A corpus-based study of reviewers’ usage of speech acts

Abstract This study examines online reviews about booking a hotel room to see what factors affect reviewers’ rating and make online reviews persuasive to the reader. Linguists study online reviews from different perspectives such as their communicative functions, their discourse features, their stylistic features, and other perspectives. The study describes the types of speech acts used by the reviewers and examines their purposes to find out how to rate, evaluate and describe a product or service, and on that basis, to provide recommendations to others for or against a particular product or service. The study is conducted on 100 online reviews from two famous social media sites TripAdvisor and Booking.com using a corpus-based approach that is used to verify expectations and assumptions about the function of speech acts. 1976) theory of Speech Acts is applied in this research as the major theory, besides Aristotle’s Persuasion theory and Vásquez’s framework that is based on Halliday’s lexico-grammatical approach. The classification includes the different types of speech act, namely, representatives, directives, commissives, expressives and declarations. The study mainly adopts a content analysis in which the collected data is analyzed by categorizing keywords and identifying the relationships among these words. Moreover, it investigates the usage of speech acts and on a more detailed study of how speech acts function as rating tools in social media reviewers. Results show that online reviews are a very rich digital genre that show a variation of usage of speech acts that perform different functions such as thanking, apologizing, complaining, describing, commanding, … etc. Thus, it can be concluded that an online review is a multi-function genre and its main purpose is being a persuasive text that affects the reader either positively or negatively.


Introduction
With the expansion of electronic commerce nowadays on the internet, it is likely that online reviews have become an important source of information that can help consumers to make decisions concerning their purchases. Dwidinawati et al. (2020) defined WOM as "direct communication from person-to-person of an opinion about a product and/or service". Online reviews are referred to as Electronic "word-of-mouth" (eWOM) and it is reported to have greater power than traditional WOM because as Dwidinawati et al. (2020) demonstrated, it reaches a large population rapidly. It is considered as a prominent source of information for potential customers. They add that the online consumer collects and compares important information when purchasing a certain product. This information helps the consumer to evaluate similar products and come up with the best choice. Therefore, eWOM is best defined as a peer-to-peer review written by a user or consumer for a product or a service based on his/her own experience as a user of the reviewed product or service. It is to be noted that online reviews include other consumers' comments that are posted on the Internet on different websites. With the help of the Internet, information is not only confined to news media or large businesses. In fact, anyone can share his thoughts with millions of Internet users and influence others' decisions through electronic word-of-mouth (Duan et al., 2011). Many information and cultural goods are experience goods that a consumer needs to taste before assessing its quality and its location with respect to his or her ideal product (Bounie et al., 2005). Therefore, eWOM is essential for the feedback written by a user about a product or service he used can help friends, family and even strangers make purchase decision. This feedback can be positive or negative depending on the user's experience about the product or service and, hence, can have the power to direct other consumers' decision. Many customers search online review sites to find comments or feedbacks about a certain product or service. According to many studies, 4 out of 5 consumers reverse their purchase decisions based on negative reviews (Table 1) (Campbell, 2016). However, it is to be noted that some companies may pay to post fake reviews in an attempt to improve their reputation or mislead consumers in their purchase decision making. They give false claims about the characteristics of a product or service or create a false impression about the goods. Therefore, studying online reviews is a critical topic, especially the factors affecting reviewers' rating.
Booking a hotel room is not an easy task since it is an experience service that needs to be evaluated in terms of the quality of the room and the location of the hotel, therefore, it is likely that one needs to read online reviews on different websites before making a decision about which hotel is suitable according to his own needs. The question is what are the factors that affect reviewers' rating while writing an online review. Accordingly, the current study examines online reviews of booking a hotel room in an attempt to determine the most important factors that should be present while rating a room in a hotel.
According to Vásquez (2014), "the main purpose of online consumer reviews is for individuals to rate, to evaluate, to describe, and on that basis, to provide recommendations to others for or against a particular product or service" (p. 25). He also states that online review writers are referred to as "prosumers". This term is coined in the 1980s, it is a blend of "producer" and "consumer", and it is used to mean a blurring of the roles of producer and consumer (Vásquez, 2014, p. 4). Therefore, throughout the study, the term "prosumers" is used to refer to online review writers, while the term "consumers or purchasers" is used to refer to those people who read the reviews to make their decisions.
Language is the most significant part in human communication. In studying language, people use language not only to say something, but also to do something. This is referred to as "Speech acts". Speech acts is the study which deals with how to do things with words. Speech acts is not only found in everyday life; however, it is also found in written texts such as online reviews which is the area of interest in this study. Purchasers, nowadays, have a great opportunity to gather data about a service or a product through other prosumers' reviews that are posted on the social media. Such reviews may include complaints, praises, recommendations, suggestions about a service or product which can have an impact on the reader. Due to the importance of reviews in that sense, accordingly, the present study examines the language used in online reviews to examine the different speech acts found in online reviews and to describe their purposes in order to shed light on the communication tools used by the reviewers to persuade the reader.
Assuming that online reviews have a great impact on the reader, it is important to refer to the word "persuasion". All language use can in a sense be regarded as persuasive (Miller, 1980). However, there is another definition for the word "persuasion" mentioned by Pelclova' and Weilun (2018) as they limit the definition of persuasion to all linguistic behavior that attempts to either change the thinking or behavior of an audience, or to strengthen its beliefs. Thus, in this study, persuasion is understood as the reviewers' attempts to influence their readers. This can be done by different rhetorical techniques. According to Bruthiaux (2000), "the urge to persuade fellow humans is powerful, the opportunities are many, and the contexts varied". The language of online reviews can be classified under the language of advertising as there are many companies use eWOM as a way of marketing. Therefore, the study demonstrates how such speech acts used in online reviews are persuasive tools that affect the reader.
Online reviews are made up of words, phrases and sentences. Words are the smallest meaningful units of messages, however their meaning could be affected by a number of factors: collocations with other words, the objective and purpose of using these words, as well as the reader who will read them. Due to the importance of words in online reviews, the study investigates certain keywords in online reviews using corpus linguistics to show the most frequently used words and their implication in the study.
This study aims to (1) find out the factors that affect reviewers' rating while writing an online review; (2) identify and describe the types of speech acts used by the reviewers; (3) determine the most dominant speech acts used; (4) investigate the pragmatic functions of the used speech acts and explain how these speech acts are persuasive tools that can affect the reader; (5) identify the rhetorical strategies used by the reviewers as a support to the findings of the study; (6) demonstrate how the use of computerized techniques can show the relationship between combinations of words used and their significances to the reviews; and finally (7) show to what extent are online reviews persuasive texts.

Conceptualization of eWOM
The concept of electronic word-of-mouth (eWOM) was first introduced in the mid-1990s, when the Internet became a way for interaction between people, in general, and consumers in particular. One of the broad definitions of eWOM is "all informal communications directed at consumers through Internet-based technology related to the usage or characteristics of particular goods and services, or their sellers" (Stephan et al.,2008 , p. 461). Another more specified definition of eWOM is that it is "any positive or negative statement made by potential, actual, or former customers about a product or company which is made available to multitude of the people and institutes via the Internet" (Hennig Thurau & Vásquez, 2014, p. 39). According to the above definitions, eWOM is one of the modes of communication for exchanging and sharing information about a certain product or service that is spread worldwide via different platforms, i.e., the Internet, social media, or online communication.
Because of the effect of eWOM on the consumer, online reviews have been the subject of many studies in different disciplines such as marketing, economics, business research, however, there are few linguistic studies found examining online reviews. Some studies have demonstrated the impact of eWOM on purchase decisions, others have examined the communication functions in eWOM, while several studies have focused on the discourse and stylistic features of eWOM. The following sections demonstrates how linguists dealt with online reviews from different perspectives.

Communicative and pragmatic functions in eWOM
One of the essential references that dealt with online reviews is a book written by Vásquez (2014, pp. 6-7) who discussed some of the common features of reviews that users are likely to encounter as they search for product information. He addressed both the linguistic properties and the social practices related to a sample of digital texts written by consumers as a discursive practice. The book is considered as a useful source for modelling a clear and concise methodological approach to analyzing online discursive practices, as it is for the important questions raised about life in digital consumer cultures, it is found that this study is both insightful and accessible. Moreover, the book has relevance across a number of disciplines from linguistics, through cultural studies, to consumer studies, online communications, marketing, and tourism studies.
Another study that investigated the communication functions of online reviews was written by De Ascaniis and Gretzel (2013) who studied online travel review titles. They adopted a pragmatic and linguistic approach. Results described characteristics of this type of metadata and provided indications for platform managers and designers.

Sentiment analysis of eWOM
One linguistic study examined online reviews using computational approaches was written by Ganesan and Zhou (2016) who examined online reviews from a sentiment analysis, predicting the polarity of texts as positive and negative at different granularity. They tried to understand the properties of complaints and praises which is an informative subset of the negative and positive categories. The study showed that complaints and praises have distinct properties that differentiate it from positive only or negative only sentences. Another important topic that is related to online reviews is hotel responses.

Hotel responses to online reviews
Two studies examined hotel responses to online reviews. The first one is presented by Sparks et al. (2016) who examined hotel responses to negative online reviews. More specifically, they focused on the effects of hotel responses on customer inferences of trust and concern. They examined the perceptions and evaluations of prospective customers toward an online negative review and any accompanying hotel response. The effects of four variables associated with a response are tested: source of response, voice of responder, speed of response, and action frame on two outcomes variables (i.e., customer concern and trust inferences). The provision of an online response (versus no response) enhanced inferences that potential consumers draw regarding the business's trustworthiness and the extent to which it cares about its customers. Using a human voice and a timely response yielded favorable customer inferences. Inferences did not vary with response source or action frame. Implications are drawn for effective management of negative online reviews.
The other study is presented by Thumvichit and Gampper (2018) in which they represented a response to a recent call for further investigations into cross-cultural variations of hotel responses to negative online reviews. Therefore, the aim was to expand the context of this computer-mediated genre to cross-cultural investigation in order to gain insight into similarities and differences between two different groups of hotels. As such, the study compared responses to negative online reviews produced by local Thai hotels to those of top UK hotels. This type of comparative genre analysis should highlight the diverse cultural conventions of the genre. The findings of this study should make a significant contribution to English language learners and practitioners in the customer service industry. Another important point in the study of online reviews is the study of their impact on consumers' decision-making. The following section presents the related studies.

The impact of online reviews on consumers' decision-making
The influence of eWOM on consumer behaviour constitutes a broader literature. Several studies are found tackling the impact of online reviews on consumers' decision-making. However, there is still a lack of understanding of the influence of online reviews. It is found that Kem, Sesia, Christy, and Matthew (2014) examined the impact of online reviews in which they based their study on a heuristic-systematic model. They identified the factors that are important to consumers' purchase decision making. The model is empirically tested with 191 users of an existing online review site. Results showed that argument quality of online reviews (systematic factor), which is characterized by perceived informativeness and perceived persuasiveness, has a significant effect on consumers' purchase intention. Moreover, source credibility and perceived quantity of reviews (heuristic factors) have direct impacts on purchase intention. The two heuristic factors further demonstrate positive influences on argument strength. This result is consistent with the proposition of bias effect in the heuristic-systematic model, which elucidates the interrelationship between heuristic and systematic factors.
In 2014, Michael, Loizos and Otterbache also considered reviews of tourist attractions at TripAdvisor (TA). Unlike social networking sites, TA review forums do not facilitate direct interaction between participants. According to this study, the theory suggested that language is guided by writers' conception of their audience, and that their style can shift in response. Therefore, they implemented a model of herding as a local transmission process, exploring the hypothesis that a reviewer is influenced by how preceding reviews manifest a given stylistic feature (e.g., pronouns, paralinguistic devices). They found that reviewers are more likely to use unusual features when such characteristics appear in their local context. The extent to which reviewers are influenced by context is correlated to attributes shared in their profiles, as well as their sentiment toward the attraction reviewed. Results showed that language can be influenced by others, even in an asynchronous environment with little to no interpersonal interaction.
Another study that examined the role of emotions and conflicting online reviews on consumers' purchase Intentions is written by Carla, Kalliopi, Rafael (2018). Drawing on dual-process theories, the study explained how the systematic and heuristic information processing of online reviews with conflicting information can influence consumers' purchase decision making. The study adopted major assumptions of complexity and configuration theory in employing fuzzy-set qualitative comparative analysis on 680 TripAdvisor users to test the complex interrelationships between emotions and the systematic and heuristic cues used in processing reviews. The results showed that the systematic and heuristic processing of online reviews can produce independent impacts on consumer decision-making.
Because customers increasingly rely on other consumers' reviews to make purchase decisions online, Ludwig et al. (2013) drew new insights into the customer review phenomenon through studying the semantic content and style properties of verbatim customer reviews to examine their influence on online retail sites' conversion rates. A dynamic panel data model reveals that the influence of positive affective content on conversion rates is asymmetrical, such that greater increases in positive affective content in customer reviews have a smaller effect on subsequent increases in conversion rate. No such tapering-off effect occurs for changes in negative affective content in reviews. These findings suggested that managers should identify and promote the most influential reviews in a given product category, provide instructions to stimulate reviewers to write powerful reviews, and adapt the style of their own editorial reviews to the relevant product category.
Another important study is written by Bounie et al. (2005) who tackled the effect of online customer reviews on purchasing decisions. The study was applied on the case of video games. The paper tried to assess, first, the impact of online customer reviews on purchasing decisions and, second, the influence of online customer reviews compared to other channels of information such as offline press and WOM. The data were used at the individual level and the electronic gaming industry are analyzed in which online customer reviews play a major role.

Discourse and stylistic features of eWOM
Investigating the discourse and stylistic features of online reviews is an important topic that motivates many researchers to write on this topic. For example, Cenni and Goethals (2017) studied negative hotel reviews on TripAdvisor. It is a cross-linguistic analysis in which they investigated whether English-, Dutch-and Italian-written negative hotel reviews on TripAdvisor show similar or divergent characteristics. The main goal was to find out whether users writing in different languages constitute differentiated speech communities with different discursive norms or rather share the same norms and discourse habits. To answer this question, they examined 100 reviews for each language and analysed three features, namely the types of speech acts that they used, the specific topics that they evaluated and the extent to which they up-scaled or down-scaled their evaluative statements. The main conclusion of the cross-linguistic analysis was that there is a general trend towards similarity between the three language user groups under examination. They found analogous (although not identical) patterns for the three features. Within this overall trend towards similarity, specific divergences can be detected, for example, regarding the status of positive comments in English-written reviews, or the status of the "interpersonal" topic in Italian-written reviews.
Deciphering consumers' sentiment expressions from online reviews has become a managerial priority to monitor product and service evaluations. Using sentiment analysis, Villarroel Ordenes et al. (2017) examined online reviews. Based on Speech Act Theory, he provided a fine-grained analysis of the implicit and explicit language used by consumers to express sentiment in text. An empirical text mining study using more than 45,000 consumer reviews, demonstrated the differential impacts of activation levels (e.g., tentative language), implicit sentiment expressions (e.g., commissive language), and discourse patterns (e.g., incoherence) on overall consumer sentiment (i.e., star ratings). In two follow-up studies, they demonstrated that these speech act features also influence the readers' behavior and are generalizable to other social media contexts, such as Twitter and Facebook. They contributed to research on consumer sentiment analysis by offering a more nuanced understanding of consumer sentiments and their implications.
Another perspective in studying online reviews was presented by Aerts et al. (2017) who examined the influence of prior reviews on what people wrote. They studied the linguistic style of prior reviews to see the effect it leaves on other people's reviews. Results showed that people use more concrete language when prior reviews also use concrete language (i.e. words that refer to tangible, qualities or characteristics), and that this concreteness leads more to more favorable attitudes towards the reviewer and the product.
Another study investigated the impact of customer reviews or influencer reviews on purchase intention. The study was written by Dwidinawati et al. (2020). Since (eWOM) is considered a reliable source of information, therefore customer reviews and influencer reviews can be considered eWOM. They represent customers' sharing of experience and evaluation of a product or service with other potential shoppers. There is abundant evidence concerning the influence of eWOM on purchase intention. A quantitative experimental study (2 × 1) was conducted. Two hundred respondents from three cities in Greater Jakarta were divided into two groups to self-rate their opinion on customer review, influencer review, trust and purchase intention. Data collected was analysed PLS using SmartPLS. Results showed that influencer review has a positive impact on purchase intention. On the other hand, customer review failed to show its influence.

Contributions of the current Work
Having reviewed all the previous related studies and to the knowledge of the researcher, it is found that none of these previous studies tackled the concept of persuasion in online reviews. The current study hypothesizes that online reviews (eWOM) are considered persuasive texts. Accordingly, the study establishes a comprehensive analysis that is based on three different theoretical frameworks in order to identify the different speech acts used and their pragmatic functions. This is in a way to show how these tools used by the reviewers can be persuasive. Moreover, the study also seeks to justify its results by demonstrating the rhetorical tools that emphasize the idea that eWOM are persuasive. The study uncovers all the persuasive tools that can help the reader to take decision. This work contributes to the growing literature on the language of online reviews, offering evidence that reviewers use different methods to make their feedback persuasive. Methodologically, the eclectic framework adopted and the use of corpus linguistics as a tool help to provide reliable results of the present study.

Research questions
The study raises the following questions: (1) What are the factors that affect reviewers' rating?
(2) What is the most frequently used speech act in online reviews?
(3) How are the different speech acts used as persuasive tool for affecting the reader? (4) What are the typical rhetorical strategies used by the reviewers in high-rated and low-rated reviews?

Data collection and methodology
To yield statistically reliable results on frequency, authenticity and trust, a retrospective study is conducted on 100 online reviews that were chosen randomly from two famous social media sites (TripAdvisor and Booking.com). It was based on a personal experience of the researcher during her visit to America in 2018 as she used these two social media sites to choose a good room in a hotel. The data was divided into two documents: 50 low-rated hotels reviews (below 3 star) and 50 high-rated hotels reviews (above 3 star). Repetitive reviews were excluded from the data. Manual annotation was done on both documents identifying the speech act used beside each utterance using<rep> to stand for representatives, <exp> stands for expressives, <com> for commissives, and <dir> for directives. It is to be noted that one review can fit into more than one category. These speech acts are communication acts that convey an intended language functions, i.e. requests, apologies, suggestions, commands, etc. Calculating the frequency of speech acts in each review is made by searching for the tags <rep>, <dir>, <exp>, <com> in the annotated texts ( Figure 1,2). The two corpora were transformed from word documents to a machine-readable format, which is called plain-text file. Antconc is a corpus analysis toolkit used to search for collocates, concordances, and word frequencies and keyword frequency generators. A comparison is made between the results from high-rated reviews and the results from low-rated reviews.
The data analysis went through different phases First, it counts the number of words the collected data of both datasets. Second, an analysis of the different speech acts found in both corpora was provided, showing the pragmatic function of each type. The study provides a primarily qualitative analysis of the online reviews.
Two approaches of analysis of discourse are adopted: Bottom-up versus Top-down. For exploring persuasion in reviews, it is helpful to combine top-down and bottom-up approaches to discourseanalysis. Some tools of corpus linguistics are used in the analysis such as word frequency lists and concordances to assist with the identification of some high-frequency lexico-grammatical items associated with persuasion. In order to see how these features function in context, this is done by highlight patterns of persuasion by including concordance lines. Concordance lines are displays that "bring together many instances of a given word from many sources, allowing us to observe the typical contexts of that word, in particular, the other words that it most usually occurs with" (Channell, 2000, p. 38). Then, moving from a top-down, quantitative approach to a more bottom- up, inductive approach in identifying some common discourse-level forms of persuasion that occur in online reviews. Some of these include features such as rhetorical devices.
First, corpus linguistic tools are used to explore some of the lexico-grammatical features of the collected data. Second, a classification of the speech acts used based on Searle's classification is presented next. Then, pragmatic analysis is also offered to identify the different pragmatic functions of the speech acts used. Finally, Aristotle's persuasion theory is applied to show how speech acts are persuasive tools that affect the reader through investigating some of the rhetorical devices that are used as persuasive tools.

Theoretical framework
The theoretical framework adopted in this study is eclectic for the researcher draws on insights from many approaches and techniques-from corpus linguistics to pragmatics and systemic functional grammar to persuasion theory. Therefore, the study adopts Searle's (1967) Speech act Theory, Aristotle's Rhetoric, 2007, and Vásquez 's framework (Vásquez, 2014). Accordingly, the following sections are divided into three parts. The first part deals with Speech Act Theory, its definition, and types. The second part tackles Aristotle's Rhetoric through shedding light on Aristotle's Persuasion Theory, and showing the different persuasion techniques proposed by him. The final part is devoted to Vásquez 's framework, which in turn is based on Halliday's lexicogrammatical approach.

Speech Act Theory
Speech acts are actions performed via utterance. According to Austin (1962, p. 108), a distinction is made in the theory of speech acts between three different types of act involved in or caused by the utterance of a sentence. They are: locutionary, illocutionary and perlocutionary. Locutionary act is what a person says. Illocutionary act is what a person does such as "ordering, warning or informing". Perlocutionary act is what a person wants to achieve by performing a speech act. The study focuses on the third type, that is, perlocutionary act. Perlocutionary act in online reviews persuades the reader to take some kind of action. Austin (1962) states that speech acts categorized depending on the characteristics of verbs and their illocutionary forces. Accordingly, he groups illocutionary acts into five sorts, despite the fact that such classification appears to be difficult to do or to comprehend, and in many cases the speaker's intentions are ambiguous. These types are presented as follows (Austin, 1962, pp. 150-151): (i) Verdicatives: This type of speech acts is used to express verdicts, judgments or findings such as to appreciate, characterize, assess, evaluate, assess, grade, etc. Eg.: I accuse you of being a dishonest journalist.
(ii) Exercitives: This type of speech acts shows the use of powers, rights or influence such as naming, sentencing, appointing, ordering, dedicating, dismissing, claiming, etc. E.g.: I dedicate this song to my husband.
(iii) Commissives: The speech act of this type implies commitment or promises of distinctive types or the acceptance of an promise or the expression of a future plan. The verbs belonging to this classification are promises, plans, guarantees, swearing in, betting, etc. They obligate the speaker to doing something in the future. E.g.: I promise to come to your party tomorrow.
(iv) Behabitives: The verbs for this class of speech acts comprise expressions of attitude and social behavior such as apologies, congratulations, compliments, welcomes, etc. E.g.: I congratulate you on your success.
(v) Expositives: According to Austin (1962, p. 152), this classification is not easily defined because the verbs of such type are not clear cut. Sometimes, they overlap with other classes, but the typical performative nature of individual utterances is rather clear. Verbs of such type are hypothesize, expect, assume, remark, concede, etc. E.g.: He's not ready to concede the election.
For Searle (1979, p. 1), the main question concerning studying a language is "how many categories of illocutionary act are there?". Searle (1979) classifies speech acts into five types: assertives (representative), directives, commissives, expressives and declaratives. Cruse (2000, p. 342) states that "assertives commit the speaker to the truth of the expressed proposition".The verbs belonging to this class are: assert, suggest, complain, report, state, conclude, claim etc.

Directives
This speech act has "the intention of eliciting some sort of action on the part of the hearer" (Cruse, 2000, p. 342). Thus, directives can either be direct utterances as ordering someone or they can be indirect as suggesting something to get the hearer to act. Verbs that belong to this type of speech act is order, command, request, beg, ask (to), etc.

Commissives
This speech act "commits the speaker to some future action" (Cruse, 2000, p. 342). This type of speech acts shows the intention of the speaker such as vows, threats, pledges, warranties, contracts, promises, agreements, and oaths. Searle (1979, p. 10) points out that both directives and commissives consist of future actions, however; with directives it is the hearer who acts, while with commissives it is the speaker who acts.

Expressives
According to Cruse (2000, p. 342), this act "makes known the speaker's psychological attitude to a presupposed state of affairs". It reflects a true wish or feeling about a condition or situation such as apologizing, thanking, congratulating someone. Cruse (2000, p. 343) states that declaratives "are said to bring about a change in reality: that is to say, the world is in some way no longer the same after they have been said". For example, if someone says I resign, then thereafter he is no longer hold the post he originally held. (ibid, p. 343).

Declaratives
It is to be noted that the study adopts Searle's classification of acts for it is broader, more comprehensive than Austin's. The first type of acts, according to Searle (1979) is representatives or assertives. By assertives, Searle means claims or reports. The second type is directives which represent suggestions, requests and commands. The third type is expressives which is used to refer to apologies, complaints or thanks. The fourth type is commissives which denote promises, commitments and offers and finally declarations which indicate announcements. After presenting the different types of speech acts, it is important also to mention that speech acts can be classified into direct and indirect speech acts.

Aristotle's Rhetoric
According to Morris (1938), pragmatics deals with the function of linguistic utterances and the propositions that are expressed by them, depending upon their use in a specific situation. It is noted that speech acts and rhetorics share one important characteristic which is the intentionality of the speaker. In order to motivate the reader to do something, the speaker uses speech acts and rhetorics to achieve the intended action. Austin's concept of a perlocutionary act can be linked with the traditional concepts of rhetoric as a technique of persuasion. Therefore, the study investigates some of the rhetorical devices used by the reviewers in their reviews which have the effect of persuasion.
Aristotle introduces three pillars of persuasion: ethos, pathos, logos. Ethos refers to persuasion by the credibility (or character) of the speaker, speaker establishes himself as an ethical person. Pathos refers to emotional connection to the audience, speaker uses persuasion by appealing to audience's emotions. Logos refers to persuasion by speech itself in which the speaker appeals to audience's sense of reasoning or logical argument.
In his rhetoric (l.Ii3-6), Aristotle explains that the "proof through character" depends on confidence, which "must be due to the speech itself, not to any preconceived idea of the speaker's character". He accounts for the "proof through emotion" by pointing out that "the judgements we deliver are not the same when we are influenced by joy or sorrow, love or hate".
Aristotle (I.Ii.7) states that there are different persuasive techniques that can appeal to the audience. These can involve "emotions, fears, desire to seem intelligent, need to protect others, desire to fit in, to be accepted, to be loved". Moreover, evidence is very persuasive as it makes the reader see the author as knowledgeable and the argument as more logical or reliable, for example, an expert opinion. Another persuasive technique is attacks as people who hold them can persuade the audience by representing views and beliefs which are opposing to the author's opinion as foolish, dangerous, uncaring or deceitful. He states that using humor to make fun of these views can be particularly persuasive. Inclusive and Exclusive language is also included under persuasive techniques. Inclusive language, on the one hand, embraces 1 st person pronouns such as "we", "our", "us" and exclusive language, on the other hand, includes such as "them" can persuade by including the reader, or by creating a sense of solidarity or a sense of responsibility. Rhetorical questions are also persuasive techniques which depends on questions that do not require an answer and are asked for effect only. They involve the audience and encourage them to think about the issue and accept the author's answer, or indicate that the answer is so apparent that anyone who disagrees is foolish. Another persuasive techniques are cause and effect, and connotations. The latter is the emotional meaning related to the word. Persuasive authors usually choose their words carefully so that the connotation meets their purpose. Other techniques are analogy, generalizations, colloquial language and repetition. Analogy is a comparison between two things in order to highlight the differences between them. Generalization is making sweeping statements about a whole group, based on only one or two members of that group. These can be persuasive if the audience believes the generalization is appropriate, yet it can weaken argument if they do not. Colloquial language is informal, everyday, conversational language that includes plain, straightforward, and tempting views that appears friendly, and can make the audience feel that the author is on the same wavelength as them. Finally, repetition is made by repeating letters, words and phrases that the author feels they can reinforce an argument and guarantee that the his opinion stay in an audience's mind. Vásquez's (2014), in his work, examines online reviews from a discourse perspective using different approaches and techniques: corpus linguistics to identify common lexico-grammatical resources used by reviewers across the dataset to assess their experiences, pragmatics, and narrative analysis.

Lexico-grammatical approach
Lexicogrammar is a term used in systemic functional linguistics to emphasize the interdependence of and continuity between vocabulary (lexis) and syntax (grammar). The term lexicogrammar was introduced by linguist M.A.K. Halliday, that is why Vásquez (2014) bases his analysis on Halliday's lexico-grammatical approach. According to Pearce (2007), the advent of corpus linguistics has made the identification of lexicogrammatical patterns easier than it once was. Accordingly, Vásquez examined lexicogrammatical patterns using corpus-based approach. Halliday (1994) defines three language functions, they are; (i) the ideational function (expressing ideas about the real world-associated with logos); (ii) interpersonal function (concerned with social relationships-associated with both ethos and pathos); and (iii) textual function (distinguishing a text with a living message from a dictionary entry-also associated with ethos). Persuasive language exhibits all these functions, and investigating rhetorical techniques and structure as adopted from Aristotle is of a great benefit to this study if bearing in mind Halliday's definitions. They are means of distinguishing between different kinds of persuasive discourse. It is to be noted that the proposed approach is selective rather than comprehensive. The aim of this study is to offer some insights into the range of persuasive devices commonly used by reviewers as well as into some of the more creative resources used by only some reviewers.
Having mentioned these three theoretical frameworks, it is worth mentioning that the study adopts an eclectic framework that is taken from the abovementioned theories. As stated before, both speech acts and Aristotle's rhetorics share one important characteristic, that is, the intentionality of the speaker. That is why the study uses Searle's Speech Act Theory and Aristotle's Rhetorics in order to study the factors used by the reviewers to motivate the reader to do something. In addition, the study identifies the common lexico-grammatical resources used by reviewers across the dataset to assess their experiences and pragmatics in order to provide a comprehensive study on online reviews and show how they are persuasive. This is done by adopting Vásquez's framework (Vásquez, 2014) which is based on different approaches and techniques. This includes Halliday's Systemic Functional Grammar and Corpus linguistics as an important tool for investigation. Based on these three approaches, the following section is devoted to the data analysis, results of the current study and the findings that are spotted and noted.

Statistical Analysis of the collected data
Using the Readability Statistics in Microsoft Word, different outcomes are provided: Number of words within each set of reviews as well as lexical density and the number of sentences in the two datasets.
Results of this statistical analysis of both dataset reveals that positive reviews contain more words than negative reviews as the number of sentences of positive reviews are 899 as opposed to 594 sentences in negative reviews. This indicates that reviewers write more about their positive feelings when they like a product or service.

Common lexico-grammatical resources used for persuasion
Using Antconc software, a top-down approach is made to display the most frequent words in both dataset. The following Figures 3,4 shows a comparison of the top 50 most frequent words in the dataset of 100 reviews in order to spot the top 50 nouns in both dataset.
A side-by-side comparison of the 50 most frequent words in both datasets shows which words are more specific to persuasion. As might be expected, because the aim of these two types of reviews is either to persuade, there are more similarities than differences between the two datasets. For example, function words such as pronouns (i, you, it), articles (the, a), conjunction (and) and prepositions (to, of, for, on) appear among the most frequent 20 words in both lists. In addition, the first-person (I, we, my, our, me) as well as second-person (you) pronouns are found in both lists. The frequency of these forms in online reviews shows an orientation to both participants in the discourse, the reviewer and the readers. First-person references are used to explore reviewers identities and second-person references are examined as an involvement strategy.
Moreover, the use of prepositions (e.g.: of, for, on, at) are very common in both lists. Prepositions typically co-occur with nouns, functioning together to describe and refer to the "world out there" (O'Keefe et al., 2007, p. 33), which is predictable in this context, considering that reviewers often combine expressive and representative speech acts as he starts by his judgment of evaluation followed by evidence represented in a description of the place and talking about his experience.  In addition, it is obvious that the two lists include higher frequencies of a handful words that are strongly associated with persuasion in the online reviews list, including negator no, evaluative adjective good, intensifying adverb very, stance adverb just and the mutli-class, multi-functional word like. Persuasion can be expressed by using a wide range of linguistic devices. However, one of the most common and explicit devices for expressing persuasion are evaluative adjectives (Hunston & Sinclair, 2000).

Top frequent nouns
After this broad view about the top frequent words in both datasets, the following Table 2 shows a comparison of the top 5 most frequent nouns in the dataset of 100 reviews.
The table above shows that room, hotel, staff, location are the most frequent nouns used in both types of reviews. The nouns "hotel" and "room" appear among top 50 words in the reviews list. Reviewers use these context-specific nouns as they offer descriptions and evaluations of these aspects of their experience. The word "room" is repeated 182 times in high-rated reviews as opposed to 145 times in low-rated reviews. Similarly, the word "hotel" is repeated 130 times in high-rated reviews, whereas in low-rated reviews it is repeated 66 times only. The word also "staff" is repeated more in high-rated reviews than in low rated reviews. On the contrary, the word "location" is repeated more in low-rated reviews than in high-rated reviews. Finally, the word "service" is one of the top frequent nouns in high-rated reviews, however, it seems that it is not important to low-rated reviews that is why it is not found among the top 5 nouns in the wordlist. This can be attributed to the fact that in high-rated hotels service is essential, whereas in low-rated hotels the priorities are different.

Top frequent adjectives
According to Hunston and Sinclair (2000), one of the most common and explicit devices for expressing persuasion is by using evaluative adjectives. Due to the significances of adjectives in reviews, it is to be noted that an examination of the most frequent adjectives in both datasets demonstrate the type of adjectives used in both reviews.
The Table 3 above shows that there are some common adjectives used in both high-rated and low-rated reviews such as good, friendly, comfortable. However, it is noted that most of the evaluative adjectives in low-rated reviews are accompanied with the negator "no", which highlights the absence of certain amenities in the hotel. Valence shifters are related to sentiment analysis. Negator (not) and the concessive conjunction (but) are quite common in negative reviews ( Figure 5). Using "but" can be very effective in highlighting the weak points of the hotel. The  adjective "good" is generally associated with positive sentiment and as such, it is used frequently to express positive evaluation which can have an effect on the reader.

Speech acts in online reviews
Results show that there are a variety of usage of speech acts both in high and low-rated reviews.
The following tables 4, 5 represent the frequency of speech acts and their percentages in both datasets.
Results show that representatives are the most frequently used speech act in high-rated reviews with percentage 77%, followed by expressives with 20 %, then commissives with 1.8% and Figure 5. The concordances of "not" in low-rated reviews.  According to the above table, it is found that representatives are the most frequently used speech act in low-rated reviews with percentage 79%, followed by expressives with 17 %, then directives with 2.4% and commissives with 0.8%, whereas there are no declaratives. Thus, in both datasets, the two prevalent types of speech acts are representatives and expressives, while the least acts are commissives and directives. This can be attributed to the fact that the reviewers' objectives are to give positive or negative statements about the product/service. They express their feelings whether approval or disapproval about the product or service, that's why expressives are used frequently. Moreover, in high-rated reviews, more commissives and less directives are found. Reviewers feel compelled to share their experiences with other customers and by using commissives can have a positive effect that can encourage other consumers to purchase the hotel, rather than directly instructing them to take an action and book a certain hotel. On the contrary, more directives and less commissives are found in low-rated reviews.Out of their bad experience they feel compelled to warn other consumers. They want to steer consumers away from making a purchase they'll later regret.

Pragmatic functions of speech acts in online reviews
Evidently, different pragmatic functions of the speech acts used in online reviews are found in this study (Table 6), e.g., thanking, asserting, warning, advising, complaining, praising, persuasion, suggestion, information, evaluation, wishes.
For each speech act, there are a number of pragmatic functions that are performed in the online reviews. Representatives are used to describe, inform, assert and suggest. Expressives are used to

Representatives
They are simple statements or claims about a certain state of affair. According to Searle, the direction of fit in representative speech acts is words to the world (Searle, 1979, p. 12). Reviewers are committed to say the truth about a product or service. The pragmatic functions that are included under representatives in online reviews are: describing, informing, asserting, suggesting.

Describing.
Describing is a discourse intended to give a mental image of something experienced (Merriam-Webster's Collegiate Dictionary, Eleventh Ed., 2003, p. 337). Many of the eWOM include description of the hotel, room, location, staff, service and room. By reviewing all the examples that fall under description, and identifying some of the common discourse-level forms of description that occur in the online reviews under investigation, there are some features found such as the occurrence of existential sentence, equative clauses and low transitivity clauses. In English grammar, an existential sentence is a sentence that asserts the existence or nonexistence of something. For this purpose, English relies on constructions introduced by "there" (Figure 6). By checking the word "there" in the corpus data and its concordances it is found that there are many constructions that involve the word there plus verb "to be" with all its forms (is, are, was, were) in high-rated reviews, and similarly in lowrated reviews, however, the difference is that the verb to be is followed by a negation to assert the nonexistence of things.

Examples
(1) There was a weird step between bathroom and bedroom.
(2) There was no heating in the room.
(3) There are only 3 lifts servicing 32 floors The above examples show that the reviewers attempt to draw a picture of the hotel what is available and what is not. They give a descriptive report details so that the reader can imagine the place where he is going to stay so that they can build their own perception about the place. The category of description includes references to accommodation (under which references to the appearance of the hotel, the rooms or some specific features such as bed, bathroom, shower, furniture, walls, etc.). A reference to service also included under description involves the cleanliness of the hotel and rooms, the quality of the food and breakfast, and the safety of the hotel. Moreover, there are references to location, price and interpersonal communications.
Unlike, low-rated hotel reviews, most of the examples that are found are made up of the construction there + verb to be followed by "no". The following Figure 7 is an extract taken from the corpus tool Antconc in which results taken from low-rated hotel reviews.
(3) There was no sign of any kind of fan.
(4) There was no towels in my room.
Most of the description found in low-rated hotel reviews are negative descriptions of the appearance and functionality of the hotel, the service or the interpersonal communication. However, it is found that some positive descriptions are included in the low-rated hotel reviews concerning the good location of the hotel near certain attractions or near city center. Positive description can also be attributed to the price of the hotel as most of the low-rated hotels have a low price in comparison to high-rated hotels. However, this construction is not the only one found in the reviews as there are a number of phrases that are also included under description. In these reviews, there is an omission of the existential word "there" and the following copula. Only the negation word "no" followed by the object (or noun). The presence of "valence shifters" which reverse the semantic polarity of an expression (Choi & Cardie, 2009;Jo & Oh, 2011). Valence shifters change the evaluative meaning of the rest of an expression. The most obvious of these are negative and include words such as not, never, none, nobody, nowhere, nothing, neither as well as but, however, and so on. (In Vásquez, 2014, p. 27). These expressions are found extensively in low-rated reviews which in turn have a direct effect on the consumer's decision-making. By searching for the concordances of the negation word "no", the following results are found ( Figure 8):
(3) No chairs in reception as i can't stand. (4) No hand soap in dispenser in bathroom.
(5) No hot water in the room All the previous examples show the absences of the amenities that should be found in the hotel. Beside describing the hotel and its appearance, there is another category that can be classified under description, that is, "extra information" as stated by Cenni and Goethals (2017). This category includes various kinds of informative sentences (when, where and with whom the trip takes place and for how long it is) as well as a vivid picture of the hotel, room, etc.

Example
(1) "The rooms décor was quite dated with old stains on carpet and ceiling. The walls to our room seemed thin as we could quite clearly hear the room next door chatting, coughing and drying their hair and returning late".
It is noted that more informative sentences are found in high-rated reviews as opposed to less informative sentences in low-rated reviews. This can be attributed to the fact that the reviewer wants to influence the reader through extra information to persuade him to purchase a certain product/service. That is why most of the high-rated reviews are full of details and descriptions which give the reviews authenticity as opposed to fake reviews that may contain certain clichés or same patterns. However, in low-rated reviews, reviewers focus on facts, especially the absences of certain amenities. They do not need to give extra information to the reader. They direct the reader to refuse this purchase through providing him with what is present and what is absent in the hotel.

Informing.
Informing is the act of giving or imparting knowledge of a fact or circumstance. Online reviews have become a significant information source that allow consumers to look for thorough and reliable information by sharing past intake experiences. The useful information in a review may assist customers to evaluate the attributes of the service so as to build confidence in the source. By examining the two datasets, it is found that a large number of reviews that fall under the category "representative" provide relevant information about the hotel namely; its location, its service, the hospitality of the staff, true incidents and details happened during the reviewers' stay at the hotel. By checking the concordances of the words: location, service, staff in high-rated reviews, the following information is provided: Example from high-rated reviews (1) The hotel is in a good location around the corner from the iconic Macy's (2) Andrew was awesome with fast service with check-in Examples from low-rated reviews (1) Booking's customer service had not been helpful (2) staff did not seem to communicate with each other All the examples above, provide a significant information source that allow consumers to find out comprehensive and reliable information about the hotel and its facilities.

Asserting.
Asserting is the act of stating clearly and firmly that something is true ("Oxford Advanced Learner's Dictionary," 2010, p. 74). All the examples found in the corpus that fall under asserting have the meaning of proclaiming that the hotel is deceiving the consumer as it promotes certain facilities, but on reality these facilities are not available.

Example
(1) "You enter the hotel, it smells of pee . . . Both rooms on the 12 the floor were awful. What they advertise on their websites was no way near the real rooms".
By checking the concordances of the phrase "as advertised", it is found that such asserting sentences are found more in low-rated reviews than in high-rated reviews.

Examples
(1) Bed was hard, not very clean overall, smell of damp or cannabis in hallway, not cable TV as advertised, no maid service as advertised, TV remote didn't work properly.
(2) Toilet and spit into the sink. Not as advertised.
(3) . . . As we just wanted to eat decent fare as advertised.
The examples above express the reviewers' assertion that these amenities are not found as advertised. Figure 9. The concordances of the word "need" in high-rated reviews. Figure 11. The concordances of the word "need". Figure 10. The concordances of the word "could" in high-rated reviews. Figure 12. The concordances of the word "could" in low-rated reviews.

Suggesting.
Suggesting is the act of mentioning an idea, possible plan or action for other people to consider. The following Figure 9 screen shot is taken from the concordances of the verb "need" as found in high-rated reviews.

Example
(1) Some rooms need upgrading

Example
(1) There could be some improvement on cleanliness ( Figure 10) It is also found that there are suggesting sentences in low-rated reviews. The following Figure 11 screen shot is taken from the concordance of the word "need".

Examples
(1) Hotel in need of a complete refit (2) Waterside restaurant staff need to be more attentive.
Checking also the modal auxiliary verb "could" as found in the following Figure 12 screen shot, it is found that there are a number of suggestions offered by the reviewers to the hotel such as:

Examples
(1) The room could have been cleaner-there were roughly 20 flies (2) There could be some improvement on cleanliness.
(3) Whole place could do with refurbishment.

Examples
(1) Hotel in need of a complete refit (Figure 13). (2) It is very old-fashioned and in desperate need of updating bathrooms Therefore, it can be said that through examining modals one can deduce one of the pragmatic functions of the representatives namely; suggesting through checking modal verbs.

Expressives
They are speech acts that express the speaker's feelings about themselves or the world (Searle, 1976, p. 12).The illocutionary point of expressive is to express the psychological state specified in the sincerity condition about a state of affairs specified in the propositional content. In expressive there is no detection of fit. In performing an expressive the speaker is neither trying to get the world match the words nor the words match the world, rather the truth of the expressed proposition is presupposed (Searle, 1979, p. 15). The expressive tags <exp> were searched with the help of the concordance software Antconc. A number of functions are found in the corpus data that are included under expressives. They are: praising, complaining, thanking, regretting, and disliking.

Praising.
It is the act of expressing the approval or admiration of somebody or something ("Oxford Advanced Learner's Dictionary," 2010, p. 1149). According to the examples found in the corpus, reviewers express their high degree of approval and admiration of the appearance of the hotel, its location, the attitude of the staff and the service in the hotel. "Evaluation can be expressed using a seemingly infinite range of orthographic (or phonological), lexical, syntactic and discourse-level resources" (p. 25).
Praising is positive evaluation, and according to Vásquez (2014), evaluation include adjectives. By examining the corpus data of high-rated reviews, it is found that all the frequent nouns have been praised by the reviewers by different positive adjectives such as: 'nice hotel, great hotel, hotel clean and welcoming, awesome hotel, comfortable hotel, fantastic hotel, amazingly friendly staff, the nicest service, friendly staff, staff interaction overall was good, location amazing, lovely location, great facilities and comfortable room, good design of the room, etc. Few of the examples found are direct speech acts, while the majority are indirect speech acts as they are declarative sentences that do not contain performative verbs to indicate praising. However, the meaning is understood indirectly. By checking the concordances of the word "hotel" many examples of expressive speech act are found such as:

Examples
(1) I am very pleased with this hotel.
(2) Very pleased with this hotel.
(4) Great hotel and great service.
(6) Great hotel for travelling around NY.
By reviewing low-rated reviews and by checking the word "hotel" there was no praising at all attributed to the hotel. All the examples found are complaints about the hotel.
Looking for the concordances of the word "service", only two examples are found fall under the category "praising", such as:

Examples
(1) Restaurant service also good, very polite and efficient staff, reasonable location, I like the location, perfect location for us, very convenient location for our purpose.

Complaining.
It is the act of saying that you are annoyed, unhappy or not satisfied about something or somebody ("Oxford Advanced Learner's Dictionary," 2010, p. 293). The reviewer expresses his displeasure or annoyance as a reaction to his bad experience at the hotel, and consequently how this affected him negatively. It is found that in high-rated reviews, very few examples of complaints are found. Some of them are direct speech act such as:

Examples
(1) Bed was rubbish Other examples are indirect speech acts in which they are representatives in which the reviewer suggests that the hotel could make some changes to look modernized, yet it could be expressive speech act as the reviewer expresses his complaint about the appearance of the hotel.

Example
(1) Hotel in general could do with some modernization.
On the contrary, the majority of complaints are found in low-rated reviews.

Examples
(1) One of the worst hotel experience ever.
(2) The area around the hotel is also a bit noisy.
(3) It was the worst hotel room i have ever had in central London (4) The restaurant was awful with staff not caring about what they were doing.
(5) The room is like a cell however we got up.
(6) Lightning in the room very bad.

Thanking.
It is the act of expressing gratitude or more often the speaker's or writer's pleasure or satisfaction to someone for doing something (Merriam-Webster's Collegiate Dictionary, 2003, p. 1294). By checking the word "thank" in the corpus data, the following results were found. In high-rated reviews, more examples of thanking were found than in low-rated reviews.

Examples
(1) Thank you for hospitality.
In low-rated reviews, there are few examples such as:

Examples
(1) . . . due to the guy in the launderette. Thank you.
(2) . . . Location and great value! Thanks! 6. 4.2.4. Regretting. It is a feeling of sadness about something sad or wrong or about a mistake that you have made, and a wish that it could have been different and better (Cambridge Dictionary.org). By checking (<exp>) in Antconc software, it is found that there are a number of expressions of regret in high-rated reviews such as:
(2) But either that was not worth the price.
In low-rated reviews, more examples of regretting and disappointing are found than in highrated reviews.

Examples
(1) Definite steal! It is what it is. Bed was hard . . .
(3) We just want to get out of there.
(4) Definitely less than a 1 star rating.
(5) Nothing good. I deeply regret booking that hotel.
(6) Thanking, regretting and disappointing 6.4.2.5. Disappointing. It is a feeling of being not as good as you had hoped or expected; not satisfactory. (Cambridge Dictionary.org). By checking the words "disappointing", "disappointment" and "disappointed" in the corpus data of high-rated hotels, few examples of disappointing are found.

Examples
(1) Very disappointing as took the great atmosphere away by allowing . . .
(4) Quite disappointing for such a high class hotel.
However, not all the examples are direct speech act containing the word "disappointing", there are also indirect speech acts which have the meaning of being disappointed such as:

Examples
(1) Not worth the money.
(2) Hit and miss I suppose.
(5) What a pain! Although it is well-known for those who choose low-rated hotels that they may be poor in their amenities, yet it is found that there are many examples of disappointing about the hotel and its service and facilities.

Commissives
It is a speech act that commits the speaker to perform some future course of action by making a statement or a wish to do something. According to Searle, commissives consist of future actions and it is the speaker who performs the action. Two pragmatic functions are found related to this type of act, namely; wish and promise. By checking the pronoun "I" to search for examples of commissive speech act, the following examples are found:

Examples
(1) I am returning again next year.
(2) I hope i have cause to stay there again.
(3) I think next time I'd try the standard . . .
(4) I will not be returning.
(5) I will not come back again Looking for the concordances of the modal auxiliary verb "would", the following examples are found:

Examples
(1) Would consider staying here again.
Looking for the concordances of the modal auxiliary verb "will" as it refers to future actions also, the following examples are found:

Examples
(1) Will definitely stay here again!
(2) Really enjoyed our stay in lion quays will definitely visit again.
(3) Was a nice stay but will look elsewhere for next visit.
Swapping with target files, the researcher checked for the pronoun "I", and the modal auxiliary verbs "would" and "will", most of the examples are followed by no, not, never as the reviewer doesn't wish to repeat this experience again. The following examples are taken from low-rated reviews:

Example
(1) I wouldn't like to live there! (2) I'll never ever return to this horrible . . .

(3) Would not stay again
Commissives are used in high-rated reviews more than in low-rated reviews as the reviewer wishes to share his future plans with the reader based on his satisfaction with the facilities of the hotel. Furthermore, there are some less predictable linguistic features associated with (commissives) in this genre, such as the spatio-temporal adverbs back and again, which are used to express the desire to repeat an experience.

Directives
It is attempts made by the speaker to get the hearer to do something such as ordering, requesting, etc. They express what a person wants or desires. Two pragmatic functions are related to directives; advising and warning. Before going through both functions, it is important to mention the difference between the two functions as mentioned by Vandervken (1990) who states that to advise is like to warn, except that the effect of what is advised is good for the hearer.

Advising
It is the act of telling somebody what you think they should do in a particular situation ("Oxford Advances Learner's Dictionary," 2010, p. 22). By checking the examples of directives that function as advising, it is found that the more advise are found in high-rated reviews than low-rated reviews. Since the hearer is involved in this type of speech act, it is necessary to check the pronoun "you" with its concordances to see how the hearer represented by the pronoun "you" is involved.

Examples
(1) Overall it's great value if you are looking for a well located, clean & safe hotel . . .
(2) We will never stay there again but if you do, do not trust their staff.
Checking the verb "recommend", the following examples are found:
(2) Would recommend (3) Highly recommend this hotel!!!! By checking the same words "you" and "recommend" in low-rated reviews, the following examples are found.

Examples
(1) Stay here only if you want to save money.
(2) Don't expect too much and you won't be disappointed.
(3) Would not recommend (4) Would never recommend this place to anyone

Warning
It is a statement or piece of information that makes people understand that there is a possible danger or problem in the future (Cambridge Dictionary.org). The majority of warning expression are found in low-rated hotels as opposed to few examples found in high-rated reviews.
In high-rated reviews, the following examples are found:

Examples
(1) Just be aware of this.
To check this pragmatic function in the corpus data of high-rated reviews, the researcher chooses the word "avoid" to see its concordances.
(3) If you do, you will likely want to avoid this hotel.
Checking also the auxiliary verb "do not", and its contraction "don't" as it indicates a warning, the following example is found:

Example
(1) Never stay there again but if you do, do not trust their staff.
In low-rated reviews, there is a direct warning that is found e as the reviewer said "be warned . . . ". By checking the same words as in high-rated reviews more warnings are found than in high-rated ones.

Example
(1) Avoid at all costs.
Examples that contain the auxiliary verb "do not" are: (1) Do not book this hotel.
It is to be noted that these findings of the different pragmatic functions of speech acts found in online reviews are in line with previous studies (Hennig Thurau & Vásquez, 2014, p. 44) that state that the motivations that reviewers may have while posting online reviews are different as some reviewers wish to provide advice for others without any apparent benefits for them. Others wish to help the company often as a result of the satisfaction of the consumer with the company's product or service. Another reason is that some consumers long for participation in and be members of an online community. Another motivation is the convenience of complaining as it is important for consumer's decision to know where to complain. Moreover, expressing negative feelings and venting positive emotions are two more motivations for writing reviews. Writing a review about a negative experience can help in reducing consumer's feeling of dissatisfaction. Similarly, very strong positive experiences may lead to feelings of imbalance, balance can be restored by writing and posting positive comments. Therefore, all the above findings meet these motivations and support them. Following Vásquez's (2014) framework, he proposes a schematic patterns of speech acts. Position in reviews is important, because opening or closing sentences often function as summaries of the entire review and accordingly, "comments at the very beginning or very end of a review are accorded more weight than remarks in less prominent positions" (Polanyi andZaenen 2004, p. 9). (In Vásquez, 2011, p. 37). Examining the reviews on Booking.com and TripAdvisor, it is found that they are made up of a title, a body and a conclusion. The whole review is a combination of different speech acts. All these speech acts intersect in a single review text; thus, the whole review can be very persuasive to other consumers. There are different schematic patterns founds in the corpus data.

Schematic patterns of speech acts
The researcher finds out that some reviewers choose to present their experience as fully formed with highly elaboration and a strong chronological organization, and offer more impersonal descriptions, and, then, end with his judgment followed by either a commitment to come back or an advice or warning to the reader to avoid this hotel or stay at this hotel. In other words, the reviewer starts by representative speech act describing his experience at the hotel with some details about the appearance of the hotel and its service. Then, he moves to expressive speech act in which he expresses his evaluation about the hotel whether by praising or by complaining. Finally, he ends either by commissive speech act by expressing his pleasure to repeat the experience at the hotel again or by directive speech act in which he directs his speech to the reader recommending the hotel to him.

"Comfortable hotel in a good location"
The staff were a friendly and helpful. The breakfast was very good with a wide range of food options. The location on the A5 made it a very convenient stopover place ( Figure 14).

Figure 14. A schematic pattern of a high-rated review.
Another example of low-rated review that is made up of a title that is expressive, body that is representative and a conclusion that is commissive is as follows.
Example (2) 6.7.2. "Not safe" I was robbed at my room early in the morning. And according to the police the circumstances indicate that there might have been an accomplice at the hotel. I do not recommend staying at that accommodation. Criminals can enter without interference. I would never stay there again.
Other schematic pattern is made up of an expressive speech act in which the reviewer expresses his feeling of dissatisfaction followed by representatives as a justification of his judgment, and, then, ends by commissive or directive speech act, either committing himself not to come back or to warn other consumers not to try this hotel ( Figure 15).

"Very clean, comfortable beds, lovely shower"
Only complaint was the doors to the balcony in our room not closing properly.
Very clean, comfortable beds, lovely shower. Very helpful staff. Couldn't fault anything apart from the doors leading to balcony in our room which were in need of renewing as they would not close properly and were s bit weather worn, Apart from this we had no complaints. Will definitely stay again.
Another schematic pattern is made up of a title that is directive in which the reviewer sends a direct message to the reader warning him not to choose this hotel. Then, he starts to inform the reader by details that support his view about the hotel and finally ends by expressing his feelings of disappointment ( Figure 16).
Example (4) 6.7.4. "If you can sleep on a bench in a park instead of this hotel, then sleep on the bench." The shared bathrooms were disgusting. The private bathrooms were no better. The overall cleanliness was very poor. There was no elevator, stairs were very high. Air conditioning was a fan, no heating system. Wifi was very poor, often disconnecting.
In some reviews, the reviewer chooses a catchy title that combines two speech acts; expressive and directive as he praises the hotel and addresses the reader to enjoy his stay in the hotel. Then, he describes the hotel trying to draw an image to the reader about his reason for loving this hotel in particular, and, finally, ends by an advice recommending the hotel to the reader ( Figure 17).
Example (5) 6.7.5. "Overall I think this hotel is a very good place to stay if one wants to enjoy Manahttan." I liked the location. It was only two blocks away from one of the main subway lines (1) which made getting to places I wanted to visit very easy. Definitely, a hotel to recommend ( Figure 18).
Example (6) 6.7.6. "Didn't actually enjoy my stay" It has no window blind. The air conditioner was good. The female receptionist that processed our payment was harsh and unfriendly. The room/bed is too small for my family, despite stating the number in my booking. No breakfast or coffee

Persuasive speech acts
In addition to taking a top-down approach and using frequency to guide the identification of recurring lexico-grammatical patterns of persuasion in online reviews, the researcher uses an inductive approach to assist in the identification of a number of discourse-level persuasive strategies. Based on Aristotle's persuasive appeals; pathos, logos and ethos, it is noted that these persuasive appeals together increase the chance of being a persuasive speaker. By examining these persuasive strategies, it is noted that reviewers rely on pathos through using expressive  speech acts to appeal to consumer's emotions and logos through using representative speech acts to appeal to the consumer's sense of logic through facts and evidences which in turn have a direct impact on consumer's decision-making. However, they do not rely on ethos because it is not applicable as the writer of a review is anonymous because of the lack of knowledge about the speaker's character. Persuasive language is used to convince others to agree with our facts, share our values, accept our argument and conclusions, and adopt our way of thinking. There are many different ways to persuade people. Beside these three appeals, there are some rhetorical devices of persuasion that are obviously seen in the online reviews, namely; rhetorical devices.

Rhetorical devices
Different techniques are included under rhetorical devices that help in creating a sense of involvement for the reader of the review. Before displaying such techniques, it essential to this study to define the term "involvement".
6.8.1.1. Involvement. Involvement is another technique of rhetorical devices. Several attempts have been made to define the term "involvement" (Bensier, 1994;Chafe, 1982;Gumperz, ;Lakoff, 1990;Norwood & Deborah, 1989;Tannen, 1984Tannen, , 1988. One common among those scholars is that the term is generally associated with the various means through which a speaker or writer interacts with his or her audience. It is an aspect of communication that refers to the engagement of the speaker and the listener. In this study, the involvement occurs between the authors of online reviews and the readers of those reviews. Historically, involvement has been associated more with speech than with writing. Recently, it is carried out in asynchronous computer-mediated communication. Forms of CMC are often characterized as "hybrid", in which they comprise linguistic features associated with both writing and speaking (Georgakopoulou, 2006;Herring, 1996c). To analyze the data, the researcher identified and coded the most common ways that writers conveyed involvement. Some of the most frequent involvement strategies that are found are: • Rhetorical questions • Declarative present-tense statements • Second-person address • The use of conventional speech acts This is besides various "non-linguistic cues", such facial expressions, gestures, and "intonation" (Bensier, 1994, p. 281), and emphatic punctuation such as (e.g., multiple exclamation marks, capitalization). There are many other ways in which involvement can be signaled, however, the focus is on those involvement strategies that are the most clearly identifiable in the corpora under investigation.
6.8.1.2. Rhetorical questions. Moving from a top-down, quantitative approach to a more bottomup, inductive approach in identifying some common discourse-level forms of persuasion that occur in online reviews. Rhetorical question is one device in which a question is asked with no answer is needed. It creates a sense of involvement for readers of reviews as it influences the reader to arrive at the desired result. It has been noted that in some online genres, rhetorical questions seem to be one of skepticism (not believing). Examples: (1) What happened to hospitality?
(2) This is rated 5 stars? No way! 6.8.1.3. Declarative present-tense statements. This is one form of implicit evaluation that is understood but not clearly stated. It is claimed that declarative present-tense statements create an impression of indisputableness (e.g., there is no room for questions or debate; Neurauter Kessels, 2011). By examining the corpus data, it is found that there are many examples of declarative present-tense statements. Examples: 6.8.1.4. Second-person address. This is another feature that includes the involvement of the reader through the use of the second-person pronoun form "you". Reviewers usually use this device to address, interact and connect with their readers. By examining the use of you in the two datasets under investigations, it is found that the second-person pronoun "you" is repeated 39 times in high-rated reviews and 42 times in low-rated reviews. By the use of the second person pronoun, the reviewer illustrates some typical ways in which he addresses, interacts and connects with their reader.
6.8.1.5. The use of conventional speech acts. Through the annotated text in which each sentence is marked by the suitable speech act, it was easy to determine the frequency of each speech act through the wordlist as mentioned previously. Results show that, in both high-and low -rated reviews, representatives are the most common speech act used. Reviewers try to persuade their readers by different methods. First as stated by Aristotle, one of the main persuasive appeals is logos in which the speaker uses facts and evidence. This is what reviewers usually do in online reviews. They provide the reader with information about the hotel, description of the appearance of the hotel and its location, giving their evaluation about the hotel, then advising them to purchase a service or by warning them against purchasing this service, and finally by committing themselves to purchase this hotel based on their experience. It is noted that often a speech act would have the form of a direct speech act, but its indirect function would be to persuade such as representatives. Although representatives are claims or statements, the influencers would often use them to their advantage to persuade their readers. This could be applied to the usage of many of the speech acts, since that reviewers use speech acts to persuade other consumers in an indirect way. Indirect speech acts are, therefore, important in the present study since they are the predominant strategies in the selected online reviews. Some speech acts are simply known as more persuasive such as combinations of representatives and commissives, or representatives and directives, representatives and expressives, or the opposite expressives and representatives as the previous examples showed.
The study is mostly qualitative and it focuses on the context of persuasion in online reviews through different techniques. In particular, it focuses on how persuasive speech acts are used in online reviews. These findings are supported with other results as the study also examined some of the rhetorical devices used by the reviewers which show how reviewers involved the readers in the reviews, and let them engaged in the review which can have a direct effect on having a better decision. These findings of the different pragmatic functions of speech acts found in online reviews are also in line with previous studies such as Hennig Thurau & Vásquez, 2014, p. 44) who state that the motivations that reviewers may have while posting online reviews are: • The concern for other consumers as some reviewers wish to provide advice for others without any apparent benefits for them.
• The desire to help the company often as a result of the satisfaction of the consumer with the company's product or service.
• Another motivation that is related to social benefits is the motive that some consumers long for participation in and be members of an online community.
• Seeking additional information and advice from other consumers who have also purchased the product.
• The convenience of complaining as it is important for consumer's decision to know where to complain.
• Venting of negative feelings and expressing positive emotions are two more motivations for writing reviews. Writing a review about a negative experience can help in reducing consumer's feeling of dissatisfaction. Similarly, very strong positive experiences may lead to feelings of imbalance, balance can be restored by writing and posting positive comments.

Conclusion
Electronic Mouth-of-Word (eMOW) is considered as an essential genre of online communication nowadays in which people write about their experiences in purchasing a product or service so that other consumers will read it and benefit from it. This could be for different motivations or reasons. The importance of these online reviews draws from the importance of mass communication. Here, in online reviews one message can reach out to countless people. Therefore, in order to be persuasive, reviewers resort to certain strategies and certain devices that help them to affect other consumers. Reviewers interact with their readers through their reviews. Persuasion is expressed in the corpus data at different levels of language (i.e., lexicogrammar, discourse, and rhetorical devices). At the level of lexicogrammar, word classes are commonly used by reviewers to express their assessment of a product and experience. There are many keywords that are found in the online reviews that can affect the reader's decision. Of such words are top frequent words that are related to the hotel and its facilities, evaluative adjectives, stance adverbs, spatio-temporal adverbs (e.g., back and again). These keywords are the factors that affect reviewers' rating of hotels whether the rating is positive or negative.
At the discourse-level, reviewers use different speech acts in the online reviews. These include representatives, expressives, commissives, and directives. Each perform more than one pragmatic function. Some had a higher frequency usage of speech acts than others. The only speech act from Searle's (1979) taxonomy that is not found in the data is declaratives. This could be because none of the reviews found in the collected data for this study changed the reality of the world. These speech acts are used by the reviewers as persuasive tools that can affect the reader. Representatives have the highest frequency of all speech acts in the online reviews under investigation. They are positive statements or claims about the hotel, and they can persuade the reader to purchase it. Commissives and expressives are used in a similar way by the reviewers. Commissives, on the one hand, are often structured in a way that they promise or guarantee the quality of something, and, therefore, they are used to persuade. Expressives, on the other hand, are also used by the reviewers to give a positive feedback about the hotel or a negative one which can have an effect on the consumer's decision. Directives have the lowest frequency of all speech acts in high-rated reviews. They are advise and warning. Therefore, they can be used as motivational commands especially when the reviewer wants the reader to purchase a particular hotel.
Most of the speech acts used in the corpus data are direct speech acts. However, it is found that indirect speech acts are also used in the online reviews. Some speech acts have one surface form, but their function is different. Most of the speech acts found in the online reviews have a certain form like declarative, yet the function of the declarative is a suggestion or recommendation. This result coincides with Searle's (1985, pp. 30-35) claim that although the structure of the sentence might show a certain speech act, it can still have an indirect persuasive meaning. This is also true for interrogatives and imperatives as well in this study. They have the forms of interrogatives and imperatives, but they have a different function. Thus, it can be concluded that reviewers use speech acts in such a way to influence the readers with their perspectives, thus affect their decision. They use language in such a way to persuade their readers to choose a certain hotel and avoid another. They use a variety of persuasive strategies through expressing their judgment or opinion to the reader, sharing their experiences, describing the place, all these in order to make a strong case to convince their readers of their position. Moreover, it is found that reviewers use certain patterns of speech acts to persuade others. Therefore, different schematic patterns of speech acts are found in this study. All together in one review text can have an impact on the consumer's decision.
Because this type of genre is a written form, therefore, it is important to have some features that attract the attention of the reader, therefore, reviewers use certain tools of persuasion that can involve and engage the readers, and accordingly, can have a direct effect on having a better decision. Such tools include rhetorical devices such as the use of inclusive language, declarative present-tense statements, second person pronoun, rhetorical questions and others. Thus, it can be concluded that an online review is a multi-function genre that involves different speech acts besides being a persuasive tool that affects the reader either positively or negatively.

Funding
The author received no direct funding for this research.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Citation information
Cite this article as: A corpus-based study of reviewers' usage of speech acts, Marwa Adel Nasser, Cogent Arts & Humanities (2022), 9: 2125155.