Formality of speech and professionalism in Italian politics

(1)

Formality of Speech and

Professionalism in Italian Politics

Author Jonathan Busnelli Student Number 11248939

First Reader G. Schumacher, Ph.D. Second Reader I. Verhoeven, Ph.D.

Date June 23, 2017 MSc

Specialization

Political Science

(2)

I - Introduction

This research is concerned with analyzing the concept of formality of speech, which is one of the most frequently discussed dimensions of verbal behavior (Li et al., 2016). In particular, I will be focusing on the definition of formal discourse provided by Heylighen and Dewaele (1999), who identified it as the attempt to avoid ambiguity by adding information about the context. Moving on from this conceptualization, I will be employing their measurement of formality, called F Score, to carry out my analysis.

While most authors seem to agree on how to describe and operationalize formality, there is still not a consensus on what lies behind the decision to use a more or less formal manner of expression. Partially borrowing Bandura's conceptual framework (1986), I can assert that linguistic behavior is usually a product of both environmental and personal determinants, clearly indicating a causal relationship between these variables. As far as formality is concerned, the following aspects, both related to the environment and the individual, have been investigated as plausible predictors of formality of speech: situation, gender, personality and education (Heylighen and Dewaele, 1999). On the one hand, there is solid evidence that the context in which a certain verbal behavior is produced has an effect on formality (ibid.), meaning that, in certain circumstances, deemed as more formal (e.g. parliament or court), the degree of formality of speech will be adjusted accordingly. On the other hand, there is only very limited or no evidence at all for the other causes of formal discourse so far (ibid.).

The aim of this research is to add another element to this list of plausible determinants of formality by investigating its relationship with professionalism in a political context. In particular, my goal is to answer the following research question:

To what extent does the level of professionalism of Italian politicians explain the degree of formality of their parliamentary speeches?

What does the term professionalism mean? It is a rather loose concept, which has been employed with different meanings in the literature (Squire, 2007). Following Bundi et al. (2016), I intend to define it as the amount of time, energies and resources a member of the Italian parliament (MP) spends on his or her mandate. My decision to select this definition among others is due to the fact that it accounts for both environmental (i.e. time) and personal (i.e. energies and resources) determinants of formality, thus respecting the conceptual framework already mentioned in this introduction.

(4)

On the basis of this multi-layered conceptualization, I will be measuring professionalism in five different ways and I will be employing them all as separate independent variables in my analysis, in order to better explore their relationship with formality of speech.

It is now clear that the more time, energies and resources a MP spends on his or her mandate, the more professional he or she is. However, given that roles such as the one in question are heavily embedded in their context (Searing, 1987), which is per se formal, it is reasonable to assume that the more time, energies and resources a MP spends on his or her mandate, the more he or she will become accustomed to this formal environment; and this could have an effect on his or her way of speaking, leading to a more formal manner of expression. Therefore, I can assume that a higher degree of professionalism will result in a higher degree of formality of speech.

This research will have the following structure. First, I will outline my theoretical framework, where I will provide the basic conceptual premises of my analysis, an accurate description of the different conceptualizations of formality and professionalism, and an outline of the main hypotheses of this research. Second, I will be discussing my sampling strategy, the data collection phase, how I built the dependent and independent variables of my analysis, and the methods I intend to use to test my hypotheses. Third, results from these tests will be presented, analyzed and discussed. In conclusion, I will provide a comprehensive answer to my research question, also indicating the main limitations of my research and plausible avenues of future studies on the matter of formality and professionalism in politics.

(5)

II - Theoretical Framework

1. Conceptual Premises

Human behavior has been regularly explicated in terms of unidirectional causation, meaning that it is affected by environmental influences as well as internal dispositions. Building on this approach, Bandura (1986) developed the so-called social cognitive theory, which is based instead on a triadic causation model. More specifically, according to his perspective, personal factors, behavioral patterns and environmental determinants all contribute to influencing human behavior in a cohesive system. This leads to the definition of individuals as "producers as well as products of social systems" (Bandura, 2001: 266).

Having said this, for the purpose of this research, I will be focusing my attention on environmental and personal determinants of human behavior, mostly leaving aside the aspect of personal agency, which could be used as a starting point for future studies on this topic. Applying these basic theoretical assumptions to my subject of interest, it is clear that the human behavior in question is purely linguistic; in particular, it has to do with formality of speech. Moreover, given my interest in exploring both environmental and personal determinants of formality, it follows that the notion of professionalism will be conceptualized in a multifaceted way, taking into account both factors (Figure 1).

After clarifying the basic theoretical grounds of this research, I can now move on to discuss in depth the main variables of my analysis.

Professionalism - Environmental Determinants Professionalism - Personal Determinants Formality of Speech - Verbal Behavior

Figure 1 - Schematization of the causal

relationship between professionalism and formality of speech

(6)

2. Definitions of Formality

Over the years several scholars have attempted to propose a comprehensive definition of formality (Li et al., 2016).

First of all, Atkinson (1982) identified everyday conversations as the paradigm of informality, and conceptualized formal as non-conversational on the basis of a number of characteristics that contrasted with everyday dialogs, such as setting or persons involved (ibid.). As a result, an example of formal language could be the sentence delivered by a judge at the end of a trial, while a prototype of informality could be represented by a relaxed chat between friends or family members (Heylighen and Dewaele, 1999).

Subsequently, organizational or institutional conventions were used as a model of formal speech, leading to the definition of formality as adherence to "officially standardized and recognized institutional conventions or prescriptions" (Andren et al., 2010: 224). More specifically, four classes of informal linguistic elements were proposed to account for formality in conversations: "informal lexical embedding (e.g. 'hi there'), colloquial style or jargons (e.g. 'what do you say to that?' instead of 'how do you plead, guilty or not?'; Linell et al., 1993), omissions of formally required parts (e.g. abbreviations), and additions to non-task talks", such as phatic expressions (Li et al., 2016: 208).

It is also worth mentioning how other authors (Graesser et al., 2014) claimed that formal speech is used when there is a necessity to be accurate, clear, eloquent and persuasive to an educated audience. Its contrary is informal speech, typical of oral conversations, personal letters and narratives, which mostly rely on a shared background (ibid.).

However, these conceptualizations only give us a general idea of what a formal situation typically entails. They all lack a thorough description of what a formal discourse truly is, thus failing to reveal the construct that lies behind the notion of formality.

2.1 Formality, Ambiguity and Context

Following the assumption that formal speech is characterized by a certain kind of particular attention to form (Labov, 1972), Heylighen and Dewaele (1999) distinguished between two different types of formality. On the one hand, there is surface formality, which is defined as nothing but attention to form for the sake of convention or form itself. On the other hand, there is deep formality, in which attention to form has the purpose of ensuring that one's expressions are not misunderstood. In line with this distinction, they decided to focus on the latter type, thus advancing a temporary definition of formality as avoidance of ambiguity (Heylighen and Dewaele, 1999).

(7)

They went on to highlight the crucial role played by context in solving semantic ambiguity and identifying different linguistic constructions (Duranti and Goodwin, 1992). In particular, context is defined as "everything available for awareness which is not part of the expression itself, but which is necessary to correctly interpret the expression" (Heylighen and Dewaele, 2002: 297). As a result, they started considering a number of expressions, called deixis or context-dependent, which must be attached to some kind of context in order to be meaningful (Levelt, 1989). Clear examples of this are simple expressions, such as 'you', 'my' or 'him', which need to be linked to a person, or 'there', 'over here', 'downstairs', which need to be linked to a specific location, or 'after', 'then', 'yesterday', which need to be linked to a certain time (Heylighen and Dewaele, 1999). Only additional information about the context will reveal which person, location or time these deictic expressions are referred to.

Consequently, they were able to update their overall definition of formality, which I intend to borrow for this research, as avoidance of ambiguity by including information about the context. More specifically, if the audience is not familiar with the content of the message, it is essential to avoid deictic words and make use instead of unambiguous expressions. At this point, it is clear that formality finds its natural opposite in the notion of context-dependence. Yet, for the purpose of this research I will be only using the concept of formality, in order to avoid misunderstandings and be as clear as possible.

Moreover, it is worth adding that formality is a relational concept (Graesser et al., 2014). This means that an expression can be more or less formal with respect to another one, but no phrase can be absolutely formal or informal. All linguistic expressions will be placed in between the two extremes of this continuum (Heylighen and Dewaele, 1999; Joos, 1961).

3. Determinants of Formality

As summarized in Figure 2, the decision of formulating a certain idea in a more or less formal way could depend on a number of both environmental and personal elements, such as situation, gender, personality or level of education (ibid.). It is now my goal to briefly analyze existing evidence of the impact that certain factors have on formality of speech.

3.1 Situation

According to Levelt (1989), the most basic concepts that lie behind each situation are represented by the people involved, the location, the time and the discourse preceding

(8)

the expression in question. Given that formality will be higher where correct interpretation is essential, it follows that a small shared context (e.g. cultural and psychological background) between interlocutors will result in a higher degree of formality (Heylighen and Dewaele, 1999). Another crucial implication is related to audience size, meaning that the larger the audience, the smaller the shared context and the higher the level of formality will be (ibid.).

3.2 Gender

Over the years, several scholars have investigated the relationship between gender and language. Most were able to achieve interesting gender-related effects, resulting in active discussions concerning the actual substance of their findings (Thorne et al., 1983). In particular, Heylighen and Dewaele (1999) were able to establish that women tend to prefer a less formal style of speech. A plausible explanation for this could be related to the fact that women tend to be more personal and engaged in conversations, while men have a tendency to be more aloof and uninvolved (Hogg, 1985; Tannen, 2003). Nevertheless, most of this still belongs to the realm of speculation; further research is needed to better assess the relationship between gender and formality of speech.

3.3 Personality

As far as personality is concerned, research shows that introverted people are more likely to engage in a conversation using a higher degree of formality (Furnham, 1990). This finding is partially confirmed by other studies, linking extraversion to a lower degree of formality and introversion to a higher degree of formality (Heylighen and Dewaele, 1999). This could be due to the fact that introverts devote more time to thinking before they speak, while extraverts are quicker at reacting, avoiding unnecessary pauses (ibid.).

3.4 Level of Education

Intuitively it is reasonable to expect that a higher level of education will result in a richer vocabulary, a more precise way of conveying one's thoughts and, as a result, to a higher degree of formality.

Nevertheless, according to the data analyzed by Heylighen and Dewaele (1999), there is mixed evidence for the assumption that people with a university degree express themselves in a more formal manner.

(9)

4. Formality in a Political Environment

Before analyzing the other crucial aspect of this research, namely the notion of professionalism, a few considerations relating to the application of formality to a political environment must be made, especially taking into account what has been said in this theoretical framework so far.

First of all, as already mentioned, in certain circumstances it is reasonable to expect a naturally higher degree of formality. That is exactly the case of speeches delivered by politicians in a parliament, in which people from different social and cultural backgrounds are placed together and expected to interact on the topic of a series of complicated legislative and political issues. It is also a context in which it is essential not to be misunderstood. Consequently, following my definition, it seems logical to expect on average a high level of formality of speech.

At the same time, however, I am anticipating to encounter a range of different levels of formality, which could be explained by both personal and environmental determinants of political professionalism. Situation Gender Personality Education Formality of Speech Figure 2 - Schematization of plausible determinants of formality (Heylighen and

(10)

5. Definitions of Professionalism

According to Squire (2007), while there has been some kind of consensus on how to operationalize professionalism, there is no specific definition that stands out. The main point of disagreement among scholars is whether it can be applied to legislators (individual level) or should only be used with respect to the legislature (institutional level).

5.1 Professional Legislature

Rosenthal (1996) claims that professionalism cannot be associated with individuals, as it is an intrinsic characteristic of the legislature. Moncrief (1994) asserts the same concept, declaring that professionalism is about institutional causes and consequences, and that its main goal is to make a legislature more efficient.

Both authors identify careerism as its corresponding concept at the individual level. Unlike professionalism, it has to do with opportunity, inner desire and ambition to achieve (ibid.). At times these two concepts may overlap, thus making it difficult to discriminate between them, but according to both scholars it is important to keep them separate.

What does a professional legislature consist of? Following Grumm (1971), it is usually identified with a variety of different components (Bowen and Greene, 2014). Moncrief (1994) relates it to compensation, staffing, session length and turnover. Rosenthal (1996) believes that it has to do with an overall abundance of resources. In particular, he refers to the so-called five S's: space, salary, session length, staff and structure.

5.2 Professional Legislator

According to Squire (2007), professionalism is a concept that can be applied to both the individual and the institutional level. In particular, he claims that political professionalism is intended to assess the capacity of both individual members and the organization as a whole. Moreover, he clearly states that individual professionalism is different from careerism. The former can be defined using the same components of a professional legislature and applying them to the individual level; the latter, as already mentioned, is about opportunity, motivation and ambition (ibid.).

Borchert (2011) confirms that professionalism can be applied to individuals, mentioning that professional politicians are more motivated by income, career maintenance and advancement than their amateur colleagues.

(11)

Searing (1987) gives its own contribution to this way of conceptualizing professionalism by investigating post-war British politics. More specifically, he claims that professional politicians can be associated with a number of features, such as "full-time commitment, service to clients, relative autonomy in authority and judgment" (Searing, 1987: 432).

6. Professionalism: Environmental and Personal Factors

For the purpose of this research, following Squire (2007), I consider professionalism to be applicable to both the individual and the institutional level.

Moreover, given my theoretical premises, I have selected five different conceptualizations of professionalism, taking into account both environmental and personal factors. My decision to employ all these different definitions in my research will allow me to explore more thoroughly the relationship between professionalism and formality of speech.

As far as environmental determinants are concerned, I have identified three main conceptualizations, which are all related to the amount of time a MP spends embedded in a certain context, that is the Italian parliament.

First, professionalism can be considered as a synonym of not having an occupation other than one's parliamentary mandate. This was inspired by Bailer and Bütikofen (2015), who, in their analysis of the Swiss party system, associated the notion of professionalism with full-time commitment to one's political career. Bundi et al. (2016) also specifically defined it as the amount of time a MP devotes to his or her mandate.

Second, I have decided to conceptualize professionalism using attendance rates with respect to the total amount of meetings in a given period. This was inspired by Moncrief (1994), who utilized sitting days, meaning how many days per year politicians meet, as a dimension of professionalism. Squire (2007) also used the number of days a legislature meets as a component of professionalism.

Third, in line with Bundi et al. (2016) and their definition of professionalism as the amount of time a MP spends in parliament, I made the decision to relate it to the length of one's parliamentary career (LPC).

Moving on to personal factors of professionalism, I have identified two main definitions, which have both to do with specific characteristics that MPs bring into the political environment, such as level of education and individual productivity.

First of all, following Searing (1987), who assumed a relationship between professionalism and high levels of education, it seems reasonable to conceptualize it as a synonym of having a university degree.

(12)

Lastly, I decided to associate professionalism with the amount of draft laws proposed in parliament in a given period. This could be explained by the fact that, according to Searing (1987), professional politicians are usually deemed as the most strenuous policy advocates; they are service-oriented and want to obtain results.

7. Professionalism and Formality

After analyzing in depth how I intend to conceptualize both variables, it now seems appropriate to discuss what kind of relationship I am expecting to encounter between professionalism and formality. As already mentioned in the conceptual premises of this chapter, I partially borrowed Bandura's framework (1986) and established a causal relationship between my variables. Yet, I have not clarified the direction of this causation.

Professionalism has been associated with time (environmental determinants) as well energies and resources (personal determinants) one devotes to his or her mandate. First, it seems logical to assume that a MP operating in parliament would be inevitably affected by this environment and adapt his or her behavior according to its features. In particular, as already stated, given the overall formality of the context, it seems plausible that being involved in a parliament for a longer amount of time will result in MPs having a higher level of formality of speech. Second, as far as personal factors are concerned, it seems reasonable to assume that a higher level of education as well as an overall stronger individual productivity will result in MPs being better prepared to be active in such a formal context. As a result, they will adapt their level of formality of speech accordingly.

To sum up, on purely theoretical grounds it seems justified to state that a higher degree of professionalism could be a good predictor of formality of speech. This leads me to the formulation of my main hypothesis (H*):

Hypothesis*: A higher level of professionalism will result in a higher degree of formality of speech.

Having outlined the overarching hypothesis of this research, I can now decline it according to the five main conceptualizations of professionalism that I have selected:

Hypothesis 1: The lack of a second occupation will result in a higher degree of formality of speech. Hypothesis 2: A higher attendance rate will result in a higher degree of formality of speech. Hypothesis 3: A longer parliamentary career will result in a higher degree of formality of speech.

(13)

Hypothesis 4: A university degree will result in a higher degree of formality of speech.

(14)

III - Research Design and Methodology

1. Data, Context and Boundaries

To test the proposed hypotheses, it is clear that parliamentary speeches as well as a series of information on individual MPs represent the kind of raw data that must be collected and analyzed.

In 1948 the Italian Constitution established a pure parliamentary system, consisting of a total of 950 members (since 1963), unequally distributed between the Senate and the Chamber of Deputies (Fedeli et al., 2014). However, considering that a citizen must reach 40 years of age to be allowed in the passive electorate of the Senate (ibid.), I have decided to focus solely on the Chamber of Deputies. This allowed me to take into account a pool of MPs of all ages. Moreover, given the enormous quantity of relevant data, only speeches delivered in the General Assembly of the Chamber of Deputies were collected, excluding those held in the Commission or other legislative bodies.

Before making a decision on which parliamentary groups to include in the analysis, a specific time period must be determined, so that an appropriate historical reconstruction of the Chamber of Deputies can be made. Official datasets of the Italian lower house provide digital transcripts of parliamentary sessions from April 28th 2006, which marks the beginning of the XV legislature, up to the present day. Due to the considerable amount of information and the laboriousness of organizing it in a proper way for the subsequent steps, I have chosen a very recent time frame, namely a one-year period, beginning on December 4th 2015 and ending on December 4th 2016. On this day Prime Minister Matteo Renzi resigned in the wake of his loss in the Constitutional Referendum.

After establishing a precise time frame, I selected five major parliamentary groups, covering the whole right-left political spectrum (Garzia, 2013): the right-wing Lega Nord e Autonomie - Lega dei Popoli - Noi con Salvini (LN), the center-right Forza Italia - Popolo delle Libertà - Berlusconi Presidente (FI), the populist Movimento Cinque Stelle (M5S), the center-left Partito Democratico (PD) and the left-wing Sinistra Italiana - Sinistra Ecologia Libertà (SEL). The M5S was placed in the middle of this ideological scale, not because it holds centrist political views, but due to the impossibility of labeling it as either a right-wing or a left-wing entity (Bordignon and Ceccarini, 2013; Franzosi et al., 2015; Mosca, 2014).

Having specified which factions are the main focus of my research, it is now worth mentioning which political blocs were left out. If we consider all parliamentary groups existing from the beginning of the XVII legislature, the following ones were excluded from my analysis: the centrist Scelta Civica per L'Italia, renamed Civici e Innovatori on October

(15)

12th 2016; the right-wing Fratelli d'Italia - Alleanza Nazionale; every component of the Gruppo Misto, which is a parliamentary group comprising MPs that do not belong to any other big political faction, but are still allowed to form subgroups within it (e.g. Minoranze Linguistiche, Alleanza Liberalpopolare - MAIE,...). Moreover, if we take into account parliamentary groups that developed during our time frame, the following ones were also left out: the center-right Area Popolare (NCD - UDC), the Christian-democratic Per l'Italia - Centro Democratico, renamed Democrazia Solidale - Centro Democratico on January 11th 2016; the centrist Scelta Civica verso i Cittadini per l'Italia; every political fraction of the Gruppo Misto (e.g. Conservatori e Riformisti, Movimento PPA - Moderati, USEI,...).

Following this scaling down of pertinent material, the total number of MPs left in the Chamber of Deputies amounts to 493 according to the information provided by the official website: 16 LN MPs, 53 FI MPs, 91 M5S MPs, 302 PD MPs and 31 SEL MPs.

At this point, it is clear that a sampling plan needs to be implemented, so as to reduce relevant data to a convenient size and make the task executable (Krippendorff, 1980).

2. Sampling Plan and Sample Description

My decision to employ a stratified sampling strategy was motivated by the fact that it allowed me to split the population of interest into subgroups or strata, from which I could select my units of analysis according to specific criteria.

Firstly, MPs were chosen on the basis of their standing and prominence within their respective parliamentary group, meaning that Presidents, Vice-Presidents, Treasurers and Secretaries were all taken into account. This first criterion accounts for 60% of the overall sample size (30 out of 50): 4 LN MPs, 4 FI MPs, 11 M5S MPs, 12 PD MPs and 7 SEL MPs. Secondly, in order to avoid prominence bias, the next criterion is based on the length of one's parliamentary career (LPC). More specifically, the two MPs with the shortest parliamentary career and the two MPs with the longest one were drawn from each group. However, a third criterion, hierarchically subordinate to the second one, had to be introduced to discriminate among MPs with equal LPCs. In particular, whenever equal LPCs were encountered, age was taken into consideration: the younger MP was selected for the shortest parliamentary career side and the older MP was picked for the longest one. The second criterion, coupled with the third one, accounts for the remaining 40% of the total sample size (20 out of 50): 4 LN MPs, 4 FI MPs, 4 M5S MPs, 4 PD MPs and 4 SEL MPs.

Furthermore, an additional scaling down of the sample had to be performed, when it became evident that no data was available for a certain number of MPs in the time frame considered. If this happened in the prominence stratum of the sample, no substitution was

(16)

or could be made. In particular, no data could be found on 4 PD MPs, namely De Micheli, Giacomelli, Velo and Bellanova, leading to total number of 12 PD MPs instead of the original 16 MPs, following the above described sampling plan.

If missing data was to be found in the other stratum, the MP next in line on the basis of either LPC or age was selected to make up for this data deficiency. Among the LN MPs, Caparini (LPC = 20) was selected instead of Bossi (LPC = 26) due to missing data. The same can be said about Prestigiacomo (LPC = 22) in FI: there was no data on either Crimi (LPC = 23) or Martino (LPC = 23). Lastly, in the PD Monaco (LPC = 17) was picked instead of Bressa (LPC = 21). Following this procedure, as illustrated in Figure 3, a final sample size of 46 was established with 26 MPs selected using the prominence criterion and 20 MPs selected using the LPC/age criterion. A full list of sampled MPs can be found in Appendix A.

3. Building the Variables

Now that the data collection phase is complete, I can focus on building the main variables of this analysis, starting from the dependent one, namely the degree of formality. On the other hand, as already mentioned, in order to construct the independent variable, the level of professionalism, more information on a series of individual characteristics of each MP will be collected.

Prominence Stratum 26 MPs LPC/Age Stratum 20 MPs Sample 46 MPs Figure 3 - Visual representation of the sampling plan

(17)

3.1 Formality - Dependent Variable

As discussed in the theoretical framework of the research, the level of formality is defined as the attempt to eliminate ambiguity by providing information about the context (Heylighen and Dewaele, 1999). The basic logic that stems from this conceptualization is to divide the lexicon into two classes of words. On the one hand, we have deictic or context-dependent expressions, which are ambiguous on their own and therefore require additional information (Heylighen, 1999). On the other hand, there are non-deictic or context-independent words, which do not necessarily need to be attached to additional context to be significant (ibid.). It follows that a high frequency of deictic expressions is related to a low degree of formality, while a high frequency of non-deictic words can be associated with a higher level of formality of speech or text (Heylighen and Dewaele, 1999). A fitting example of context-independent words is provided by nouns, whose meaning does not automatically vary with changes in the context (ibid.). It is true that they still depend on a shared context to be thoroughly interpreted, but it is distinctly less specific than the one required by context-dependent words. The frequency of nouns is positively correlated with the frequency of adjectives and prepositions, thus allowing us to place them both in the non-deictic category (Dewaele, 1996). Moreover, according to Kleiber (1991), articles do not make any particular reference to context and tend to covary with nouns, thus making it clear that they belong to the context-independent category.

Moving on to the non-deictic group, verbs can be identified as typical context-dependent words, as they indirectly indicate a specific time through their tense (Levelt, 1989) and a specific subject through their inflection (Heylighen, 1999). This is especially true for a language like Italian, in which subject pronouns do not have to be necessarily expressed, as the referent can be clearly inferred from the form of the verb (Heylighen and Dewaele, 1999). The frequency of verbs is positively correlated with the frequency of adverbs and pronouns, which therefore can be both placed in the context-dependent category (Dewaele, 1996). Furthermore, due to the lack of a fixed referent and their subsequent high degree of ambiguity, interjections belong to the deictic class of words (Heylighen, 1999).

Lastly, it is worth noting that conjunctions will not be taken into account. They do not belong to either the deictic or the non-deictic category, as they lack reference to a specific context or to an objective meaning (Dewaele, 1996; Heylighen and Dewaele, 1999).

To sum up, nouns, adjectives, prepositions and articles are non-deictic words, whose frequency is expected to increase with the formality of speech, while pronouns, verbs, adverbs and interjections are deictic words, whose frequency is expected to

(18)

decrease with the formality of a text. This leads us to the following indicator, named F Score (Heylighen and Dewaele, 1999):

F = (noun freq. + adjective freq. + preposition freq. + article freq. - pronoun freq. - verb freq. - adverb freq. - interjection freq. + 100)/2

The above-mentioned frequencies are essentially percentages of the amount of words pertaining to a certain category with respect to the total amount of words in a text (Heylighen, 1999). The F Score can vary from 0 to 100, but will hardly ever reach those extreme values. The more formal the language, the higher it will be (ibid.).

As far as its applicability is concerned, the F Score has been successfully tested with a variety of alphabetical languages, such as English, French, Italian and Dutch (Heylighen and Dewaele, 1999). It has also been employed with Chinese, after a series of necessary adjustments to the formula, so as to account for the different structure of a non-alphabetical language (Li et al., 2016).

3.1.1 POS Tagging and Building the F Score

Having established how to operationalize the dependent variable of this analysis, I can now move on to apply it to the data I collected. My aim is to obtain individual F Scores of each MP included in the sample. To do so, I first tagged the whole corpus of parliamentary discourses using a software called TagAnt, which provides automatic part-of-speech (POS) tagging for all the grammatical categories contained in the formula with the exception of interjections: Adj. (adjectives), Det. (articles), Nom. (nouns), Prep. (prepositions), Adv. (adverbs), Pron. (pronouns), Ver. (verbs).

Subsequently, all tagged files were uploaded on AntConc, a text analysis software, which allowed me to search for each tag through its concordance tool, providing the total number of hits per tag for every uploaded file as well as for the whole corpus.

After obtaining results on every tagged word category, I uploaded on AntConc an official list of interjections from the Hoepli Online Dictionary (Gabrielli, 2015), in order to make up for the fact that interjections could not be tagged on TagAnt and in order to obtain all the data that was required to complete the formula. In the end, the total number of hits per category of each MP was transformed into a percentage with respect to the total amount of words per file, thus making it possible to calculate the overall F Score for each unit of analysis.

(19)

3.1.2 F Score: Descriptive Statistics

The degree of formality pertaining to the sampled MPs ranges from 59.37 to 76.70. Its standard error is 4.28 and its mean value is 66.61, indicating an overall above average level of formality. This could be due to the fact that subject pronouns do not have to be explicitly stated in Italian, which leads to a lower total amount of pronouns and, as a result, to higher formality scores (Heylighen and Dewaele, 1999).

The lowest value in the sample belongs to Matteo Mauri (PD), while Marco Bergonzi (PD) holds the highest F Score. Of particular interest is the fact that they both belong to the same parliamentary group (the center-left PD) and share other similar characteristics, such as being male and over 45 years of age, holding a university degree and having less than 3 years of LPC. It is also worth noting that, while Mauri has a second occupation, Bergonzi is a full-time politician. Figure 4 shows the mean value of the F Score for each parliamentary group, whereas a complete list of MPs with their respective F Score and file length can be found in Appendix A.

Notes: X = Parliamentary Group; Y = F Score.

Even if the overall F Score is a more reliable measurement than any word category considered by itself (ibid.), it is still interesting to analyze which classes correlate best with the levels of formality that I have obtained. First of all, as expected, formal categories (nouns, adjectives, articles and prepositions) increase with an increase in formality (R =

62,00 63,00 64,00 65,00 66,00 67,00 68,00 69,00 70,00 LN FI M5S PD SEL

(20)

.180), while informal categories (pronouns, interjections, adverbs and verbs) decrease with an increase in the F Score (R = -.315). Correlation for the formal categories is not significant, but in the expected direction; correlation for the informal categories is significant at the 0.05 level (2-tailed) and in the expected direction.

If I start comparing individual categories, results show that pronouns and adverbs are the ones that perform best (R = -.327 and R = -.314 respectively). Correlation for both classes is significant at the 0.05 level (2-tailed) and in the expected direction.This is not surprising, since they are clearly contextual categories, which makes them very likely to have a strong correlation with the F Score (ibid.). Nevertheless, as previously said, given the overall lower number of pronouns in Italian compared to other languages (ibid.), their general effect will be less significant.

Table 1 shows descriptive statistics on each word category as well as on the length of the corpus files I analyzed. A full list of the results of each word category for all MPs can be found in Appendix B (formal classes) and C (informal classes).

* The total here provided does not correspond to the sum of all the above listed word categories, as it also takes into account conjunctions, which are beyond the scope of this analysis.

3.1.3 F Score: Examples

To further illustrate how the F Score was applied to the sample of parliamentary speeches, I have decided to include two extracts with different levels of formality.

First of all, I have selected the following sentence, rich in nouns and adjectives, from the Bergonzi file (highest F Score in the sample), expecting a high level of formality:

"Il documento, in particolare, deve indicare gli sviluppi del processo di integrazione europea alle questioni istituzionali, alla politica estera e di sicurezza comune nonché alle relazioni esterne

Table 1 - Word Categories and File Length

Adj. Det. Nom. Prep. Adv. Interj. Pron. Ver. Total*

Min ₁₆ ₁₈ ₄₀ ₂₈ ₁₂ ₀ ₈ ₂₉ ₁₆₂

Max _5,155 _4,263 _12,601 _8,869 _4,578 ₈₅ _6,204 _9,121 _52,819 Sum _{55,318 4,472 150,121 100,294 38,099} _1,032 _{54,723 85,415 556,515}

(21)

dell'Unione Europea; poi, la partecipazione dell'Italia al processo normativo e l'attuazione in Italia delle politiche di coesione economica, sociale e territoriale, l'andamento dei flussi finanziari verso l'Italia e il loro utilizzo, il seguito dato alle iniziative assunte in relazione ai pareri, alle osservazioni

e agli atti di indirizzo delle Camere".

I then selected the following sentence, rich in verbs and adverbs, from the Mauri file (lowest F Score in the sample), expecting a low degree of formality:

"Ma chiediamoci, anche ascoltando il dibattito di oggi, a volte in alcuni passaggi surreale: che ne sarebbe stato di noi se, proprio in questi anni difficili, non ci fosse stata l'Europa? Chi lo dice, non

sa di cosa parla o fa finta di non saperlo, per avere solo argomenti da sventolare in qualche dibattito".

Before delving into these two extracts, it is worth bearing in mind that the F Score of a single sentence has very little meaning by itself and does not correspond in any way to the overall F Score of a file. Moreover, these two extracts have been specifically selected with the intention of showing a practical example of higher and lower degrees of formality.

As far as the F Scores are concerned, my selection has proven to be accurate: results show a degree of formality of 90.25 for Bergonzi and of 46.29 for Mauri. These values, especially the former, are quite extreme and unlikely to be reached when analyzing a complete corpus file. However, they allow us to fully understand the dynamics behind the F Score and how it was applied to the sample.

On the one hand, there is the Bergonzi extract, in which nominalization (formal) is preferred to verbalization (informal): nouns account for 33.77% of the total word count, while verbs represent only 5.19% of it. Given that nouns are usually positively correlated with adjectives and verbs with adverbs (Heylighen and Dewaele, 1999), if follows that we will encounter a high number of adjectives and a low amount of adverbs. This assumption is correct, as adjectives comprise 15.58% of the word count, while adverbs only account for 1.30% of the total.

On the other hand, in the Mauri extract verbalization is clearly preferred to nominalization, as nouns make up 14.81% of the word count, while verbs represent 25.93% of the total. Moreover, in line with the expected direction of the relationships, there is a higher number of adverbs (12.96%) than adjectives (7.41%).

(22)

* The total here provided does not correspond to the sum of all the above listed word categories, as it also takes into account conjunctions, which are beyond the scope of this analysis.

3.2 Professionalism - Independent Variable

Having obtained the F Score of every MP, I can now move on to build the independent variable of this research. Most scholars went on to measure professionalism with an aggregate index, combining a number of different characteristics (Grumm, 1971; Moncrief, 1994; Squire, 1992). However, according to Bowen and Greene (2014), it is a heterogeneous concept both theoretically and empirically. Therefore, under certain conditions it can be useful to disaggregate an index into its distinct components or build multidimensional measures.

In line with the definition of professionalism as the amount of time, energies and resources a MP dedicates to his or her mandate (Bundi et al., 2016), I was able to identify five different conceptualizations, as I explained in the previous chapter. The former three have to do with the environmental side of professionalism, while the latter two are concerned with the personal side of professionalism. My goal now is to clarify how I intend to measure them.

First, I investigated whether each MP had a second occupation or not. To do so, I utilized the information available on the official website of the Chamber of Deputies. In particular, I consider politicians who do not see their parliamentary mandate as their main occupation as less professional than those who do, since they invest less time on their political career (ibid.). In this instance, I created a dummy variable, with 0 indicating the absence of a second job and 1 the presence of a second job.

My next conceptualization relates professionalism to MPs attendance records in the Chamber of Deputies with respect to the total amount of times the lower house met within the time frame. The required data was collected from the OpenParlamento portal, a special project of the OpenPolis association, dedicated to making Italian politics more transparent. In this case, I built a ratio variable, consisting of the percentage of attendance of each MP. Third, I chose to operationalize professionalism with the LPC, a concept already employed in the sampling strategy section. More specifically, MPs with a long parliamentary career are to be considered as more professional than those with a short one, as they have spent

Table 2 - Word Categories and File Length

Adj. Det. Nom. Prep. Adv. Interj. Pron. Ver. Total*

Bergonzi ₁₂ ₈ ₂₆ ₂₂ ₁ ₀ ₁ ₄ ₇₇

(23)

more time on their elective mandates (ibid.). In this case, after getting the necessary information from the official website of the Chamber of Deputies, a ratio variable, consisting of the actual number of years spent in Parliament, was created.

I then decided to take into account the level of education of each MP. Those that hold a university degree are considered more professional than those who do not. A dummy variable was created with 0 indicating the absence of a university degree and 1 indicating the presence of a university degree.

Lastly, I decided to link professionalism with the amount of draft laws presented in the Chamber of Deputies in the time frame considered. The more draft laws submitted, the more professional a politician is, as he or she devotes more time, energies and resources to his or her political career. After collecting official data from the OpenParlamento portal, I built a ratio variable consisting of the actual number of draft laws proposed by each MP in the given time period.

3.2.1 Measurements of Professionalism Applied to the Sample

After discussing how to operationalize my variables, I can now move on to analyze the results of the application of different measurements of professionalism to our sample. In particular, 22 MPs (47.8%) do not hold a second job (0) and, following the definition provided in the theoretical framework, are considered more professional, while 24 MPs (52.2%) devote part of their time to another occupation (1) and are therefore deemed as less professional.

As far as the attendance rate is concerned, 9 MPs (19.5%) have attended less than 50% of the parliamentary sessions held in the time frame considered, while 37 MPs (80.4%) have an attendance rate that is higher than 50%.

Considering professionalism as LPC, a clear element stands out in its application to our sample: 26 MPs (56.5%) hold a relatively short parliamentary career (from 1 to 3 years), resulting in an independent variable that is rather left-skewed in its distribution.

Moving on to H4 which links professionalism to a higher level of education, we have 11 MPs (23.9%) that do not have a university degree (0 - less professional) and 35 MPs (76.1%) that have a university degree (1 - more professional).

Lastly, if we take into account professionalism defined as the amount of draft laws proposed, values range from 0 to 32 with an outlier of 95, represented by Caparini (LN). Table 3 sets out descriptive statistics for all the main variables included in my research. A list of frequencies of all independent variables for each MP can be found in Appendix D.

(24)

Notes: Mean, max and min have been reported for ratio variables; frequencies of 0 and 1 have been reported for dummy variables.

4. Statistical Analysis and Control Variables

Having built all the main variables of this analysis, it is now possible to fully investigate the relationship between professionalism and formality of speech by testing the main hypotheses of my research. To do so, I intend to employ a multiple regression analysis, in which the F Score is my dependent variable and the five different ways of operationalizing professionalism are my independent variables.

Other potentially significant variables will be incorporated in a second multiple regression analysis, in order to control for other plausible effects and explanations.

In particular, the age of each MP will be taken into account, using the middle of our time frame (June 5th 2016) as a benchmark to calculate it. In addition to age, place of birth will also be included in the analysis. More specifically, I created a dummy variable, named north, in which 0 indicates 'south' and 1 indicates 'north'. Gender will also be taken into consideration: I built another dummy variable and named it male, in which 0 stands for 'female', while 1 stands for 'male'. Moreover, a fourth control variable, named ideology, will be added to the analysis. To build it, an ideological left-right scale from the Chapel Hill Expert Survey (2012) was used. In particular, I selected the 'General Left-Right' indicator, which gives information on the ideological positioning of parties/parliamentary groups on a general left-right dimension, and ranges from 0, extreme left, to 10, extreme right (Bakker et al., 2012).

In the end, I will run a third multiple regression analysis, leaving the main variables unchanged, but replacing ideology with a new control variable, called radicalism. More specifically, I created a dummy variable, in which 0 indicates affiliation to FI, M5S and PD, while 1 indicates that a certain MP belongs to a radical group, such as LN or SEL. This

Table 3 - Dependent and Independent Variables: Descriptive Statistics

Name F Score Occupation Attendance LPC Education Draft

Type Ratio Dummy Ratio Ratio Dummy Ratio

Mean 66.61 - 68.09% 7.20 - 9.87

Max 76.70 - 97.56% 24 - 95

Min 59.37 - 11.21% 1 - 0

0 - 22 - - 11 -

(25)

decision is mainly due to the fact that the F Score seems to be lower at the extremes of the left-right political spectrum, suggesting an overall inferior degree of formality for radical parliamentary groups, as illustrated in Figure 4. In the end it is worth mentioning that my choice to run a third regression analysis, instead of incorporating radicalism in the second one, is essentially an attempt to avoid problems of multicollinearity, which would be detrimental to the overall quality of the model.

Table 4 summarizes descriptive statistics for all control variables included in the analysis. A full list of frequencies of all control variables for each sampled MP can be found in Appendix E.

Notes: Mean, max and min have been reported for ratio/interval variables; frequencies of 0 and 1 have been reported for dummy variables.

Table 4 - Control Variables: Descriptive Statistics

Name Age North Male Ideology Radicalism

Type Ratio Dummy Dummy Interval Dummy

Mean 45.93 - - 4.95 -

Max 69 - - 1.29 -

Min 29 - - 8.86 -

0 - 20 18 - 31

(26)

IV - Results and Analysis

1. Professionalism as a Predictor of Formality of Speech

A multiple regression analysis was run to predict the value of the F Score on the basis of five operationalizations of professionalism. As far as the overall model is concerned, results reveal an Adjusted R Square of .318, meaning that the different measurements of professionalism explain 31.8% of the overall variance of the dependent variable.

Moving on to the effect that each independent variable exerts on the F Score, my empirical analysis highlights a moderately strong correlation between the dependent variable and professionalism operationalized as not having a second occupation. More specifically, the b coefficient is -3.238, indicating a negative relationship between the two variables; the standard error is 1.098; the significance level is below 5%. This gives solid support to my H1, meaning that full-time MPs tend to employ a more formal language in their parliamentary speeches than their part-time colleagues.

A plausible explanation for this can be found directly in the literature. Language is inextricably bound to its context, which could be defined as the social and spatial framework within which a certain linguistic behavior is produced (Ochs, 1979). Given that certain contexts have a superior level of formality than others (e.g. parliament or court), in order to minimize the chance of a misinterpretation, individuals are likely to adjust their language and increase their own degree of formality (Heylighen and Dewaele, 1999). Therefore, it is reasonable to assume that the more time one spends embedded in a certain environment, the more likely he or she is to adapt to its features. It follows that in this instance the more working time one devotes to his or her parliamentary mandate, without being involved in other professions, the more one becomes accustomed to a context that is per se more formal, and this has an effect on one's way of speaking, meaning that the degree of formality will be naturally raised above average.

Turning my attention to H2, which assumes that a higher attendance rate will result in a higher degree of formality, the unstandardized coefficient is .073, pointing in the direction of a very weak relationship between the two variables in the expected direction. Nevertheless, the significance level is slightly above 5%, thus leading to the overall rejection of this hypothesis.

Evidence for H3, which assumes that a longer parliamentary career will result in a higher level of formality, is just as unconvincing. The unstandardized coefficient is .099, possibly

(27)

suggesting a very weak positive relationship between the two variables. However, the significance level is distinctly above 5%, making it impossible to confirm H3.

There could be support for H4, according to which a higher level of education will result in a higher degree of formality, since the unstandardized coefficient is 1.909, indicating a moderately strong positive relationship between the variables. Yet, even in this instance, the lack of significance leads to the rejection of this hypothesis.

Lastly, findings show that there is no support for H5, which assumes that a higher amount of draft laws will result in a higher degree of formality. The b coefficient is -.043, which possibly suggests a very weak negative relationship between the variables, thus pointing in a different direction from the one assumed in H5. Nevertheless, its significance level is markedly above 5%: this hypothesis must be rejected.

Notes: coefficients are unstandardized; standard errors in parentheses; significance levels: *p < 0.05, **p < 0.01, ***p < 0.001.

To sum up, this first multiple regression analysis, whose results are presented above in Table 5, helped me prove H1. Yet, it also led me to reject the other four hypotheses I formulated. Furthermore, in certain cases (i.e. H2, H3, H5) the significance level was

Table 5 - Multiple Regression Analysis: Predictors of Degree of Formality

Second Occupation -3.238* (1.098) Attendance Rate 0.073 (0.031) LPC 0.099 (0.098) Education 1.909 (1.273) Draft Laws -0.043 (0.038) Constant 61.569*** (2.779) Adjusted R Square 0.318 N 46

(28)

considerably above the 5% borderline, thus suggesting a non-linear relationship between the variables.

At this point, it is also worth mentioning that my results are based on a very small sample, which inevitably means that individual observations could have a strong influence on the findings and could therefore lead to a higher significance level. Consequently, future studies could increase the overall sample size, in order to better explore the relationship between degree of formality and professionalism.

2. Introducing Control Variables in the Analysis

In line with the strategy discussed in the methodology section, I will now compute a multiple regression analysis, aimed at investigating the relationship between our main variables, also taking into account four control variables: age, north, male and ideology. As far as the overall model is concerned, findings reveal an Adjusted R Square of .309, meaning that the model explains 30.9% of the variance of the dependent variable, which represents an expected slight drop in value compared to the previous case (31.8%).

Moving on to analyzing each variable and its corresponding hypothesis, previous findings concerning H1 can be confirmed. There is a moderately strong and positive relationship between the degree of formality and not having a second occupation. The unstandardized coefficient has slightly increased (-3.364), even if the standard error is also greater than before (1.134); the level of significance has remained unchanged (below 5%).

I can also reinforce the lack of evidence for H2: a marginally lower unstandardized coefficient of 0.068 is coupled with a marginally higher standard error of 0.033. The level of significance has also slightly increased, thus remaining above the 5% borderline.

Findings also show that there is no support for H3, given the distinct surge in its p value, clearly indicating the lack of a linear relationship between the two variables.

Evidence for H4 remains unchanged after controlling for age, place of birth, gender and ideology. The b coefficient is 1.675, possibly suggesting a relationship in the expected direction between education and formality; yet, the significance level remains above 5%, leading us to reject H4.

Lastly, I reassert my inability to prove H5. Results show a slight decrease in the value of b (-0.028) as well as a minor increase in the value of the standard error (0.040). However, due to the high significance level, H5 is rejected.

To sum up, the overall findings of this second model, which are presented in Table 6, have remained largely unchanged if compared to the first regression analysis. In

(29)

particular, I can confirm support for H1 as well as the lack of evidence for the remaining hypotheses due to significance levels exceeding the 5% borderline.

3. Controlling for Radicalism

One last multiple regression analysis was computed with the F Score being the dependent variable and my five measurements of professionalism being the independent

Second Occupation - 3.364* (1.134) Attendance Rate 0.068 (0.033) LPC 0.029 (0.119) Education 1.675 (1.315) Draft Laws -0.028 (0.040) Male -1.619 (1.154) Age 0.085 (0.066) North -0.515 (1.172) Ideology 0.085 (0.259) Constant 59.472*** (3.970) Adjusted R Square 0.309 N 46

(30)

variable. I also added age, north, male and radicalism, instead of ideology, as control variables. As far as the overall model is concerned, findings show an Adjusted R Square of .307, meaning that this third model explains 30.7% of the variance of the dependent variable, which is virtually identical to the explanatory power of the previous model. Turning my attention to the specific variables and their corresponding hypotheses, findings show little relevant changes with respect to the two previous models. More specifically, all non-significant results have remained such, leading to the rejection of H2, H3, H4 and H5. The only relevant finding emerging from the other regressions (i.e. professionalism as not having a second occupation) has now been rendered borderline non-significant by the addition of radicalism as a control variable. Its unstandardized coefficient is -3.343, higher than the one in the first model, but lower than the b encountered in the second regression. Yet, its significance level has slightly increased, surpassing the 5% borderline. Table 7 sums up results of this last multiple regression analysis.

Second Occupation - 3.343 (1.136) Attendance Rate 0.066 (0.033) LPC 0.037 (0.117) Education 1.712 (1.315) Draft Laws -0.024 (0.041) Male -1.538 (1.196) Age 0.078 (0.064) North -0.428 (1.190) [Continued]

(31)

4. Discussion

As already mentioned in the theoretical framework of this paper, the notion of professionalism has been conceptualized in multiple ways in the literature (Squire, 2007). Consequently, my decision to employ different operationalizations, instead of an aggregate index, seems justified. However, this inescapably means that, in order to prove my overall hypothesis (i.e. higher levels of professionalism resulting in a higher degree of formality), the majority of my sub-hypotheses relating to different measurements of professionalism (i.e. H1, H2, H3, H4, H5) ought not to be rejected.

Yet, my empirical analysis only allowed me to confirm H1, which assumes that not having a second occupation will result in a higher degree of formality; and this happened only in two out of three models. Due to significance levels exceeding the 5% borderline, I have been forced to reject the rest of my hypotheses, thus leading to an overall rejection of the overarching assumption of this research. Bearing in mind that increasing the size of the sample could possibly lead to different outcomes, I will now investigate the implications of my findings.

4.1 Professionalism is not a Predictor of Formality

In line with the conceptual framework presented in the theory section of this research, formality of speech is affected by both environmental and personal determinants. As already discussed, there is some empirical evidence, even if little theoretical interpretation on a variety of factors, relating to the situation (e.g. context) or to individual characteristics, such as gender, personality and level of education (Heylighen and Dewaele, 1999). In particular, a context in which a certain linguistic behavior is produced has proven to be a solid predictor of formality, meaning that, in circumstances

Radicalism -0.145 (1.337) Constant 60.195*** (3.501) Adjusted R Square 0.307 N 46

(32)

where it is important not to be misunderstood, formality of speech is likely going to be raised above average (ibid.). Evidence concerning the effect of individual features is limited or mixed: more research needs to be carried out, in order to fully assess the impact of personal factors on the degree of formality of speech.

My aim was to make a contribution to this list, possibly adding another determinant of formality in a political context; and, in order to enhance my chances of success, I conceptualized professionalism as both an environmental and a personal feature. It is also worth adding that the only significant result of my research had to do with the environmental side of my definition of professionalism (i.e. second occupation), possibly further validating existing evidence on the fact that the context itself is indeed the best determinant of formality.

Nevertheless, although theoretical assumptions about high levels of professionalism resulting in high levels of formality could be made, findings clearly indicate that professionalism is in fact not a good predictor of formality.

At this point, in light of the results of my research, it seems appropriate to briefly consider which different definitions and measurements of our variables could have been used, in order to possibly obtain more significant findings.

4.2 Different Measurements Leading to Different Results?

First of all, the F Score itself could be improved by partitioning each word category into more specific ones (e.g. distinguishing between different kinds of pronouns), thus creating a more refined operationalization of formality (Heylighen and Dewaele, 1999; Li et al., 2016). However, a number of different measures, other than the F Score, have been proposed throughout the years, including a composite formality index, called Coh-Metrix Score (Graesser et al., 2014). While the F Score relies on syntactic word categories to measure the degree of formality, this operationalization is based on a multi-level theoretical framework, consisting of words, syntax, text base, context model, genre and rhetorical type (ibid.). It takes into consideration both linguistic and non-linguistic factors that affect the degree of formality, thus helping uncover the real structure behind it.

Nevertheless, as already mentioned, there is definitely more of a consensus on how to define and operationalize formality than on how to conceptualize and measure professionalism. More specifically, if most authors have agreed on creating an aggregate index consisting of a variety of elements relating to professionalism, their theoretical definitions all seem to differ (Squire, 2007). My decision to employ separate operationalizations has led me to reject my main hypothesis. Consequently, it sounds

(33)

logical to suggest building an aggregate index in future studies, that are interested in exploring whether professionalism has any effect on the degree of formality of speech.

Formality of speech and professionalism in Italian politics