An Open Diachronic Dataset of Requests in Italian: Annotation, Challenges and a Case Study

(1) Introduction: context and motivation

Making requests lies at the very core of human interaction, revealing the interdependent and cooperative nature of everyday social experience. Rossi (2015, p.1) defines the need to appeal to others as “a primordial and pervasive component of social interaction”, highlighting a key feature of requests in interpersonal relations: reliance between individuals, understood as a structural dimension of our social existence. The ability to coordinate and guide others’ behavior constitutes the interactive fabric that makes social life possible, as well as efficient and organised, and seems to have a universal character (Drew & Couper-Kuhlen, 2014).

In addition to being an inherently social action, making requests is also a linguistic act, typically – though not always – performed through specific linguistic means. Languages provide a wide range of constructions and strategies to shape requests according to different socio-interactional variables, which has led Drew and Couper-Kuhlen (2014, p. 2) to refer to “social forms of requesting” as interacting with a range of linguistic forms, highlighting the complexities that can emerge from such combinations.

Requests are illocutionary acts in which a speaker communicates to a hearer a desire for the hearer to perform an action that benefits the speaker (see, e.g., Trosborg, 1995, p. 187). They are socially delicate actions, involving a complex interplay of demands and obligations: thus, speakers often feel compelled to render their linguistic expressions in a polite manner, softening their imposition and thereby reducing potential conflict and preserving relational balance. In more technical terms, according to Brown and Levinson’s seminal work (1987), requests are face-threatening acts that inherently challenge the addressee’s negative face – that is, their wish for autonomy and freedom from external impositions. In this light, speakers may opt for indirect (off-record) requests, drawing on a repertoire of linguistic and contextual resources – such as politeness markers, mitigation devices, minimizers, and supportive moves like justifications – to limit the impact of the imposition.

The aspects briefly outlined above – namely, the fundamental role of requests within our systems of cooperation and social solidarity, and the rich repertoire of linguistic strategies that languages have developed to perform this key action – make this speech act particularly compelling to explore from a variety of perspectives. As a result, requests have become one of the most extensively studied speech acts, especially from a cross-cultural point of view, but also through socio-interactional, acquisitional, and, to a lesser extent, diachronic approaches. Within this latter strand of research, studies focus primarily on English. Important studies examining the “old synchrony” of the English language are for example Kohnen (2000) on Old English and Culpeper and Archer (2008) on Early Modern English. A notable example of a diachronic study is Jucker (2020), who investigates the evolution of politeness norms in the history of English and demonstrates how these norms directly shape the linguistic realization of requests, illustrating how strategies of mitigation, indirectness, and deference vary across historical periods.

Studies on Italian, by contrast, are still very limited and thematically and/or chronologically circumscribed. Fedriani (2020) offers a first pilot study focusing exclusively on the mitigation of requests and based on a very small corpus of texts sampled at roughly two-century intervals. The doctoral dissertation by Parodi (2025), on the other hand, examines requests in Italian in the diachronic span from the seventeenth to the nineteenth century, primarily adopting a contrastive perspective, with Italian requests examined in systematic comparison with Spanish and Genoese.

Based on these premises, further research is needed to enrich our understanding of pragmalinguistic changes in the speech act of requesting across the long diachrony of Italian. To foster and support research aimed at filling this gap, we compiled a pragmatically annotated dataset of requests drawn from texts ranging from Old Italian to Contemporary Italian. We first present the dataset (Section 2), the annotation scheme and the annotation procedure (Section 3), and then discuss the main annotation challenges (Section 4). We also present a case study based on the dataset (Section 5) and outline the dataset’s potential applications (Section 6).

(2) Dataset description

The Diachronic Dataset of Requests in Italian contains 1,667 requests extracted from dialogues in Italian literary texts from the thirteenth to the twentieth century. For each request, the author, the title of the source work, and its century of publication are provided. Each request is pragmatically annotated into its constituent elements, following an annotation scheme based on Blum-Kulka et al. (1989) and described in Section 3.1.

(3) Method

(4) Sources of annotation disagreement

An examination of the cases of disagreement shows that, aside from occasional annotation oversights, divergences largely stem from intrinsic interpretative difficulties in the data. This is hardly surprising. As Weisser (2014, p. 84) notes, “[a]ny type of linguistic annotation is a highly complex and interpretive process, but none more so than pragmatic annotation”. A key reason lies in the lack of a one-to-one correspondence between pragmatic forms and functions: “the form–function mismatch of most pragmatic phenomena means that automatic assignment of tags will often lack precision, and manual implementation of tags (which is time- and resource-intensive) is unavoidable” (Rühlemann & Aijmer, 2014, p. 11). In addition, many pragmatic categories have fuzzy boundaries and cover functional domains that overlap with neighbouring categories. As we will discuss in this section, these properties frequently give rise to cases of annotator disagreement, stemming from ambiguity and from the availability of multiple plausible interpretations. We discuss here some illustrative examples, beginning with the cases of disagreement that emerged in the annotation of head acts, which are less than half of those in the annotation of modifiers (21 vs. 48). This in itself is an interesting point, to which we will return later.

One case of disagreement in the annotation of the head act is given in (1). The first annotator classified it as a direct request in the form of a want statement expressed by the verb intendere in the sense of ‘to wish/want’. The second annotator interpreted intendere as a performative verb (‘to declare’) and therefore classified the request differently.

• (1)   acciò che tua sorella senza dote non sia, io intendo che egli e non altri abbia questo beneficio che il re promette così grande per te […] (Boccaccio, Decameron, 14th c.)

        ‘So that your sister may not remain without a dowry, I intend that he, and no other, should receive this great benefit that the king promises for you.’

In this case, the interpretative difficulty stems from polysemy. The verb intendere already exhibited the two meanings (‘to wish/want’ and ‘to declare’) in fourteenth-century Italian (see GDLI, s.v.). Closer examination of the broader context, together with in-depth discussion among the annotators, led to the conclusion that, in this instance, intendere conveys a want-based request, a reading supported by the interlocutor’s subsequent compliance in the following stretch of text.

As noted above, the level of disagreement in the annotation of modifiers proved to be higher, and this functional domain presented a wider range of annotation challenges, which were also more complex. A recurrent type of divergence concerned expressions such as the one highlighted in example (2), annotated by both authors as an external modifier, but interpreted by one as expressing motivation and by the other as a downgrader.

• (2)   Per che io vi priego, per cotanto amore quanto è quello che io vi porto, che voi non neghiate il vostro verso di me […] (Boccaccio, Decameron, 14th c.)

        ‘Therefore I beg you, for the great love I bear you, that you do not deny me yours [love] toward me’

The categories of “motivation” and “downgrader” may indeed partially overlap, because providing a reason or justification for the request can also function as a mitigating strategy that reduces its impositive force. Adopting a prototypical approach, we agreed to annotate such expressions, which are particularly common in Old Italian, as downgraders, since their core function appears to be the modulation, in terms of mitigation, of the request, while the motivational component remains more implicit.

Another case of divergence concerned the classification of the pragmatic marker per carità (‘for heaven’s sake’) in example (3), interpreted by both annotators as an internal modifier but considered a downgrader by one and an upgrader by the other.

• (3)   Per carità, lasci perdere quel “primo consigliere”. Non ci tengo (Fo, Morte accidentale di un anarchico, 20th c.)

        ‘For heaven’s sake, drop that “chief adviser” business. I don’t care about it.’

The pragmatic marker per carità has historically been used in Italian both in polite and impolite requests. To resolve this disagreement, we relied on the existing literature on the topic. In particular, according to Paternoster (2023), who adopts a combination of methods, both qualitative (focusing on the mention of the formula in contemporary etiquette manuals) and quantitative (examining the use of the formula in a diachronic corpus), impolite uses of per carità emerged in the nineteenth century and are predominant today, often being used to express disagreement – as in (3). For this reason, we ultimately annotated this twentieth-century occurrence as an upgrader.

A final noteworthy case is illustrated in example (4), concerning come sia tuo piacero:

• (4)   priegoti che mi doni oro o argento o robbe, come sia tuo piacero

        ‘I pray that you grant me gold, silver, or other items, as it pleases you’ (Anonymous, Novellino, 13th century)

One annotator did not mark this expression, interpreting it purely propositionally as attached to argento o robbe – that is, ‘silver or other items, as you wish’. The other annotator, however, interpreted come sia tuo piacero as an instantiation of one of the pragmatic rules for polite expression in Lakoff’s politeness model, namely “give options”, which entails phrasing utterances so as to allow the addressee flexibility in responding to or interpreting the request (Lakoff, 1975). Following discussion, it was decided to annotate it according to this line of reasoning, classifying it as an external downgrading modifier.

Space constraints prevent us from discussing further cases of interpretative disagreement. These examples nevertheless highlight the challenges involved in classifying pragmatic modifiers, whose context-dependent and multifunctional nature makes their identification particularly susceptible to individual analytical judgments.

An in-depth discussion of these cases allowed us to clarify problematic instances and then proceed independently with the annotation of the remaining requests, consulting each other whenever further ambiguities arose.

(5) Results: diachronic trends in direct vs indirect requests

In this section, we provide an illustrative example of research based on the dataset just described. In particular, as a testing ground, we have decided to focus on one of the most widely studied aspects of requests across different languages and historical periods, namely the degree of (in)directness of requests. Blum-Kulka et al. (1989) show that, in the five languages examined in their study (English, French, German, Hebrew, and Spanish), conventionally indirect strategies are by far the most frequent (between 58% and 82%, depending on the language), followed by direct strategies (between 10% and 40%) and non-conventionally indirect strategies (between 2% and 8%). Culpeper and Archer (2008), analysing trial records and drama texts from Early Modern English (1640–1760), observed, by contrast, a marked predominance of direct strategies (about 73%; note, however, that their category requests is used in a broader sense, also encompassing commands; consequently, their findings are not directly comparable to ours). In any event, this discrepancy led us to explore what happens in the history of Italian with regard to the parameter of (in)directness. Let us now thus see what the data from our dataset show from this perspective.

We organised the data into six consecutive periods, partially following the periodization adopted in the MIDIA diachronic corpus of Italian (D’Achille & Grossmann, 2017): Period 1 (13th c.–1375), Period 2 (1376–1532), Period 3 (1533–1691), Period 4 (1692–1840), Period 5 (1841–1947), Period 6 (1948–1978). Figure 1 shows, for each period, the proportion of direct, conventionally indirect, and non-conventionally indirect requests relative to the total number of requests in that period.

Figure 1

It can be observed that up to the mid-20th century requests are overwhelmingly direct. This result matches the trend also observed by Culpeper and Archer in their analysis of Early Modern English. In the second half of the 20th century the situation changed: conventionally indirect requests increased substantially, reaching 30% of the requests in Period 6. Non-conventionally indirect requests, by contrast, remain consistently marginal, as is also the case both in Blum-Kulka et al. (1989) and in Culpeper and Archer (2008).

It therefore appears that, whereas in historical data from Italian (and English) requests were predominantly formulated directly, in contemporary Italian (as in other present-day languages) the gap between direct and indirect strategies has narrowed. The trend that emerges here, although limited to the dataset used, seems to suggest a developmental tendency that may be cross-linguistically valid, at least for European languages that have shared similar historical trajectories and cultural developments. For a more in-depth, culturally and historically informed account of the reasons underlying this evolutionary tendency, we refer the reader to Fedriani and Strik-Lievers (in preparation).

(6) Applications and potential reuse scenarios

Building on the illustrative case study on (in)directness in Italian requests presented in Section 5, we now turn to other potential reuse scenarios of the dataset, which, in our view, support multiple lines of research across different fields.

First, it can be used to investigate a wide range of pragmalinguistic phenomena from a historical pragmatic perspective, including the use of performative verbs, (im)politeness strategies, mitigation and intensification, as well as diachronic changes in support moves – for instance, the increasing conventionalization of ritual apologies or the gradual pragmaticalization of Italian politeness markers over the centuries. A related area of application concerns the study of dialogic interaction patterns, such as the use and evolution of terms of address and endearment, as well as variation and change in attention-getting forms and their communicative functions.

In stylistic and literary studies, the dataset allows the identification of recurring stylistic devices employed by specific authors. For example, earlier research on the requests included in the dataset highlighted Andreini’s strong preference for the discourse marker di grazia in his work Lo schiavetto, which contributes to a marked increase in the use of discourse markers in seventeenth-century requests included in the dataset (Fedriani & Strik-Lievers, 2025). At the same time, the dataset provides a valuable resource for examining authorial preferences in dialogue construction and characterization.

In computational linguistics and NLP, it can support the training of language models on pragmatically annotated data and the development of automated systems for request detection and classification. More broadly, this aligns with recent work exploring the use of LLMs for pragmatic annotation tasks, such as Yu et al. (2024) on apologies, Brocca et al. (2026) on cancellations after accepting an invitation, and Bianco et al. (2026) on humour.

Finally, the dataset also has educational and digital humanities applications, serving as a resource for courses in pragmatics, stylistics, and historical linguistics, as well as for pragmalinguistic annotation exercises – an approach the authors have implemented in research seminars with PhD candidates and in MA-level courses.

Author Contributions

Chiara Fedriani: conceptualization, data curation, investigation, methodology, writing of Sections 1, 5, and 6.

Francesca Strik-Lievers: conceptualization, data curation, investigation, methodology, writing of Sections 3 and 4.