A Corpus Assisted Comparative Analysis of Donald Trump’s and Hillary Clinton’s Speeches during the 2016 Presidential Election

1 
          Introduction 
Political language is something that concerns everyone and is the emblematic example 
of the use of language with a persuasive purpose. It is a reality to which we are constantly 
confronted, on which everyone has prejudices and that everyone questions. In 2016, the world 
experienced the unfolding of a historic election in the United States, an election that would take 
either the first female president into the White House, or a president who had no previous public 
office experience and who repeatedly broke canonic political norms. Donald Trump's ultimate 
victory over Hillary Clinton vanquished all popular perceptions and shocked the whole world. 
Generally regarded as the most striking election result in modern U.S. history, Trump's victory 
sent people seeking out for answers and explanations in all sectors of society. In an effort to 
better understand Trump's vocabulary and his attitude, the main characteristics of 
his rhetoric have been the subject of many discourse studies. Obviously, most of these 
inquiries take Trump as the main subject of investigation, assuming that his rhetoric reflects a 
deviation from existing norms and therefore worthy of special investigation. 
This inevitably pushes studies on his rivals to the periphery of academic consideration, 
thus failing to highlight their specific differences. This work aims to investigate, from the 
linguistic-argumentative point of view, the discursive production of both the candidates Donald 
Trump and Hillary Clinton in the context of the American Presidential Elections, accentuating 
their campaign themes, and exploring their underlying political beliefs. This objective will be 
pursued through the analysis of two distinct corpora, each one containing the speeches given 
by each candidate over the period of the campaign events, with a particular textual analysis 
software. Computational analysis’ tools, introduced by Sinclair and Partington, are combined 
with the tools of argumentative linguistics and of the Critical Discourse Analysis, in order to 
thoroughly investigate the communicative methods and the main differences of the subjects 
examined.
2 
I. Theoretical Frameworks: Theories and Concepts 
1.1 Introduction 
Typically, political speech is thoroughly prepared, rhetorically elaborate, and read 
from a written manuscript. As already mentioned in the introduction, this paper analyzes the 
transcripts of speeches by Trump and Clinton in the political and institutional context of the 
United States. In the composition of such discourses a particular use of language is made, 
which will be investigated thanks to the joint use of Corpus Linguistics and Political 
Discourse Analysis. The main reason behind the choice to carry out this kind of analysis 
comes from the fact that “a corpus of political speeches makes it possible to analyse the 
idiolect of specific politicians (although note that many politicians use speech writers), for 
example in terms of rhetorical style or the typical connotations of specific keywords” (Ädel, 
2010, p. 591). Moreover, another aim at the heart of this work is to grasp and identify 
elements of gender references, persuasive techniques and social inclusion and exclusion. 
In this chapter, the theories and authors who have contributed to deepening the 
knowledge necessary for the development of the present work will be illustrated. Furthermore, 
some key concepts in the linguistic-argumentative field, preparatory to the conduct of an 
autonomous linguistic analysis, will be introduced. 
 
1.2 Corpus Linguistics 
At the beginning of the text 'Patterns and Meanings', Alan Partington provides a 
definition of the concept of corpus, namely "a collection of texts assumed to be representative 
of a given language, dialect or other subset of language, to be used for linguistic analysis". The 
discipline that derives from it is corpus linguistics, that is "the study of language based on 
examples of ‘real life' language use" (McEnery & Wilson, 2001, p.1). This kind of approach 
has become popular since the advent of personal computers in the 1990s and has widely been 
accepted as a relevant way of analyzing language. The literature on corpus linguistics describes 
several important advantages of corpus linguistics as a methodological approach, especially 
when used to investigate political speech. First of all, the advantage of using big datasets "allow 
patterns to emerge providing evidence about language in use which was not previously 
available" according to John Sinclair (1991, p.4). Second of all, corpora allow researchers to 
"find differences that intuition alone cannot" (Tognini-Bonelli, 2001, p.55). Currently, 
computer corpora can store many millions of running words whose features can be 
analyzed using concordancing programs and software, and this surely is one of the main
3 
advantages of utilizing such collections of texts. There have of course been a great number of 
studies of language in politics carried out by linguists or political scientists, without any resort 
to corpus analysis. These have traditionally been focused on a limited number of texts. 
However, it is clear that in politics an individual text does not exist in isolation from other texts 
of the same institutional discourse type, whether it be a parliamentary speech, a campaign event 
or even a session of speeches. Single speeches are part of a much wider ongoing discourse. As 
John Sinclair, one of the founders of corpus linguistics commented: "the language looks rather 
different when you look at a lot of it at once" (Sinclair, 1991, p.100). As a result, large 
collections of tokens of a discourse type would seem to be a valid way of reflecting the nature 
of political discourses. The statistical techniques of corpus analysis enable indeed researchers 
to reveal patterns and features of language use that are not immediately evident to the language's 
users, teachers, or analysts of that language. 
As Norman Fairclough states, language is vital to the process of converting political will 
and power into social governance and all political actions are prepared, guided, and controlled 
by language. Politics, he says, is not just conducted through language, but much of politics is 
language: "politics partly consists in the disputes which occur in language and over language" 
(Fairclough, 1989, p.23). Given this definition and once established the nature of politics, it is 
evident why the use of a corpus is one of the most suitable for investigating the candidates' 
characteristics and political thought. 
In the book Trust the Text (2004, p.21) Sinclair explains how language is not fully 
described by dictionaries, vocabularies, and glossaries; but how language is a living tool of 
communication, which is continually updated and modified through the use that speakers make 
of it. This characteristic of continuous renewal makes it impossible for any traditional 
instrument to record any particular use, whether grammatical or not, until the actual recognition 
by dictionaries, which occurs only when almost all speakers "accept" universally a new nuance 
of meaning or a new phenomenon. In this regard, Sinclair himself in The lexical Item (1998, p. 
13) draws attention to the difficulties of a non-native speaker of a language in tracing the true 
use of words through canonical instruments. In order to better explain this aspect, a key concept 
must be introduced: semantic prosody. This is a term created by Sinclair that is used to denote 
how some terms that may appear neutral can be interpreted with a negative or positive meaning 
through frequent occurrences with certain words. We can have a better understanding of this 
term through the study of the word "budge", that the author proposes as a fitting example.
4 
 
With this instance Sinclair shows through a study of its collocations, namely "the 
occurrence of two or more words within a short space of each other in a text", that this term has 
a clear negative semantic prosody. This, however, is not recorded at all by common dictionaries, 
which only provide a generic meaning of the term, not mentioning any reference to the actual 
communicative capacity of the word. This ability does not escape the speakers, however, who 
carefully choose, albeit often unconsciously, a certain term to express a specific concept. 
Corpora’s usefulness consists above all in the number of texts that can be analyzed, once 
compiled based on genre or domain, and the opportunity to construct and analyze them with a 
semi-automatic process, as Biber explains: 
Text corpora provide large databases of naturally occurring discourse, enabling empirical analyses 
of the actual patterns of use in a language, and, when coupled with (semi-) automatic computational 
tools, the corpus-based approach enables analyses of a scope not otherwise feasible. (Biber et al. 
1994, p.169, in Partington 2003, p.6). 
Within Corpus Linguistics two main methods for analyzing a collection of texts are 
distinguished: The corpus-driven method, which finds the material to be analyzed based on the 
first results extrapolated from the corpus, and the corpus-based approach, which involves the 
early selection of expressions that appear interesting for the study. (Gray, Biber, 2015, p.126). 
In this analysis, both approaches will be used to investigate the various features of the 
language used by Trump and Clinton. The reason for this choice will be explained later in the 
chapter on methodology. 
 
1.2 Critical Discourse Analysis and Political Discourse Analysis
5 
As already mentioned in the introduction, this paper analyzes the transcripts of speeches 
delivered by Donald Trump and Hillary Clinton in the context of the 2016 United States 
Presidential Elections. In the composition of such speeches, particular use of the language is 
made, on which a study can be conducted through the use of Political Discourse Analysis 
(PDA). It is, therefore, appropriate to explain briefly, given the vastness of the subject, what is 
meant by Political Discourse Analysis. Before doing so, however, it is necessary to explain the 
more general concept of Critical Discourse Analysis (CDA), and how such an analysis can be 
conducted. Machin and Mayr describe CDA as a school that has traditionally been "concerned 
with ideologies that are hidden in language" (2012, p.15). CDA has grown quickly over the last 
decade of the 20th century as a continuation of critical linguistics that appeared in the 1970s 
(Barletta Manjarrés, 2007, p. 35). It is, in fact, a relatively new field developed in language 
study in which discourse is considered as "a form of social practice" (Fairclough & Wodak, 
1997, p.258). CDA investigates the relationship between language, power, and society, and 
pays much attention to the crucial role that context plays in discourse (Wodak, 2001, p.8). It is, 
in other words, "discourse analysis with an attitude" (Baker, 2006, p.71) and an interdisciplinary 
analytical viewpoint which looks into the relationship between power and discourse, and 
particularly it investigates the way in which authority and social inequality are constructed, 
sustained, reproduced and resisted in the discourse of written texts and spoken words 
(Kazemian & Ashemi, 2014, p.4). This type of analysis is considerably relevant as "it bridges 
the gap between the structural form of language and the external social world it tries to 
represent" (Kazemian & Ashemi, 2014, p.4). In this way, a deeper understanding of discourse 
can be obtained. In this study, CDA will be used in association with Corpus Linguistics, but 
this will be explained in the next subchapter, first, it is necessary to go deeper into CDA. 
CDA takes into account different aspects of the language, depending on what the 
researcher intends to focus on. Generally, however, it can be said that there are two main ways 
to follow: the "top-down" approach and the "bottom-up" one (Woods, 2006, p.11). As Woods 
reminds us, there is no clear distinction between these two approaches, but an explanation, 
although simplified, is necessary to clarify the position adopted in the analysis of the speeches 
of the two candidates. When the "top-down" method is used: 
discourse analysts begin from an understanding or a conceptualization (their own) of the context in 
which the discourse is taking place. [...] they look ‘down' from this position into the utterances 
produced and rehearsed, in the expectation of finding evidence – linguistic evidence – of the 
assumptions, expectations and social constructs that create and define that context. (Woods, 2006, 
p.11)