this meeting based on sheer
numbers, this is due to the fact that they held the floor more and had more turns to speak
than women did. However, when you consider normalized percent men hedged much
less than women, the scaled proportion for this is reflected in the normalized scores.
The results could possibly be different with different participants and different power
relations, for example if women occupy higher positions in the council. Another
interesting aspect of the distribution of hedges in Table 2 is that the amount of hedges is
relatively low for both men and women in regard to the amount of words spoken by
each sex. Coates’ explanation for such a situation is topic sensitivity, that hedges
become a valuable resource for the speaker when a sensitive topic is discussed, in that
they mitigate the force of the utterance (2004). Seeing as the number of hedges was low
in this meeting, a contributing factor could be the lack of sensitive topics during this
particular meeting. Thomson offers a similar explanation to Coates’; in his study he
found that women and men tend to hedge similarly in gender-neutral matters (2006).
Table 3 below shows the normalized percentage values and counts of
hedges by gender in each of the categories, in other words how frequently hedges
occurred in each category by gender.

24
Table 3. Normalized Percent of Hedges by Gender for Each Category, Counts
(Left); Normalized Percent (Right)
Epistemic Affective Confirmation Shields
Modal Function Seeking
Function
Men 12; 0.085 12; 0.085 10; 0.07 11; 0.78

Women 11; 0.195 4; 0.07 6; 0.11 10; 0.18

Table 3 illustrates that men hedged more than women in categories affective function
and shields. The results support what Bradac, Mulac and Thompson found in their
study, namely that women do not always hedge to signal uncertainty (1995).
Confirmation Seeking, which signals certainty about the response from the addressee,
illustrates women’s more frequent use of this category compared to men’s. Coates
discusses how participants in interaction sometime act as facilitators, that is one or
several people take on the role of ensuring that the conversation goes smoothly.
According to Coates, women tend to take on this role more often than men, which could
explain why women dominate the category of confirmation seeking (Coates, 2004).
However, Prince, Brosk and Frader’s results do not agree with the ones in
this study, as they found that women used more shields than men in conversation
(1982). In this study, women used relatively few hedges in relation to the total amount
of words issued by women, though proportionally shields are used frequently, second to
the epistemic modal function. Prince, Brosk and Frader also implied that women and
men’s communication goals are different, namely that women strive to create a positive
relationship with the other speaker through hedging, while men hedged simply to signal
uncertainty about a proposal. There seem to be no such implications in this study, as
men actually used more hedges in the affective category than women did. However, this
might have been due to the choice of domain, if the domain had been another than
council meetings the results could have been different. The most frequently occurring
hedges in men’s speech belonged to the two categories which signal uncertainty, the
epistemic modal function and the affective function. There were also two categories
where the normalized percentage indicated that men hedged more than women did,
namely the affective function and shields. Again, the results of this study disagree with
the study carried out by Prince, Brosk and Frader, who found that women used more
shields than men (1982). The results do not match the claim by Coates either, about how
25
women usually dominate the affective function by acting as facilitators. This counters
Lakoff’s claim that men do not use any hedging devices at all in their speech. However,
the epistemic modal function was also the most frequently occurring category for
women, which signaled uncertainty more for women than for men. This agrees with
Holmes’ results where she too found that women used hedges in this category to a much
larger extent than men (1990).
There were also some instances where two hedge types belonging to
different categories appeared in combination during the meeting, that is, hedges which
belong to multiple categories. Figure X below illustrates this. Table 4 illustrates the
frequency of the occurring combinations and percentage by gender.

Table 4. The Frequency of Combined Hedge Types for Men and Women, Counts
(Left); Normalized Percent (Right)
Epistemic Modal Epistemic Modal Affective Affective Function +
Function + Affective Function + Function + Epistemic Modal
Function, Types 1.5 + Affective Confirmation Function, Types 6 + 3
6.5 Function, Types Seeking,
1.5 + 6 Types 5+8

Men 4; 0.028 1; 0.007 1; 0.007 1; 0.007

Women 2; 0.35 1; 0.18

26
Figure X. Normalized Percent of Co-Occurring Hedges by Gender
0,4
0,35
0,3
0,25
0,2
0,15
Men
0,1
Women
0,05
0
Epistemic Epistemic Affective Affective
Modal Modal Function + Function +
Function + Function + Confirmation Epistemic
Affective Affective Seeking, Types Modal
Function, Function, 5+8 Function,
Types 1.5 + 6.5 Types 1.5 + 6 Types 6 + 3

Figure X and Table 4 illustrate how types 1.5 and 6.5 from categories epistemic modal
function and affective function, uncertainty in combination with face-saving, are the
most frequently appearing combination for both men and women. It is interesting that
men felt the need to save their own face as much as they did, seeing as men hold the
majority of spots in the council. This might indicate that men felt less powerful than
women even though men were a majority. Bradac, Mulac Thompson found in their
study that hedges seemed to be bound more to powerlessness than gender, which could
explain men’s face-saving use of hedges (1995). The information about who holds the
most powerful position in the council was left out of the website of the Tampa City
Council. If it had been included the results in this study could have been further
analyzed. Furthermore, it is notable that men used types five and eight in combination,
which communicate an affective function with confirmation seeking, more than women
in this study, since according to Coates women usually hedge using the affective
function more frequently than men. Again, this might be because men felt powerless, as
explained by Bradac, Mulac and Thompson (1995).
The overall result from this study shows that men hedged less than women
in the Tampa City Council meeting. Men had a normalized percentage value of 0.32
compared to women who had a value of 0.8. The hypothesis for this study was that
women do not hedge simply to signal uncertainty, and to investigate whether or not they
hedged more than men in political discourse. The results indicated that women do hedge
more than men in this particular meeting. However, the results disprove Lakoff’s claim
27
that men do not use hedging devices in their speech, and that women do not hedge
simply to indicate uncertainty.
In the next section the conclusions will be presented.

28
5. Conclusion
This section will conclude the results of the analysis of the Tampa City Council
meeting, as well as discuss how well the thesis of this paper has been answered. The
aim of this paper was to investigate whether or not women used more hedging devices
than men in political discourse, and to address Lakoff’s claim that women hedge to
signal uncertainty. The hypothesis was that women do not hedge more than men in
political discourse, which counters Lakoff’s claim. The data for this study was a
transcript from an American city council meeting, and the hedges occurring in the
meeting were annotated into categories of hedges according to a taxonomy of hedges
based on previous research on hedging discussed in the literature review. The taxonomy
constructed based on the communicative functions of hedges described in the literature
contained a number of categories, each of which consisted of some mutually-exclusive
subtypes. These categories consisted of the epistemic modal function, which
communicated the speaker’s imprecision or uncertainty about the utterance. The second
category was the affective function, which created a more casual conversation by, for
instance, face-saving and politeness. The third category, which was hedges which seek
confirmation, or confirmation seeking, signaled the speaker’s certainty about the
expected response from the addressee. The final category, shields, empowered the
speaker’s commitment to the utterance by phrases such as I think or similar.
The results from the analysis in the preceding section indicate that men
hedged less than women, since the total normalized percentage for men was 0.32%
whereas for women the total normalized percentage was 0.8%. While women hedged
more than men, this counters Lakoff’s claim that men do not use hedging devices at all
in their speech, and more specifically, that they do not signal uncertainty. In fact, the
results indicated that men had an equal number of hedges in both categories which
signal uncertainty, i.e. the epistemic modal function and the affective function. Women
had a lower amount of hedges than men in two of the categories, the affective function
and shields, but a higher amount in the epistemic modal function and hedges which seek
confirmation, indicating that women signaled uncertainty and wanted confirmation
more often than men.
Based on the results from the analysis, the question which formed the
investigation for this paper has been answered. In this study, men hedged less than
women. Women hedged more to signal uncertainty than men, however, not at all times.
This disproves Lakoff’s claim about women’s speech habits, as they did not always
hedge to signal uncertainty. This study illustrates an example of a domain where

29
Lakoff’s claims do not hold. However, more accurate results could have been achieved
by taking into account information about the level of power of the council members and
their social background. This information could have helped to shed further light on the
results in this study. Future work on power in political discourse should attempt to
investigate not only a sample of people but a population, and also to collect as much
sociological information about the participants as possible. In order to ascertain if it is
gender which causes women to hedge more than men, there need to be an equal amount
of men and women in equal positions of power participating in the study.

30
References
K., Bailey. 1994. Methods of Social Research. New York: The Free Press.
J., Blommaert. 2005. Discourse: A Critical Introduction. New York: Cambridge
University Press.
JJ., Bradac , A., Mulac & S.A., Thompson. 1995. Research On Language And Social
Interaction. Wales: Lawrence Erlbaurn Associates, Inc.
P. A., Chilton. 2004. Analysing Political Discourse: Theory and Practice. London:
Routledge.
J., Coates. 1986. Women, Men and Language. Great Britain: Pearson Education
Limited, 2004.
H.P., Grice. 1995. Logic and Conversation. New York: Academic Press.
H., Gunnarson. 2003. Vältalaren. Malmö: Almqvist & Wiksell.
O., Hargie. 2004. Skilled Interpersonal Communication. Great Britain: 4th edn:
Routledge.
L., Hickey. 1998. The Pragmatics of Translation. Great Britain: Cromwell Press Ltd.
J., Holmes. 1990. Hedges and Boosters in Men’s and Women’s Speech. Language and
Communication, 10 (3): 185-203. 10th edn. Great Britain: Pergamon Press
K., Hyland. 1998. Hedging In Scientific Research Articles. John Benjamins Publishing
Company.
R., Lakoff. 1973. Language and Woman’s Place. Cambridge University Press.
R., Markkanen. 1997. Hedging and Discourse: Approaches to the Analysis of a
Pragmatic Phenomenon. Germany: de Gruyter.
Merriam-Webster Dictionary. <[Link] [Accessed on 18
February 2013.]
W. M., O’Barr, B. K., Atkins. 1980. "Women's Language" Or "powerless Language"?
U., Okulska. 2010. Perspectives in Politics and Discourse. Amsterdam: John Benjamins
Publishing Company.
K., Precht. 2008. Ohio: Sex Similarities and Differences in Stance in Informal
American Conversation. Journal of Sociolinguistics, 89–111.
R., Reppen, S. M., Fitzmaurice, D., Biber. 2002. Using Corpora to Explore Linguistic
Variation. Amsterdam: John Benjamin Publishing Company.
R., Wardhaugh. 2010. An Introduction to Sociolinguistics. Great Britain: 6th edn: Wiley-
Blackwell.

31