Facultat de Filosofía i Lletres

Memòria del Treball de Fi de Grau

From the Fantastic to Computing : A Corpus- Based Study of Troll in Recent American English

Laura Muñoz Vicens

Grau d’Estudis Anglesos

Any acadèmic 2019-20

DNI de l’alumne:41541717Z

Treball tutelat per Lucía Loureiro Porto Departament de Filología Espanyola, Moderna i Clàssica

S'autoritza la Universitat a incloure aquest treball en el Repositori Autor Tutor Institucional per a la seva consulta en accés obert i difusió en línia, Sí No Sí No amb finalitats exclusivament acadèmiques i d'investigació x

Paraules clau del treball: troll, slang, folklore, lexical change, corpus linguistics.

Abstract Lexical and semantic change has long been studied by many researchers and, with the introduction of the internet in our lives, our language has been even more exposed to these changes. Many slang terms resulting from the use of this phenomenon have been studied and analyzed, one of them being troll. Although there have been some works carried out to study the perception of this words, some questions are yet to be answered such as which meaning is more frequent now or how is it that this change occurred. This paper studies troll in the Corpus of Contemporary American English (COCA) by creating a database with a total of 1738 tokens where they have been analyzed according to different variables. This has been done in order to prove whether the new slang term is more used than the already existing meaning attached to troll.

Keyword Troll, slang, folklore, lexical change, corpus linguistics.

Table of Contents 1. Introduction…………………………………………………………..p.1 2. Theoretical background………………………………………………p.1 2.1. Semantic and lexical change………………………………….….p.1 2.2. Troll………………………………………………………………p.2 3. Methodology………………………………………………………….p.4 4. Results………………………………………………………………...p.7 5. Discussion and Conclusions…………………………………………..p.12 6. Works Cited…………………………………………………………...p.14

1. Introduction Many works have been written on semantic and lexical change in the English language (Singleton 2000; Stockwell and Minkova 2001; Kay and Allan 2015) and, taking into account the speed with which the internet is advancing and influencing our lives and our language, others have already studied the language change conditioned by this same medium (Crystal 2006). Many slang terms have been created as a result of a change produced by the internet, to the point of making their way into our everyday offline language so we can hear someone say “” and still understand what they mean, although there might be some exceptions (Cimarusti 2013). Some of these words have been roaming in our language for some time, but they just added a new meaning during the last decades. Such an example is troll, which was adopted in English in the 19th century. Some researchers have written about the use of this word and how it is perceived by the public as a slang term (Dunlop 2013; Hardaker 2013; March 2019), but the questions about how this word has changed its meaning over the last decades and which meaning is it that prevails over the other are still open in the air. Thus, this paper endeavors to study troll as means to answer those questions by creating a database where the tokens will be analyzed by postulating the hypothesis that troll as a slang term has increased in frequency and is now more used in that sense than with its already existing meaning. This paper will examine the change produced in troll by using a corpus of contemporary American English to show whether the slang term is overused in relation to the previous existing meaning, i.e. troll as a Scandinavian folklore being. And therefore, the structure of this paper will be divided into sections where section one will consider previous studies about semantic and lexical change as well as other researches made on troll as a slang term. Section two will explain the methodology followed, that is, the database used to analyze the tokens of the corpus. Section three aims at showing the results extracted from the database that will be later discussed in Section four along with a conclusion of the findings.

2. Theoretical Background 2.1. Semantic and lexical change The study of language change is common among researchers as it is basic to understand the language as we use it today. As the “knowledge of the world moves on from one point in time to another” (Singleton 2000, 143) it is common to find semantic change, as the understanding people have of their surroundings varies with time. “Semantic change can be occasioned by

1 social and political restructuring as well as by technological developments” (Singleton 2000, 143). For example, as expressed by Stockwell and Minkova, “the development of new material and social conditions may cause words to become unnecessary. … A shift of attitudes may render some words socially unacceptable, while others become highly fashionable or socially relevant” (2001, 149). However, not all changes are “observable in all speakers or writers at the same time or may not take place in all varieties of a language” (Kay and Allan 2015, 75). Some scholars (Stockwell and Minkova 2001; Kay and Allan 2015) have divided the semantic change into different forces in order to classify it. Stockwell and Minkova differentiate external forces, accidental associations, internal forces and loss of specificity. In the first mechanism, external forces, they include changes due to technology and current relevance, with examples such as tools, screen or icon, which have obtained a new meaning with the development of computing. In internal forces, they make reference to analogy and how it “involves the perception of similarity between some concrete object or process and some abstract concept or process” (Stockwell and Minkova 2001, 152). Kay and Allan coincide in some classifications such as the external factors, and the internal factors, where they consider polysemy, homonymy and synonymy, but they add the stylistic factors, without classifying the loss of specificity or accidental associations as Stockwell and Minkova did in their book. Following the classifications above mentioned and more importantly, those changes conditioned by the development of new technologies, linguist David Crystal focuses his theory on what is called Internet linguistics, which studies how the internet influences the language. He defends that “the linguistic consequences of evolving a medium in which the whole world participates … are also bound to be far-reaching” (Crystal 2006, 5), making reference to the global reach of internet and its applications and uses and the implications this has on language.

2.2. Troll Many researches have been conducted on the many different changes some words have undergone to be what they are today. The interest on slang terms has also grown as it has taken an important part of our vocabulary, partly influenced by the use of the internet and . One of these words is troll. According to the Oxford English Dictionary, s.v. troll n.1. (computing slang) is “a person who posts deliberately erroneous or antagonistic messages to a newsgroup or similar

2 forum with the intention of eliciting a hostile or corrective response. Also: a message of this type.” This might have been influenced, as stated by the same OED, by the word carrying the meaning of a Scandinavian supernatural being. The first piece of literature in which this type of trolls can be found is Prose Edda (Snorri 1916), a 13th century work written in Icelandic and it is, up to date, a source of knowledge about the Norse mythology. Different websites (McKay 2018; Furey n.d.) describing Old Norse troll coincide in some aspects; trolls are unfriendly, ugly and brutish, and they tend to live in dark places such as in caves or underground. These behaviors of the mythical trolls are somewhat similar to the definition given by the OED under troll n.1. However, contrary to the OED definition for the slang term, some have stepped up to show that the meaning of troll is not always as negative and it can be perceived differently depending on who we are asking. Some say that “the word once had quite a specialized meaning limited to a particular sort of disruptive behavior, but it has now become a catch-all term to describe any behavior that some journalists and editors deem inappropriate” (Dunlop 2013, n.p). Therefore, some group of people would tell us that someone is a troll if they do not like the attitude others have. This is just an example that would prove Hardaker’s claim correct when she states that “trying to discover [a definition of troll] or, if one didn’t exist, to create a clean, robust, working definition that everyone would agree with would be close to impossible” (Hardaker 2013, n.p). In the same article, she exposes the different definitions given to the word in different sources such as the OED, The Guardian or Urban Dictionary and the different ways in which it is actually used, which may vary depending on the point of view given and the interpretation of each speaker and interlocutor. A survey was conducted (March 2019) in order to find out how people perceive and define the action of trolling. As expressed by Hardaker, “users develop their own particular understanding of what [a] word means” leading to a “semantic instability.” (2013, n.p) In this survey, people were given different internet situations, one with a “mischievous and comical behaviour” and another with a “malicious and antisocial behaviour” (March 2019, n.p) and they had to say whether they considered the comments as trolling. Even though many people coincided with the second case being an act of trolling, not all agreed upon the first situation were the intention was more comical. Hence, we have another example of how people perceive the same words differently and they even have both, negative and not so negative implications.

3

3. Methodology This is a corpus-based study using the Corpus of Contemporary American English, or COCA (Davies 2008-). It is the most widely-used corpus of English as it collects more than one billion words from texts from 1999-2019. Since it is a recollection of contemporary American texts, there are different genres: spoken, fiction, magazines, newspapers, academic texts, TV, movies subtitles, and webpages (Davies 2008-). The study will be about the analysis of the word troll in the singular form, which has a frequency of 4786 occurrences. For practical purposes, other forms such as trolls, trolling or trolled will not be analyzed as the number of occurrences to analyze would grow exponentially. Two different samples of 1000 occurrences each have been examined, total of 2000 occurrences, and out of them, repeated occurrences have been taken out of the list in other to avoid unnecessary repetitions of tokens, some proper names have also been left out of the database. The resulting number is of 1738 tokens. These have been analyzed by using a database where the different occurrences have been studied with variables that would later help expose the results. There are, among them, some examples in which the word troll appears as a proper noun of either characters or places. In some of these cases, most of the variables have been left blank because the information given is not enough. To analyze the sample with all the occurrences, the definitions provided by the Oxford English Dictionary have been used since, as already mentioned, different definitions can be found depending on the used sources. Since the sample includes troll as a noun and as a verb, all the entries have been taken into account when allocating their meaning. In some cases, the meaning is ambiguous to allocate it under the different entries on the OED Online as they might have similar connotations in some contexts. An example is the difficulty to differentiate the meaning of troll as under the troll n.1 entry “draft additions March 2006” and the one under troll n.2 “daft additions September 2008”. In these ambiguous cases, the text-type has been the final factor to determine whether it is one or the other. Moreover, there were some cases in which the meaning was not included in the OED. This problem has been solved by categorizing them as “look for” or “look through” in the meaning (OED) variable instead of writing the OED entry that would correspond. However, a total of 17 tokens have been sorted as NC (No Context), this means that these could not be properly analyzed because they lacked context to understand its meaning. The database used to analyze the 1738 occurrences has been divided into different variables:

4

- Text-type. - Year. - Grammatical category (noun or verb). - Meaning. - (If noun) Syntactic function. - (If noun) Main verb. - Collocates. - Special typography. - Active or passive voice. - Gender pronoun. - Apposition. - Connotation (positive, negative or neutral). - Relation. First, the tokens have been categorized by year and text-type following the information already provided by COCA. Then, through another variable, they have been separated as nouns or verbs (N and V) and, after having done so, they have been assigned a number that represent the entry in the OED so that “1”, “2” and “3” represent troll n.1, troll n.2 and troll v. respectively. Those tokens in which the meaning was related to the draft additions of either of the three entries “(add)” was used and “(fk)” was used for the tokens that have troll with the Scandinavian folklore meaning. In the cases where troll is a noun I have indicated their syntactic function as subject (S), object (O), predicative complement of the subject (PCS), predicative complement of the object (PCO), agent of the sentence, vocative or complement (place complement, manner complement and such similar cases are included in this classification). Moreover, in the case of nouns, the main verb of the sentences has been indicated if possible in order to find out if there are any verbs that accompany a meaning or the other. Collocates have also been analyzed by taking into account the most general words used with troll to know if there are any special collocations. Later, a variable for special typography was included for those cases in which troll appeared with uppercase letters or with symbols like an exclamation or inverted comas to find out if they influence the meaning of the word or if it serves as emphasis. The voice of the sentences has also been taken into account to do the analysis in other to find out if the examples were in the passive or the active voice. Another variable was included to determine the gender pronoun used to refer to the word and, in those cases in which there is no gender pronoun such as he, she, his or her but there is a name or another

5 word that gives us a clue to know the gender such as a name or a title, they have been written as they appear in the text. This will be used to know which gender is used to accompany troll. Another variable was left to indicate if there is apposition in the tokens. In this column, I have also indicated if there was any kind of explanation or definition given to troll. Finally, a variable where to put the connotation of the occurrences has been included and they have been divided into positive, negative and neutral. Occurrences concluded as having a negative connotation are those that refer to troll as in example (1) and (2). (3) is an example of troll considered with a neutral connotation. Most cases in which troll is a reference to a folklore troll have been taken as neutral unless the negative connotation was noticeable such as in example (4). The occurrences in which the connotation of troll was somewhat ambiguous, they have been labeled as neutral such as in example (5), where it seems that the words with the negative connotations is being used positively by the subject. Lastly, a case such as (6) has been deemed to have a positive connotation but cases such as this one are very limited. The last variable that has been used is to easily know what troll makes reference to, and only those tokens which meaning was related to angling, toys, names, games and the movie of the same name have been included in this variable independently of their word class. (1) Fortunately he was banned for being for being a name-morphing troll. (COCA, , 2012) (2) You don’t think, troll. (COCA, web, 2012) (3) A troll held him in its arms. (COCA, fic, 1992) (4) Treating him like a dimwitted and violent troll. (COCA, web, 2012) (5) A self-described Internet troll. (COCA, mag, 2013) (6) He’s like a plump little troll. (COCA, mov, 2006)

The data has been analyzed diachronically and the time unit considered is the decade. Since COCA includes texts produced from 1990 to 2019, I have decided to divide the sample into three groups: (1) the occurrences from 1990 to 1999, (2) from 2000 to 2009, (3) and from 2010 to 2019, in order to have three roughly equivalent groups. This will help me to find out in which period troll has a higher frequency as well as to identify the turning point in which one meaning started to be used more often than the other. I have also analyzed the meaning of the words in respect to the text-type and the year in order to know whether this has an influence on the use of the word and also with the aim of identifying in which text-type(s) exactly each of the two meanings prevail, that is, troll as a supernatural creature or as a slang term.

6

4. Results The data were firstly analyzed across time, with the aim of finding out any diachronic tendency. As mentioned, the time division followed was the decade. Because each period is similarly represented in COCA (roughly 248,000,000 words per decade), in what follows I will offer raw frequencies. Table 1 represents the total number of occurrences in each decade independent of the other variables. Table 1. Frequency of occurrences by decade. Dates Occurrences

1990 - 1999 243

2000 - 2009 217

2010 - 2019 1278

The number of tokens in the first two decades is very similar (in fact, the difference between them is found not to be statistically significant) but there is a significant increase1 of tokens in the third decade, from 2010 to 2019. A finer diachronic analysis revealed that this is due to the fact that 1043 tokens proceed from the year 2012 from different blogs and other webs. This very unbalanced distribution of troll led me to explore different hypotheses regarding the different meanings this word may exhibit. An a priori hypothesis one could posit in order to explain this has to do with the fact that COCA includes a considerable percentage of words taken from blogs (where troll is expected to be often used as a computer slang term), so I first thought that maybe this year was the “golden age” of blogs. However, this hypothesis seems to be incorrect as there is evidence to prove the contrary. Some sources (Cohen-Setton 2015; Warnica 2017; Murphy n.d) agree that the “golden age” of blogs went from 2003 to 2009 approximately, just three years before the year in question. Another hypothesis explores an avenue where the original meaning of troll prevails, i.e. the magical creature that lives in the forest. In this sense, it was hypothesized that videogames with fantastic creatures such as trolls gained its momentum in 2012 and, therefore, more talk about the characters and gameplay would make the number of tokens increase. However, only 36 tokens seemed to correspond to the gaming register and, taking into account the total number of occurrences, it is not a relevant number. Therefore, this hypothesis has, as well, proved wrong. Having rejected these two hypotheses, the only plausible explanation of this accidental disparity in the results may probably be because the randomizing tool in COCA has some

1 According to the online calculator Log-likelihood. Accessed July 30, 2020. http://ucrel-api.lancaster.ac.uk/cgi- bin/llsimple.pl?f1=243&f2=1278&t1=248395708&t2=248145425. 7 kind of bias that leaps towards 2012 results. Only a complete analysis of all the tokens of troll in COCA could test whether this hypothesis holds true or not, but that falls out of the scope of this thesis. For the purpose of the paper, it is also important to know in which text-type is more common to find troll being used with either meaning and, later, what is the predominant meaning in each type. Table 2 shows the results obtained to know the total amount of occurrences found of each text-type.

Table 2. Total number of occurrences per text-type

TEXT-TYPE

620 405

232 Total

111 111

100

73

54 32

BLOG WEB FIC MAG MOV TV ACAD NEWS SPOK

The predominant meanings of troll are those that are listed as draft additions in the OED entries with a total of 1049 occurrences of which 887 are nouns with the meaning of computing slang and 125 are verbs with the same computing slang meaning. The remaining 37 are those tokens that mean “an unpleasant or ugly person”, as it appears under troll n.2 in the OED. 530 of the occurrences are those that mean troll as in the Scandinavian folklore fantastic creature and the rest of occurrences have other meanings relating to angling, movement or meanings not found in the OED (listed in the data as “look for” and “look through”). Only 17 are listed as NC for not having enough context to determine its meaning.

Table 3. Predominant meaning in each decade.

Meaning per decade

NC 10 218 Others 31 1013 Decade 3 Decade No OED 6 3 2 (fk) 154 32 slang 23 Total Decade 2 Decade 5 NC 4 158 Others 63 13 Decade 1 Decade No OED 5 0 200 400 600 800 1000 1200

8

As seen in Table 3, in the first two decades, 1990 to 2009, the predominant meaning of troll is the folklore one and the few tokens making reference to the word as a slang term started being used in 1994 with only two tokens. In the decade that goes from 2010 to 2019, the number of occurrences with the draft additions’ meanings increased significantly2 in contrast to the previous decades. Moreover, 902 of the total tokens with such meaning appeared in blogs and webs and the remaining 147 are distributed among academic texts, magazines, movies, news, TV, fiction and spoken texts.

Table 4. Syntactic function of troll as noun.

FUNCTION

553

441 379

236 Total

69

47

9

3 1

AGENT COMPL MOD O PCO PCS S VOCATIVE VERBS

The cases in which troll appeared as a noun were analyzed in terms of syntactic function in order to know whether troll is active or passive. Most of the occurrences show the word as the subject of the sentence, the object or the predicative complement of the subject (PCS). It also appeared to be used as a vocative, although there only 69 occurrences of such function. 48 cases were found to be complements, with place complement as the most common one in examples such as (7), where troll makes reference to the name of a place. Only one of the tokens is a modifier of the verb (8). Moreover, only 10 cases are in the passive voice, hinting us that troll is mainly used in the active. (7) Let Ragweed lead her back to troll country. (COCA, fic, 2012) (8) These lovers are already troll stamping you. (COCA, blog, 2012)

The most common verb among the occurrences is the verb to be and its conjugations and it is the only verb that is repeated more than 50 times. This verb is, moreover, most used in occurrences during the third decade, from 2010 to 2019, in utterances such as (9) and (10).

2 According to the online calculator Chi-Square. Accessed August 02, 2020. https://www.socscistatistics.com/tests/chisquare2/default2.aspx 9

Although the other verbs used do not imply any important change to the analysis, it is important to mention verbs such as feed and engage and other variants of the same such as the gerund forms. These verbs are used mostly during the third decade and go together with the noun troll as computing slang. In the case of feed, it is mainly used to refer to the fact that the troll is being given reasons to continue with its messages and hostile behavior towards the rest of the users. (9) I am not a Troll. (COCA, blog, 2012) (10) He is a live internet troll. (COCA, spok, 2016) (11) I’m not going to engage a troll like you. (COCA, web, 2012) (12) Stop feeding the troll. (COCA, blog, 2012)

The collocates and words that are most commonly used with troll are different but there are some that are to be highlighted. Collocations such as concern troll, , patent troll and internet troll have been found to be the most common and noticeable. There are 34 occurrences of concern(ed) troll in the sample meaning “someone opposing an idea or viewpoint, yet acting like they are an advocate for the cause” as defined in dictionay.com. There are 21 troll farm(s) occurrences and most of them are most specifically Russian troll farm(s). This collocation is used in the third decade, in the years 2017 to 2019, and the reason behind it is because, in 2016, year of the election for the President of the United States of America, Russia was allegedly trying to manipulate the votes of Americans by using social media (Dave 2018; Alba 2020). For patent troll, there are 26 tokens from the second and third decade using this “pejorative term for a company” (Gregersen 2015, n.p). Lastly, the combination of internet and troll appears 24 times, although not all of them appear as the collocation internet troll as in example (10), to remark that the meaning gone after when writing troll is that of computing slang. Other collocations to be mentioned are bridge troll, cave troll or forest troll, with 13, 10 and 4 occurrences respectively and meaning, in this case, folklore troll.

Table 5. Gender specifications Others He / his / him She / her Names Male references Female References

125 27 31 24 7

10

In the variable of gender pronoun, 125 results have appeared with a male gender pronoun, compared to 27 for female gender pronouns. There is one case in which, instead of using he or she, he/she is being used not to choose a gender over the other being the gender of the subject unknown as in example (13). As many were the cases in which troll appeared with a name, 31 cases, they were specified and 23 of them are male names, two are female names and 6 are names that could not be identified as belonging to one group or the other. (13) He/she is a troll. (COCA, blog, 2012)

With the aim of finding out if there is some kind of special typography used for the word troll, I have considered different options. Firstly, I paid attention to whether the word was spelled in the lower or in the uppercase letters. Secondly, I considered any other notational marking, such as inverted comas as shown in Table 6. Out of the 1738 total number of examples, 136 tokens are written as Troll, these cases are either using the word as a name of a place or to refer to a person, 18 are written with all uppercase letters as in TROLL, with the aim of giving emphasis, as well as with the previous example. In one case the word can be found on its own with an exclamation symbol as Troll! On 27 cases the word is spelled between inverted comas. Some of these cases uses double inverted comas to use troll as a word as in example (14). As mentioned in Section 3, the database also considered whether the word troll was followed by an apposition, because that was interpreted as evidence of the speaker’s need to clarify the meaning of the word for the listener. The results render no examples of such syntactic construction, but, interestingly enough, some occurrences did include a definition or an explanation of the word such as in example (14), where the sentence is followed by a definition or an explanation, as found in example (15) as well. (14) The actual definition of “troll”. (COCA, web, 2012) (15) A troll is a web poster who attempts to disrupt discussion of the subject in hand. (COCA, blog, 2012) Table 6. Tokens with special typography. Troll TROLL Troll! “troll” 136 18 1 27

Table 7. Connotation Negative Neutral Positive 976 748 2

11

Connotation-wise, only 2 cases have been found with a positive connotation. Most of the occurrences in which troll meant the folklore creature have been analyzed as having a neutral connotation unless it was explicit that it had to be negative such as in example (16) because in this case, Troll is something to be afraid of and be protected from. Just like that, most occurrences in which troll had some of the draft additions’ meaning was categorized as negative with the exception of those cases in which it has been categorized as neutral because of ambiguity of connotation. Cases where troll made reference to names, games or angling have also been included in the neutral group. 36 occurrences have been categorized as referring to a game, 45 to a name (either given to a person who is an internet troll, or places with troll in the name), 17 to a toy (troll dolls), 89 related to angling and 2 cases to the movie of the same name. (16) He assisted us during the Troll invasion. (COCA, mov, 2018)

5. Discussion and conclusions Although the results may have been slightly altered by the fact that the randomizing tool of the corpus shows more results from the year 2012 in comparison to the rest, figures show how the use of troll has been increasing during the last decade. This increase is due to the fact troll is used with the new meanings on 1049 tokens with 902 out of them being used in blogs and other web pages. These types of texts seem to be the birth of this new meanings for troll, as a troll in is a person who posts messages to create a dispute between users. But, why has troll been chosen as the word to describe this type of people and messages instead of another one? The answer seems to be behind the original meaning of troll (looking back at the Scandinavian trolls) and the behavior these fantastic creatures were thought to display. As explained in section 2.2, trolls are unfriendly towards humans and try to deceive them, they are ugly and tend to live un dark places like underground or caves. Internet trolls are, in some aspects, quite similar, they deceive other internet users by posting erroneous information and writing sometimes infuriating texts to create a debate. Although they do not literally live in dark places or caves (although some might if we are to consider their bedrooms as such), they do hide their identity behind the computer screen by creating, most of the times, false names and profiles. is their darkness and it is through it that they feel free to behave in such a manner. There were even some cases in the sample where this juxtaposition is present as in example (17), where, talking to an internet troll, they make reference to a fantastic troll

12 and its cave. Moreover, trolls are ugly and, as the draft addition under the troll n.2 entry in the OED defines, we can tell someone is a troll and understand that it is because they are ugly. (17) How is it you have an internet connetion in your troll cave? (COCA, web, 2012)

Then, this semantic change may not only be caused because of external forces, that is technology, following Stockwell and Mikova’s division. Although there is no denying that the new meanings acquired by troll are influenced by technology, analogy may have also influenced in this change, especially taking into account all the similarities between the fantastic creature and the internet troll. However, this does not necessarily mean that we can discard the fact that technology placed an important role in this semantic change. Therefore, we could consider that this change has been influenced by both external and internal forces. Moreover, as seen in the results, the change does not mean that the already existing meanings of troll have disappeared but, instead, the word has undergone further widening of meaning producing, at the same time, more homonymy. The results also show that the new added meanings are more used than the others but neither has disappeared completely. During the third decade, troll as a slang term seems to have taken over, compared to the previous two decades were the folklore and angling related meanings of troll had a higher frequency than the additions. The use of troll from 2010 to 2019 also seem to be directly linked to specific types of texts. Blogs and websites have a higher frequency use of the slang term while academic writings and fiction tend to talk about Scandinavian trolls. There are, however, some references to these beings in blogs and webpages. In most of these cases, these trolls are characters of role-playing games (or RPG) that are famous for using magical and fantastic creatures such as elves, trolls, giants, and others, or are a reference to troll dolls. This is proof that it is unlikely that troll as a mystical being will become a word only used in academic texts or fiction because, although in a limited number of occurrences, they will still appear in blogs about videogames and similar webpages. There also seems to be a generalization of the slang term troll. Some tokens that were analyzed as pertaining to that category showed a change as it was used to refer to a person, but not necessarily in an internet context. This means that troll is generalizing to also include people who, in real life, also make comments that can be inciting to a debate and infuriating to the other interlocutors. In example (18), for instance, it is a troll because of the response given, typical of an internet troll, but given in person. This probably means that, in the next

13 decades, this will be more accepted and more people will use troll to refer to any person who either writes or says something that is not correct or is impolite and with bad manners, independently of the of the context where it has happened. (18) But that’s troll doctors for you—no bedside manner. (COCA, fic, 2006)

This paper has analyzed the word troll from 1990 to 2019 in order to determine its frequency and whether the meaning used has changed through time. By trying to answer the research question of how has the meaning of troll changed and which meaning is more used, I have shown that, indeed, the slang term has increased its frequency to be more used than the previous meaning. It has also been discussed a possible way how the meaning might have changed taking into account the results and other different sources, to conclude that different factors have taken a role in the change. Although it remains to be seen, a new change can occur within the sang term to become more generalized and include more people other than internet users, but more time would be needed in order to obtain the data required to prove this hypothesis.

6. Works Cited Alba, Davey. 2020. “How Russia’s Troll Farm Is Changing Tactics Before the Fall Election.” . Accessed July 28, 2020. https://www.nytimes.com/2020/03/29/technology/russia-troll-farm-election.html. Cimarusti, Nick. 2013. “Internet slang meshes into everyday language.” Daily Trojan. Accessed August 02, 2020. https://dailytrojan.com/2013/10/01/internet-slang-meshes- into-everyday-language/. Cohen-Setton, Jérémie. 2015. “Is Blogging Dead?” Bruegel. Accessed July 27, 2020. https://www.bruegel.org/2015/02/is-blogging-dead/ “concern troll”. Dictionay.com. 2020. Accessed July 28, 2020. https://www.dictionary.com/e/slang/concern-troll/. Crystal, David. 2006. “A Linguistic Perspective” In Language and the Internet. New York: Cambridge University Press. Davies, Mark. 2008-. The Corpus of Contemporary American English (COCA): One billion words, 1990-2019. https://www.english-corpora.org/coca/.

14

Dunlop, Tim. 2013. “How The Word ‘Troll’ Has Been Redefined By The Powerful.” The Guardian. https://amp.theguardian.com/commentisfree/2013/aug/16/trolls-journalism- editors. Furey, Lauren. n.d. “Exploring the Mystery of Scandinavian Trolls.” Scandification. https://scandification.com/exploring-the-mystery-of-scandinavian-trolls/ Gregersen, Erik. 2015. “Patent Troll.” Encyclopædia Britannica. Encyclopædia Britannica, Inc. Accessed July 28, 2020. https://www.britannica.com/topic/patent-troll. Hardaker, Claire. 2013. “Web of Words: A Short History of The Troll.” ESRC Centre for Corpus Approaches to Social Science (CASS). http://cass.lancs.ac.uk/621/ Kay, Christian and Kathryn Allan. 2015. “How and Why Words Change Meaning” in English Historical Semantics. Edinburgh: Edinburgh University Press. http://www.jstor.com/stable/10.3366/j.ctt1g0b29p.10. Lee, Dave. 2018. “The tactics of a Russian troll farm.” BBC. Accessed July 28, 2020. https://www.bbc.com/news/technology-43093390. March, Evita. 2019. “Online Trolling Used To Be Funny, But Now The Term Refers To Something Far More Sinister.” The Conversation. https://theconversation.com/online- trolling-used-to-be-funny-but-now-the-term-refers-to-something-far-more-sinister- 110272. McKay, Andrew. 2018. “The Mythology of Norwegian Trolls.” Life in Norway. https://www.lifeinnorway.net/norwegian-trolls/. Murphy, Justin. n.d. “The Second Golden Age of Blogging.” The Other Life Now. Accessed July 26, 2020. https://theotherlifenow.com/the-second-golden-age-of-blogging/. Singleton, David. 2000. Language and the Lexicon. New York: Oxford University Press. Snorri, Sturlson. 1916. The Prose Edda, trans. Arthur Gilchrist Brodeur. New York: The American-Scandinavian Foundation. Stockwell, Robert and Donka Minkova. 2001. English Words: History and Structure. New York: Cambridge University Press. "troll, n.1". OED Online. June 2020. Oxford University Press. Accessed July 21, 2020. https://0-www-oed- com.llull.uib.es/view/Entry/206613?rskey=u9VKzu&result=1&isAdvanced=false. "troll, n.2". OED Online. June 2020. Oxford University Press. Accessed July 21, 2020. https://0-www-oed- com.llull.uib.es/view/Entry/206614?rskey=u9VKzu&result=2&isAdvanced=false.

15

"troll, v.". OED Online. June 2020. Oxford University Press. Accessed July 21, 2020. https://0-www-oed-com.llull.uib.es/view/Entry/206615?rskey=u9VKzu&result=3. Warnica, Richard. 2017. “The brief life of blogging: Golden age of blogs was both exhilarating and exhausting.” National Post. Accessed July 26, 2020. https://nationalpost.com/life/a-brief-life-of-blogging-the-golden-age-of-blogs-was- both-exhilarating-and-exhausting.

16