Repetition and Reduction. Two Contrastive Characters of Politeness Formulas in a Gender Perspective The Case of the Italian cià cià and Similar Expressions

This study is about politeness formulas as a class of fixed expressions and, in particular, it analyses the coexistence of two opposing phenomena – namely repetition and phonetic reduced variants – which contrasts with the typical stability and immutability of formulaic aspects in spoken communication. These aspects are discussed using the method of correlational sociolinguistics with reference to language and gender. The data show that men tend to use politeness formulas affected by repetition and reduction more than women. This tends to support the thesis according to which men are more exposed to linguistic variation and innovation while women would tend to a conservativeness and a tradition of the forms.


Introduction
Spoken communication is characterized by phenomena of mutability and instability due to the different speakers who use the language in different situations (Voghera 2017). Two of these phenomena are repetition and reduction. Both of them are very common aspects of language, and involve both content and functional words. In this study the two phenomena are analysed in the field of politeness formulas in spoken communication, and a gender perspective is adopted.
In particular, the repetition and the reduction phenomena -and their coexistence in specific contexts -represent the main issue in the field of formulaic language due to their constituent characters of fixity and stability in spoken communication too. In addition, this perspective can introduce a key factor for the study of language and gender according to the matter of variation and innovation of language forms in men and women.
The present paper has the following structure. After the introduction, the second paragraph deals with politeness formulas, their meaning and their theoretical foundations. The third paragraph deals with the phenomenon of repetition, while the fourth is about reduction, as related to formulaic language and gender variation. The fifth section concerns the methodology of our research and the presentation of the data on repetition and reduction. Then, in the sixth paragraph we present the results from the data, particularly the case of ciao, while in the seventh we show that, following the perspective of correlational sociolinguistics as regards the field of language and gender, men tend to use more repetition and reduction in politeness formulas than women. Finally, our conclusions are presented in paragraph eight.
be made of verbs (e.g. it. scusa, ʻsorryʼ; prego, ʻyou are welcomeʼ), nouns (e.g. it. auguri, ʻbest wishesʼ), or adverbs (e.g. a presto, ʻsee you soonʼ). 2 But they can also be made of a single word (e.g. it. salve, ʻhiʼ, en. bye), collocations (e.g. it. grazie mille, ʻthank you so muchʼ, en. good morning), locutions (e.g. it. in bocca al lupo, ʻgood luckʼ, en. see you soon) or sentences (e.g. ti ringrazio, ʻthank youʼ -lit. ʻI thank youʼ -en. how are you?). They can be effectively described by the construction grammar paradigm, according to which a construction -i.e. a combination of a form and a function -can represent diverse entities of varying complexity/simplicity and specificity/schematicity placed in the same continuum, without clear boundaries between the traditional levels of lexicon, syntax and morphology (Croft 2001;Goldberg 2006). In addition, politeness formulas are characterized by fixity as regards phonetics, lexicon-syntax, pragmatics and intonation and are therefore included in the macro class of 'formulaic language' (Coulmas 1994;Wray 2006;. According to Bardovi-Harlig: The use of the term formula in contemporary empirical pragmatics refers to recurrent strings or expressions used for specific pragmatic purposes. Formulas often succinctly capture the illocutionary force of a contribution by virtue of the fact that the speech community in which they are used has tacitly agreed on their form, meaning, and use. (2012,207) The social and pragmatic role of politeness formulas, according to the theory of linguistic politeness, is to limit conflict and encourage harmony in the community through the communicative interactions, also varying the linguistic choices in relation to the context, the style, the register, the channel and the means of communication (Lakoff 1973;Leech 1983Leech , 2014Brown, Levinson 1987;Watts 2003). Traditionally, it has been argued that women are more polite than men, because they make greater use of minimal responses to indicate support to the speaker (e.g. hedged statements and linguistic forms related to politeness), while men talk more, use a greater amount of offensive language and use imperative forms to face other men (Coates 1993;Holmes 1995;Mills 2003). This statement would be worth considering more carefully. However, this goes beyond the aims and limits of this paper. The Phenomenon of Repetition Repetition is a complex, ubiquitous and multidimensional phenomenon that affects all aspects of human existence. Daily linguistic and non-linguistic practices consist of routines in which humans do things and speak words in the same way day after day (Fischer 1995), without being aware of it. Indeed, Bazzanella (1996) theorizes a scale of consciousness of repetition that runs from the unconscious repetitions of the discourse markers to the conscious repetitions of rhetorical strategies of poets and advertisers. She also distinguishes between self-repetition and other-repetition and recognizes in this phenomenon different macro-functions (cognitive, textual, argumentative, stylistic and ethnic) and micro-functions (conversational and interactional). Aitchison (1995) provides two groups of variables useful for a classification of repetitions. The first group refers to means (written vs spoken), participants (self-repetition vs other-repetition), scale of fixity (degree of repetition, total or partial), time scale (immediate or delayed repetition) and size (phoneme, morpheme, word, phrase or sentence). The second group refers to function, intentionality and optionality of repetition. With regard to functions, repetition is used for intensification, cohesion, comprehensibility and conversational interaction. Intensification involves an increase in quantity and in quality, whereas repeated lexical items promote textual cohesion and avoid misunderstandings, and finally serve to maintain conversations. With regard to the distinction between intentional and unintentional repetition, Aitchison (1995) distinguishes between intentional repetition, which is more easily identifiable in writing and poetry, the so-called 'covert controlled', which occurs above all in informal conversations, and the unintentional repetition, which occurs to a different extent in both normophasic and pathological speakers. Finally, as regards optionality, the parameter provides four alternatives: obligatory repetition, which is necessary in cases where the phenomenon is a grammatical one, as in reduplication; optional repetition, related to the alteration of the words in the conversation; dispreferred repetition, when it is avoided to favour the pronominalization; and impossible repetition, when it is unacceptable.
Furthermore, repetition is particularly connected to the phenomenon of formulaic language. In particular, Tannen (2007) links the phenomenon of repetition in conversation to the issue of fixed, prefabricated language, as opposed to ex novo creation. In fact, the analysis of repetition indicates that much human language is not produced from scratch every time but relies on resources already available, such as the range of idiomatic and of formulaic expressions. This does not mean that speakers are unaware users of a collection of fixed expressions, but that the language they use actually arises from the in-terplay between the fixity and the novelty. In this perspective, Tannen (2007, 58) recognizes four functions of repetition in conversation. The first one is production, as repetition allows the speakers to express themselves more efficiently and with less use of energy. The second one is comprehension, because through redundancy the listener can receive and understand information at the same time or about in which it is produced. The third one is connection, as repetition links the different parts of a speech. The fourth one is interaction, because repetition serves to manage the conversation by tying their participants to each other.
Finally, Wang (2005) analyses the connection between formulaic sequences and repetition, considering the three dimensions of morphology and syntax, semantic and pragmatics. He distinguishes three groups of repetitions: full copy (e.g. on and on), partial reduplication (e.g. here and there) and triplet and others (e.g. tic-tac-toe).

The Phenomenon of Reduction
In speech every token of a word differs from other tokens of the same word in many aspects: pronunciation variants are present at all levels of spontaneous and semi-spontaneous speech and speakers tend to articulate what is necessary for understanding what they are saying, while simultaneously reducing any articulatory gesture that is not explicitly necessary. Accuracy of pronunciation depends on the linguistic and non-linguistic context and on the in-/formality of the speech situation. Even so, speakers can more or less consciously change their level of articulation in every moment (Albano Leoni, Maturi 1995;Voghera 2017). Reduced variants can be defined as words absent in their entirety, characterized by incomplete articulatory gestures or fewer segments as compared to the typical variants. All speakers use reduction and reduced variants are common both for function and content words in many languages. Moreover, it can be assumed that reduction processes are gradient rather than categorical. Indeed, sounds may be very short and weakly articulated, but still present. Furthermore, reduced sections of speech often contain clear cues to some phonological features that cannot be definitively localised.
Differences in speech situations may affect the quantitative (frequency of occurrence) and qualitative (type of reduction) effects of reduction. Previous studies have claimed that segments may be weakened or even completely silenced due to two completely different reasons. On the one hand, segments may be reduced during the articulation process. For instance, a segment may be inaudible because its articulation is hidden by the articulation of surrounding segments. On the other hand, segments may be weakened or silenced due to higher level processes, which include both phonological reduction process and the selection of a reduced pronunciation variant of the word from the mental lexicon (Ernestus, Warner 2011;Ernestus, Hanique, Verboom 2015).
Previous studies assert that there are differences among social groups. In particular, as regards the variable of gender, many studies affirm that, in general, men reduce more than women, even if there is not a considerable variance. Byrd (1994) states that the speaker-specific characteristic of sex influences speech rate, stop releases, flapping, central vowels, laryngeal state, syllabic consonants and palatalization processes. Many of these phonetic characteristics, produced more frequently by men, are typical of reduction in speech. The study of Bell et al. (1999) analyses the phonetic (speech rate, following consonant/vowel, accent, disfluency) and non-phonetic factors (planning problems, predictability and collocation, 3 position in turn and utterance, age and sex) that affect the reduction of frequent function words. The authors claim that the greater disfluency and the higher speech rate found in men are partially, but not completely, responsible for a larger amount of reduced variants in their speech. Finally, Keune et al. (2005) and Strik et al. (2008), state that men reduce more often than women due to their higher speech rate among other social and linguistic factors.
In particular, as regards politeness formulas, Jespersen (1922, 266-7) already observed that pronouncing tendency to hypoarticulate, which is a common phenomenon in all languages and which in extreme cases leads to a complete unintelligibility of what was said, also affects formulas of politeness (e.g. How d(e) do for How do you do).

5
Methodology of Research and Data

Aim of the Research and Corpus
In light of prior studies, this paper means to fill the existing research gap among specific phenomena of spoken language, and to analyse an indipendent class of pragmatic expressions and the gender perspective. Indeed, the aim of this study is the necessity to analyse not only the two phenomena of repetition and reduction in the politeness formulas, but also to consider the gender perspective evaluating the tendency to repetition and reduction -and so to innovation -in linguistic elements that concern social relationships.
The current study considers repetition and reduction in politeness formulas as the co-occurrence of two opposite phenomena against the stability and immutability of formulaic characters like fixity, conservativeness and conventionality in a gender perspective. Indeed, this study starts from the method of correlational sociolinguistics within the field of language and gender that takes binary sex as a legitimate starting point for analysis, that exists previously as a potential analytical category, and relies on quantification to identify general patterns (Hultengren 2008;Dittmar 2010). In particular, it addresses the following questions: 1. Is the repetition of politeness formulas more present in men's or women's speech? 2. Is the phenomenon of reduction in repeated politeness formulas more present in men's or women's speech? 3. Finally, is it possible to consider one of the two gender categories as more innovating than the other?
The corpus used for the current study is a corpus of 3325 politeness formulas of Italian speech (Pagliaro 2018) collected from VoLIP (De Mauro et al. 1993;Voghera et al. 2014) and C-ORAL-ROM (Cresti, Moneglia 2005) Italian spoken corpora. Especially, the VoLIP is an about 500,000 tokens corpus, organized in five sections: (a) face-toface conversations; (b) telephonic conversations; (c) communicative exchanges with constrained turn-taking; (d) unidirectional exchanges; (e) television or radio broadcasts. The Italian C-ORAL-ROM, instead, consists of about 300,000 tokens and its corpus design includes a first division between the informal register (50%) and the formal register (50%). The informal register is divided between family-private domain and public domain, both domains are divided into sections of monologue, dialogue and conversation. The formal register, instead, is divided into mass media (interviews, weather forecasts, news, documentaries, scientific press, sports and talk shows) and natural contexts (conversations on economics, law, conferences, political debates, professional explanations, religious sermons, political speeches, lessons). The formal register also includes the telephone domain divided between private conversations and human/machine interactions. 4 The VoLIP and the C-ORAL-ROM corpora are homogeneous from the gender point of view. Initially, items with single or multiple repetitions of a word are extracted from the corpus of politeness formulas. Then, politeness formulas with reduction are extrapolated.

Politeness Formulas with Repetition
The politeness formulas containing items with single or multiple repetitions of a word collected from the total corpus are 151 and they have been manually extracted. Table 1 shows the items in Italian, their translation into English and the number of tokens. The data are analysed through tree variables: the type of politeness formula, the speech situation and the gender of the speaker who uttered them. The classification of the type of politeness formulas and of the speech situation in which the items occur partially follows the taxonomy used in Pagliaro (2018, 128-9).
As regards the type of politeness formulas, the current study distinguishes among formulas of apologies, farewells, greetings, greetings to others (i.e. not addressed to the direct interlocutor), ironic farewells, minimizations, thanks and wishes as indicated in Table 2.
Concerning the speech situation, the present classification unifies and reorganizes the two similar organizations of the VoLIP (De Mauro et al. 1993;Voghera et al. 2014) and the C-ORAL-ROM (Cresti, Moneglia 2005) Italian spoken corpora, and adopts three parameters: face-to-face presence of speaker and addressee, freedom of turns taking and frequency of turns' succession (cf. Voghera 2017, 67). So that, it distinguishes among:

Repetition and Reduction. Two Contrastive Characters of Politeness Formulas in a Gender Perspective
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 99 Language, Gender and Hate Speech, 91-110 a. face-to-face conversations; b. telephone conversations; c. bidirectional communicative exchanges with constrained turn-taking (i.e. interviews, debates, classroom interactions, oral exams); d. unidirectional exchanges (i.e. lectures, sermons, speeches); e. telephone conversations broadcasted on radio; f. television or radio broadcasts.
The number of the tokens for each category is shown in Table 3. Table 4 indicates the number of tokens as produced by men or women.

Politeness Formulas with Combination of Repetition and Reductions
As regards the politeness formulas, it is firstly observed that the phenomenon of repetition can occur in combination with the phenomenon of reduction. So, the items with reduction identified through a perceptive and a spectroscopic analysis by PRAAT (Boersma, Weenink 2018) are extracted from the 151 politeness formulas containing items with single or multiple repetitions of a word. Figure 1 and 2 show the spectroscopic analysis of two example items. In particular, Figure 1 shows the analysis of the Italian informal farewell ciao ciao, 'bye bye' in its complete form, while Figure 2 shows the representation of the same item in a reduced form, with the final syllable of the second item reduced (i.e. ciao cià, 'bye bye').
The formulas with reduction are 62, so that they represent the 41% of the 151 politeness formulas with repetition.
With regard to the type of reduction, the items are divided into five categories. The first category includes the items with the reduction of the final syllable. These represent the larger group and include expressions such as, for example, the farewell ciao ciao (/ˈʧaoˈʧao/), 'bye bye', that becomes cià cià ([ʧaˈʧa]). The second category includes the items with an internal syllable reduced, like in the farewell arrivederci arrivederci (/arriveˈderʧiarriveˈderʧi/), ʻgoodbye goodbyeʼ, that becomes arrivederci arriderci ([arriveˈderʧiarriˈderʧi]). The third category comprises the items with the reduction of the final vowel, as in the English farewell bye bye (en. /bɑɪˈbɑɪ/; it. /baj'baj/) that becomes it. ba ba ([baˈba]), 'bye bye'. The fourth category includes the items with the reduction of the final vowel again and the final syllable reduced as in the apology ma scusa ma scusa (/maˈskuzamaˈskuza/), 'but sorry but sorry' that becomes ma scuz ma scu ([maˈskuzmaˈsku]). Finally, the fifth category comprises the items with an internal vowel reduced, such as the greeting buongiorno buongiorno (/bwonˈʤornobwonˈʤorno/), ʻgood morning good morningʼ, that becomes bongiorno bongiorno ([bonˈʤornobonˈʤorno]). Table 5 provides a summary of the data with the number of tokens for the different categories. The distinction among the different types of politeness formulas, the speech situation and the gender of the speaker who uttered them follows the same criteria used for the previous classification of the repetition. In particular, Table 6 shows the items divided into the different types of politeness formulas; Table 7 displays the different speech situations in which the politeness formulas occur; Table 8 presents the data as produced by men or by women. Finally, Table 9 offers a summary of the data.   In what follows we will present the results of the previous analysis, starting from the variables used to examine the two phenomena of repetition and reduction, i.e. the type of politeness formulas, the speech situation and the gender of the speaker who uttered them.
As regards the type of politeness formulas, the type mostly affected by the phenomenon of repetition is farewells. Indeed, farewells with repetitions represent the 78% of the items, followed by thanks (8%), apologies (7%), greetings (3%), and other types. Moreover, farewells are the type of politeness formulas mostly influenced by the phenomenon of reduction too, with a percentage of 93% of all the items. Also, the two phenomena usually both occur in the item ciao, used as a farewell in 114 tokens out of 151 items. Ciao is repeated once and more times and it is affected by reduction in 46 tokens out of 62. It is usually the final syllable that undergoes reduction. In this way, the combination of repetition and reduction creates new forms of the farewell ciao, like ciao ciao [ˈʧaoˈʧao] . Table 9 for more examples). According to Bazzanella (2013, 42) for the cases of ciao, but also for other politeness formulas, this means that the speakers are reacting to the fixity of formulaic and routine language and are looking for novelty and intensification, creating new forms of politeness formulas that produce new values of linguistic politeness. In fact, the modified forms of the item ciao can be generally present only at the end of a communicative exchange (as is the case with farewells), while the no-repeated and no-reduced forms of ciao are used both for greetings and for farewells. In other words, these modifications seem to be responsible for the specialization of certain forms for new meanings.
With respect to the speech situation, the results indicate that the two phenomena of repetition and reduction in politeness formulas are more present in telephone conversations. Namely, as regards repetition, telephone conversations have the 54% of the tokens, followed by television or radio broadcasts (18%), telephone conversations broadcasted on radio (16%), face-to-face conversations (9%) and other speech situations. Similarly, the phenomenon of reduction is more present in telephone conversations (63%), followed by telephone conversations broadcasted on radio (16%), television or radio broadcasts (11%) and face-to-face conversations (10%).
Generally, farewells are the most used politeness formulas, because speakers need to point out when the conversation ends more than they need to point out for other moments of the communicative exchange to occur. Also, when closing a telephone call, speakers tend to use more politeness formulas due to lack of face-to-face interaction patterns, especially facial and body gestures. This means that speakers may tend to modify, through repetition and reduction, the politeness formulas they use most frequently.
As regards the parameter of gender, the results of the previous analysis indicate that both repetition and reduction are more present in the speech produced by men. Indeed, politeness formulas with single or multiple repetition of a word are the 57% of tokens produced by men, while only the 42% of those produced by women. 5 Similarly, the 62 politeness formulas with repetition and affected by reduction are produced in the 56% of cases by men and only in the 44% of cases by women. Moreover, with regard to the phenomenon of the reduced variants, the data are confirmed by previous studies who claim that, generally, men reduce more than women due to phonetic characteristics (Byrd 1994;Bell et al. 1999;Keune et al. 2005;Strik et al. 2008).

Discussion
As the literature states, both repetition and reduction are widespread phenomena in language, and they affect politeness formulas too, which are marginal elements of the lexicon. Their co-occurence also represents the manifestation of two opposite tendencies through the modifications of illocutionary force of a given speech act within a communicative exchange. Indeed, on the one hand, the reduplication functions as an intensification phenomenon; on the other hand, the presence of phonetic reduced variants operates as a mitigation phenomenon (Labov 1984;Bazzanella, Caffi, Sbisà 1991;Bazzanella 2004;Bazzanella, Gili Fivela 2009).
Furthermore, the two phenomena are strictly related to the context. Repetition, in particular, is an iconic strategy in which the exact repetition always represents a more developed sign in terms of a semiotic and diagrammatic perspective. Regarding reduction, on the other hand, previous studies have claimed that segments may be weakened or even silenced due to articulation or phonological processes (Ernestus, Warner 2011;Ernestus, Hanique, Verboom 2015). For politeness formulas, it can be assumed that segments may be weakened or silenced due to both mechanisms. Namely, on the one hand, the reduction is due to the repetition and the segments are silent because they are hidden by the articulation of the surrounding segments and because of the production speed of two or more forms rather than one; on the other hand, it is possible to affirm that in specific contexts the speaker directly selects the reduced pronunciation variants.
Specifically, regarding the participants in the contexts and adopting the perspective of correlational sociolinguistics, it is then possible to consider the two sexes as variables for the analysis (Hultegren 2008; Dittmar 2010). And, if it is assumed that the presence of the two phenomena stands for a reaction against the fixity and immutability of formulaic language and for a disposition towards the renewal of traditional and conservative forms, it can consequently be stated that men are more innovative than women in their use of this aspects of language. In this sense, the issue seems to support the traditional thesis according to which men are more exposed to linguistic variation and innovation than women, who on their side would prefer conservativeness and the tradition of the forms (Berretta 1983;Eckert 1989;Coates 1993).
Nevertheless, Coates (1993, 185) affirms that linguistic variation echoes social variation and that it is therefore incorrect to affirm that either women or men are the linguistically innovative sex. Indeed, it would be better to claim that some linguistic change is initiated by female speakers and some by male speakers. Moreover, studies of gender and variation actually go beyond the description of who does what, and proceed into more worthwhile investigations on how systematic variation in language gives rise to social meaning (Meyerhoff 2014, 100). Also, from a sociophonetics perspective, as Podesva and Kajino (2014, 104) affirm, we need to remember that phonetic output is heavily influenced by physiology, as asserted by previous studies about reduction, but the roles of culture, social convention and gender ideology cannot be underestimated.
Also Ochs (1992) deals with the theme of the relation between language and gender, and the way this is mediated by social activities, and she suggests that three aspects of the language-gender relation must be taken into account. The first aspect is that the relation is non-exclusive, in the sense that various features of language may be used by both sexes, but it is possible that some features are preferentially employed more by one than by the other sex. The second aspect is that the relation is constitutive, because linguistic features may guide social meanings, which in turn help to establish gender meanings. The third aspect is that the relation is temporally transcendent, because societies establish norms against which men and women can verbally recontextualize the past and precontextualize the future.
It would therefore seem that the reaction to the fixity of politeness formulas characterizes both sexes but men to a greater extent than women, and that this relation is mediated by men's tendency towards innovation and women's tendency towards conservativeness as social meanings.

Conclusions
The aim of this study was to analyse the phenomena of repetition and reduction in a gender perspective, adopting the viewpoint of correlational sociolinguistics. Repetition and reduction are two pervasive aspects of human communication that can be variously affected by the social and the linguistic context. In this paper we analysed the two phenomena in the field of politeness formulas -a marginal part of the lexicon -through the variables of the type of politeness formula, the speech situation and the gender of the speaker who uttered the expressions in discussion. The analysis of the latter variable revealed that men tend to use more politeness formulas affected by repetition and reduction. The results were interpreted as a tendency to react to the immutability of the formulaic aspects that influence men's speech more than women's.