Introduction

Within the scope of the Spoken Portuguese Grammar Project (Castilho, 1990), a group of researchers responsible for studies on the textual construction was created and led by Ingedore Grünfeld Villaça Koch. This group faced the challenge of defining theoretical-methodological foundations for textual analysis from a textual-interactive perspective, culminating in the elaboration of a theoretical proposal for textual analysis, known as Textual-Interactive Perspective or Textual-Interactive Grammar. In this challenging work of delimiting principles that are specific to the analysis of texts and also pertinent to the level of textual organization of the language, the definition of a textual analysis unit was a fundamental task, consistent with the proposed theoretical foundations.

In this sense, when studying the macrostructure of texts, the group found that a basic process of textual construction is topicality, that is, the organization of texts into discourse topics, since, “throughout a communicative event, the interlocutors focus their attention on certain themes, which constitute the focus of verbal interaction” (Jubran, 2015a, p. 28, our translation).¹ Guided by this observation, this group established an analytical category of texts, the discourse topic, understood, in general, as “‘about’ what is spoken” (Jubran, 2006a, p. 35, our translation),² a category that allows the division of texts into so-called topic segments. Each topic segment is a textual portion that materializes the abstract category of the discourse topic and finally constitutes the theory’s unit of analysis.

Thus, a methodological principle defined within Textual-Interactive Grammar (TIG) is that textual phenomena must be analyzed in relation to topicality, which makes distinguishing topic segments of a text an elementary procedure in textual-interactive analysis. Evidence of the need to operate with topic segments in the description of textual phenomena is the study of parenthetical insertion, one of the text construction processes recognized by TIG. As Jubran (2015b) points out, parenthetical insertion occurs in the context of the topic segment, corresponding to a brief deviation from the discourse topic, which does not cause the establishment of a new topic within the one which was in progress, thus not allowing the delimitation of a new topic segment. Another example of the methodological importance of recognizing the topic segments of a text to study a textual-interactive phenomenon is the work with discourse markers (DMs), a set of linguistic resources responsible for sequencing textual portions and marking the presence of the instance of enunciation in the utterance (Risso, 2015; Urbano, 2015). In the analysis of the functioning of linguistic expressions as DMs, an essential step is to delimit the text in topic segments, since this delimitation is crucial for an accurate establishment of the behavior of expressions in relation to the features of topic sequencer, sentence sequencer and non-sequencer, features that are crucial for the characterization of an expression in terms of its performance as DM.

In this context of delimiting methodological procedures for the textual study linked to the category of the discourse topic according to TIG, we aim, in this paper,³ to present methodological contributions to the analysis of texts based on this category, which will imply contributions to TIG itself. More specifically, our work comprises the following objectives: (i) to argue that the construction of texts may include the establishment of meta-discourse topics which are centered on the definition and indication of the discourse topics to be developed in the text; (ii) to discuss and propose a methodology for naming the referents which, in a text, assume the status of discourse topics. The material explored in our discussions comes from descriptions, experience narratives and opinion reports collected from the IBORUNA Database Census Sample (Gonçalves, 2007), which brings together linguistic samples of Brazilian Portuguese spoken in the Northwest of the São Paulo state.

The paper is organized as follows: after this introduction, in the next section we present TIG and its conception of discourse topic; in the sequence, we explain the notion of meta-discourse and its performance in topic construction, which will help us understand how a topic can be recognized as a meta-discourse topic; after that, we address the referencing process and the ways of activating textual referents, seeking a theoretical basis for the discussion on ways of identifying referents that are topics; reaching our discussion of data, in the following section we discuss the two issues involved in our objective, offering methodological contributions to textual-interactive analysis; finally, we present the conclusions and references, in that order.

Textual-Interactive Grammar and the notion of discourse topic

TIG is a branch of Text Linguistics, being, therefore, an approach that takes the text as its object of study, focusing on its interactional dimension. As Jubran (2007) explains, TIG understands the text as a dynamic socio-communicative unit that mobilizes, in addition to linguistic-textual factors, interactional knowledge that takes place in the game of communicative action carried out by language and which is inscribed in the textual materiality itself. In a grammatical approach to text, such as TIG, it should also be noted, as postulated by TIG, that the rules of textual construction are not equivalent to those at the sentence level, once the text presents properties based on its own constitutive order. In line with the conception of text in question is the notion of language of TIG, understood as a form of verbal action, through which at least two interlocutors perform communicative tasks of exchanging representations, goals and interests, in the context of a discourse space that is always directed to communication partners (Jubran, 2007).

From these conceptions of language and text, which emphasize the importance given to interactional factors in textual construction, a set of guiding principles of TIG emerges (Jubran, 2015a, 2007). The first one is that the phenomena considered in it have their properties and functions defined in use, in concrete situations of interlocution, co-involving the circumstances of enunciation. A second essential principle for TIG is that interactional factors are constitutive of texts and inherent to linguistic expression, with a natural presence of pragmatic data in textual processing, which are shown on the textual surface, a principle that largely resumes the notion of text as a socio-communicative unit based on interactional knowledge inscribed in the materiality of the text.

Assuming the intrinsic presence of pragmatic factors in the construction of texts, TIG, in the analysis of the phenomena involved in textual construction, does not dichotomize textual and interactive functions, but combines them, according to the dominance of one or the other, which already demonstrates a third theoretical principle of the approach, the gradient principle. According to this principle, a given textual construction procedure sometimes acts dominantly in the informational organization of the text, sometimes dominantly in the enunciative activity, signaling the domain of the interactional function, sometimes in intermediate classes, throughout which there is a progressive projection of the textual function on the interactional and, conversely, a progressive projection of the interactive function on the textual one.

As mentioned in the introduction, TIG establishes the discourse topic as an analytical category, understood as what is spoken about (Jubran, 2006a, Pinheiro, 2005), as the basic semantic informative content of a text (Bernárdez, 1982; Brown; Yule, 1983; Van Dijk, 1980). More specifically, it is considered that the topic manifests itself in a text through utterances formulated by the interlocutors regarding a set of referents related to each other and relevant at a certain point in the message (Jubran, 2006b, 2015c; Pinheiro, 2005).

From the observation of the convergence of a group of utterances to the same set of discourse objects, two particularizing properties of the discourse topic are defined – centering and organicity. The property of centering covers the features of semantic interdependency, relevance and delimitation. According to Jubran (2015c), semantic interdependency consists of the interdependence relation between textual elements, established by cohesive mechanisms of sequencing and referencing, which promote the integration of these elements into a specific set of explicit or inferable discourse objects, established as the target of the interaction. Relevance, in turn, is related to the prominence of textual elements in the constitution of this referential set, which are projected as focal in the interaction. Finally, the feature of delimitation concerns the concrete location of this set of discourse objects at a certain point in a text, based on the semantic interdependency and relevance of its elements.

The example in (1), from Jubran (2015c), shows the features of centering, as well as the notion of discourse topic as an abstract category and the conception of topic segment.

According to Jubran (2015c), in (1), semantic interdependency is certified in the construction of a reference set related to the professions of L1’s husband. The integration of the constituent elements of this set of discourse objects is established, for example, by the semantic association between lexemes such as “escritório/office”, “carreira/career”, “advogar/advocate”, “lecionar/teaching”, “procurador/prosecutor”, “professor/teacher”, which circumscribe the theme “profissão/profession”. The topic relevance, on the other hand, results from the focal position assumed by topic referents, which can be verified in this case by observing the themes and sentential rhemes. The themes often refer to the element “marido/husband”, putting this discourse object in prominence, and linked to these themes are rhemes that express the dominance of the topic “profissões/professions”, as in the excerpt “ele se dedica muitíssimo a ... tanto à carreira de procurador como de professor/he is very dedicated to ... both to the careers of prosecutor and teacher”. After identifying a set of semantic interdependent and relevant utterances about the professions of L1’s husband, we can find, in the materiality of the text, a textual unit that concretizes the discourse topic pointed out, which already demonstrates the third feature of centering, delimitation. Therefore, this is how it is said that the discourse topic is related to what is spoken about, as in (1), in which the professions of L1’s husband are thematized.⁴ This is also how it is assumed that the abstract category of discourse topic is materialized by a topic segment, a textual unit that corresponds, in (1), to the entire excerpt of the example, which materializes the topic “Atividades profissionais do marido de L1/professional activities of L1’s husband”, from the beginning of the documenter’s speech to the end of L1’s speech.

The second particularizing property of the discourse topic, organicity, is manifested, according to Jubran (2015c), by relations of topical interdependency that are established on two planes – linear and hierarchical. In the hierarchical plane, there are dependencies of superordination and subordination between topics that are implied by the degree of comprehensiveness of the subject discussed in a text, while on the linear plane, there are inter-topic articulations in terms of adjacency or interpositions of topics in textual linearity. Figure 1 exemplifies the hierarchical plan of topic organicity, which will help us to understand, in the next sections, the functioning of meta-discourse topics. We will not include examples that demonstrate topic linearity because we will not address, in our analyses, inter-topic relations in terms of their distribution in textual linearity.

Figure 1
– Topic hierarchy

Still on the concept of discourse topic, in addition to the notion of topic proposed by Jubran (2006a, 2006b, 2015c), it should also be recognized, as in Garcia (2022), that every discourse topic can be assumed as a referent.

Regarding the notion of referent, Lyons (1977b) points out that it is common to say that words or expressions, but not concepts, refer to things. In this context, dealing with the reference, the author states that when we use a simple descriptive sentence, it is often appropriate to say that what we are doing involves talking about someone or something, about a particular entity (or entities), or group(s) of entity(ies) in the world. For example, in a situation where an expression such as “Napoleon is a Corsican” is used to make a statement, the speaker is said to have referred to a certain individual (Napoleon) by the use of a referential expression. If the reference is successful, the referential expression will allow the listener to correctly identify the individual in question, who is the referent.⁵ Still for this author, in sentences with only one referential expression, the expression used to refer to what we are talking about (the referent) is typically the subject of the sentence (Lyons, 1977b).⁶ Regarding this notion Brown and Yule (1983), based on Lyons (1977b), say the following: “The term reference can […] be […] reserved for that function whereby speakers (writers) indicate, via the use of a linguistic expression, the entities they are talking (writing) about.” (Brown; Yule, 1983, p. 205).

Based on this conception of referent, it is admissible to recognize the discourse topic as a referent because the topic, as the theme of a text – “theme” conceived as what is spoken about in a discourse (Brown; Yule, 1983) – can also be assumed as something about the world, (re)constructed in a text and targeted by the textual-interactive process. This referential view of discourse topic, presupposed in essential works for the definition of the category (Jubran et al., 2002; Jubran, 2006a, 2006b, 2015c), can be attested by means of the example in (2).

The topic segment in (2) composes a text in which the speaker describes different parts of his ranch, such as the bedrooms, a topic focused on in the given example. The centering on this topic can be verified by the semantic interdependency established through the semantic relation established between phrases such as “quarto/bedroom”, “cama de casal/double bed”, “[cama] de solteiro/[single]bed”, “guarda-roupa/wardrobe” and “quartinho/smal bedroom”, delimiting the theme “quartos do rancho/ranch bedrooms”. The topic relevance, on the other hand, can be attested by the explicit activation of the speech object “quarto/bedrooms” at the very beginning of the topic segment (“aí nos quarto”, L. 1/so in the bedrooms),⁹ indicating the part of the ranch to be focused on in that point of the text, as well as by the recategorization of this referent throughout the topic segment, mainly by phrases such as “um quarto/a bedroom” (L. 1, 4, 6), “o outro/the other” (L. 2), “o quarto/” the bedroom(L. 2-3), “o único quarto/the only bedroom” (L. 5), “o outro quarto/the other bedroom” (L. 6) and “o quartinho/the small bedroom” (L. 7). Once we identify a set of semantic interdependent referents which are relevant, at a certain point in the text, concerning the ranch bedrooms, this discourse object (the ranch bedrooms) can be recognized as a discourse topic.

Based on this topic analysis, the case in (2) attests that a discourse topic can be seen as a referent because it is about the entity of the world “Quartos do rancho do informante/Informant’s ranch bedrooms”, reconstructed in textual processing, that it is spoken about in the topic segment. In other words, it is about an entity (that is, a referent) – “Quartos do rancho do informante/ Informant’s ranch bedrooms” – that a centering is constituted (that is, a topic is established); hence the understanding that a topic is always a referent. As a referent, the topic can then be identified by three basic procedures, as discussed in the next sections.

Meta-discourse in topic construction

According to Risso (1999), the linguistic strategies for expressing meta-discursivity are among the multiple verbal procedures that signal the instance of enunciation in textual structure, showing, on the utterances’ surface, a self-reflexive movement that makes the discourse unfold on itself, in a process in which it refers to its own discursive “doing”. Thus, the author argues that the basic particularizing property of meta-discourse is self-reflexivity, as attested by the example in (3), taken by Risso (1999) from an interview with the then leader of Brazil’s Landless Workers’ Movement (MST), promoted by TV Cultura, in 1997, in the Roda Viva show.

As Risso (1999) explains, in (3) the meta-discourse segments in bold highlight the verbal act requested to the interviewee (enumerar/to list), labeling it as a specifying act of the proposed topic (mudanças sociais/social changes) and defining the format, not necessarily detailed (em termos genéricos/in general terms), of the required statements. Thus, by referring to what is going to be said, the discourse, simultaneously, is established as an event and as an object of mention, which allows us to say, as we mentioned, that the meta-discourse refers to the discursive process itself.

As for the marks of meta-discourse segments, Risso (1999) points out that the status of meta-discourse is usually marked, within a text, by verbal expressions, such as “say”, “ask” and “answer”, or nominal expressions, such as “statement”, “affirmation” and “response”, which refer to the verbal act. In addition, according to the author, it is possible to have an ellipse of metalinguistic terms, which can be recovered with the support of data from the communicative event. Thus, excerpts such as “Segunda coisa: o governo precisa priorizar a reforma agrária/ Second thing: the government needs to prioritize land reform”” or “Senhor líder do MST, o seguinte.../ Mr. leader of the MST, it’s this.” can be decoded as “Second thing to be said” and “I want to ask you this”.

Characterizing different modalities of meta-discourse, including the one involving the text construction scheme, Risso and Jubran (1998) show that meta-discourse prefacer can indicate the proposition of discourse topics to be addressed by the speakers, as we see in the example in (4), in which the bold excerpt, said by the documenter of a dialogue between two informants from the Cultured Linguistic Urban Norm Project (NURC), suggests that its subsequent segment acquires topic status.

In the same vein, Jubran (2022) recognizes the reference to topic structuring of texts, in terms of textual assembly and progression, as a type of meta-discourse, which can: (i) indicate the discourse topic to be addressed; (ii) mark the text composition scheme; (iii) signal the introduction and completion of a discourse topic; (iv) signal interruption, insertion of parentheses and topical resumption; (v) indicate the discourse status of a text fragment, such as summary, thesis and definition. An example of this modality of meta-discourse is the case in (5), in which it is possible to observe not only the naming of the topic to be developed (o problema da seca/the problem of drought), but also the indication of the structure of the topic development, organized into two parts (duas linhas/two lines).

Although there is relationship between topic structuring and meta-discourse (in the sense that the reference to topic structuring is a modality of meta-discourse), it is necessary to recognize, according to Risso (1999), that meta-discourse does not directly participate in the set of discourse objects involved in topic propositions, being an important factor in the pragmatic anchoring of informational content, thus marking the instance of enunciation in textual construction, as it was said at the beginning of this section. This aspect of meta-discourse can also be noted in the work of Jubran (2022), who states that, “in the process of metalinguistic or meta-discourse labeling, objects-of-discourse do not concern topical elaboration” (Jubran, 2022, p. 229, our translation),¹⁰ because they consist of: (i) linguistic activities, designated by words such as “description”, “explanation”, “report”, “comment” etc.; (ii) mental processes, designated by “analysis”, “evaluation”, “opinion” etc.; (iii) attributes of the scope of language or text, meta-linguistically designated by “sentence”, “question”, “word” etc.; (iv) illocutionary acts, designated by lexemes such as “order”, “promise”, “advice” and “affirmation”, as we see in (6).

According to Jubran (2022), in (6) the expression “a afirmação/the statement” anaphorically refers to Carlos Menen’s speech without reference to its propositional content, as it takes it as a speech act, categorizing it by its illocutionary force of assertion. Thus, the author explains that the highlighted anaphoric expression focuses on the plane of enunciation, leaving its functionality in the informational flow in the background, considering that it denominates the verbal action practiced by the enunciator of the cited discourse and, at the same time, predicates it by its asseverative content. Similarly to these considerations of Risso (1999) and Jubran (2022) about the fact that meta-discourse does not act directly on topic construction, in the data analysis section we will discuss the topics we call “meta-discourse”, seeking to show that, in these topics, the informational development of an ongoing text decreases, while the interactional function of this text relatively increases, which corroborates the gradient principle of TIG.

Referencing and ways of activating textual referents

According to Koch (2015, 2018), referencing is a discursive activity through which objects of discourse are constructed and reconstructed, an assumption that implies a non-referential view of language, which moves away from the idea of direct correspondence between words and the external world. The author addresses the issue in the following terms:

Based on this conception, which indicates a shift from the notion of “reference” to that of “referencing” in Textual Linguistics studies (Mondada; Dubois, 2003), Koch (2018) explains that our brain does not photograph or mirror the extralinguistic world, so that our way of seeing and saying the real does not coincide with the real. As the author attests, our mind actually re-elaborates the sensory data for the purpose of apprehension and understanding and this re-elaboration takes place essentially in the discourse, in a process that is not subjective and individual, but the re-elaboration must obey restrictions imposed by cultural, social, historical conditions and, also, by the conditions of processing the use of the language. In this sense, Koch (2018) argues that discourse builds what it refers to, at the same time that it is subordinate to this construction.

In the context in which referencing is seen as a process of construction and reconstruction of discourse objects, which are not in a specular relation with the extralinguistic world, Mondada (2001) understands that discourse objects are entities interactively and discursively produced by the interlocutors throughout enunciation, being placed, delimited, developed and transformed in and through the discourse that they belong to. Following this concept, the author also adds that objects of discourse do not preexist, nor are they fixed, but emerge and develop gradually in the discourse dynamics.

According to Koch (2018), every discourse builds a representation that functions as a discourse memory, which is fed by the discourse itself, and the successive stages of this representation are responsible, at least partially, for the selections made by the interlocutors, especially with regard to referential expressions. Thus, in the constitution of discourse memory, the author identifies the involvement of three basic operations, which are established as referencing strategies: the construction/activation of discourse objects, through which a textual “object” not previously mentioned in an ongoing text is introduced into this text, with such an object being prominent in the textual construction; the reconstruction/reactivation of discourse objects, through which a node already present in the discourse memory is reintroduced into the operational memory, with the discourse object remaining in focus in the text; the defocus/deactivation of discourse objects, in which a new discourse object is introduced, starting to occupy a focal position, but the object removed from focus can return to the focal position at any time of the textual construction.

In view of the second purpose of this paper, which is to discuss a methodology of topic naming, showing how the referents that are topics are identified in a text, we present, in the continuation of this section, some ways of activating objects of discourse, which will support the discussion of data pertinent to this second objective of the work (for this reason, the other referencing strategies are not focused here).

In this regard, according to Silva and Cortez (2020), the most classic mode of referential introduction is through the use of a referential expression, understood, according to Cavalcante, Custódio-Filho and Brito (2014), as a linguistic structure used on the surface of the text to formally represent a referent, which, in general, is of substantive nature, being formed by a noun or by a name in a substantive function. In the example (7), from Cavalcante (2003), the phrases “um homem/a man”, “na mesa/on the table” and “uma mulher/a woman” function as referential expressions, activating, according to the author, referents without any type of association with elements of the discourse context or the immediate situation of communication.

In addition to the referent being formally represented by a referential expression, being introduced without association with the discourse context or with the communication situation, it can also emerge and be activated through an implicit referencing process, based on an indirect relation with the cotext, constructed inferentially, as it is the case of the indirect anaphora, according to Koch (2018).

According to Marcuschi (2022), indirect anaphors are usually constituted by defined and undefined nominal expressions and pronouns that are interpreted referentially without any type of explicit (or subsequent) antecedent in a text. In addition, the author explains that this phenomenon constitutes an endophoric strategy of activation of new referents, motivated or anchored, therefore, in the textual universe, and not a reactivation strategy of already known referents, which reaffirms its operation as an implicit referencing process and demonstrates its functioning in the continuation of the global referential relation. The example in (8) illustrates an occurrence of implicit referencing by indirect anaphora, activating the pronoun “eles/they” a new referent, based on the previous element “pescando/fishing”.

In this context of implicit construction of referents, it is also possible that the referential activity occurs without a noun phrase that explicitly represents the object of discourse established, as happens in (9), analyzed by Cavalcante (2011).

According to Cavalcante (2011), in (9), there are discourse objects that are constructed without any explicit mention of them, such as the object “entrevista de emprego/job interview”. In this case, there are other elements that reveal the establishment of the referent, such as the beginning formula of the interview – “antes de começarmos/before we start” –, the allusion to a previous job, the social deixis “o senhor/sir”, indicating the form of respectful and also formal treatment. These indications are added to the common knowledge of what is the communicative ritual of job interviews, favoring the establishment of the identified discourse object.

Another type of activation of discourse objects is through the process of nominalization, which, according to Koch (2018), is carried out from related elements of the cotext, as well as indirect anaphors. It consists of referring, through a noun phrase, to a process or state signified by a proposition that previously did not have the status of an entity. Thus, the introduction of this new referent encapsulates and labels the information present in the previous or subsequent cotext, moving from the level of the sentence to that of the noun phrase (Koch, 2008). The excerpt in (10) demonstrates a case where nominalization is prospective, summarizing and labeling all subsequent information.

Dealing with nominalization from a functional perspective, Dik (1997b) understands it as a type of embedded construction that has one or more properties in common with a primary nominal term. In (11a) and (11b), from Camacho (2011), we show a case of nominalization, pointing out that this process occurs through formal and semantic adjustments of the verbal predication embedded in the nominal term.¹²

According to Camacho (2011), the following formal and semantic adjustments occur in (11a) and (11b), which are very common in the nominalization process:

For the discussion we propose in this paper, it is important to emphasize that, despite the formal and semantic adjustments of the verbal predicate into a nominalization, nominalization preserves, by inheritance, syntactic-semantic features of the verbal term, such as the semantic category represented by (11a), state-of-affairs,¹³ which is the same designated by (11b).¹⁴ This is an evidence that the transformation of a predication into a referential expression, through a nominalization – a transformation that allows a referent to be recognized – can represent the fundamental semantic content of that predication,¹⁵ and it is, therefore, an effective procedure for naming discourse topics, as we will discuss in the next section.

Methodological contributions to topic analysis

As established in the introduction, in this paper we aim to offer methodological contributions to the textual analysis practiced in TIG, a modality of analysis based on the category of the discourse topic. To this end, we will discuss two specific issues, which are: (i) the presence, in certain texts, of a meta-discourse topic centered on the definition between the interlocutors of the discourse topic to be established in the text; (ii) the methodology for naming the referents that are discourse topics, which involves, as we will show, three basic procedures.

Also, we will analyze data extracted from descriptions, experience narratives and opinion reports from the IBORUNA Database Census Sample (Gonçalves, 2007),¹⁶ which brings together linguistic samples of the variety of Brazilian Portuguese spoken in the northwest region of the São Paulo state, collected between 2004 and 2007. In IBORUNA, descriptions, experience narratives and opinion reports are predominantly monological texts, but there are some moments of explicit interaction between the documenter and the informant, which highlights the textual-interactive processing of these texts, as can be seen during our discussion of data. From the Database, we then selected 10 descriptions, 10 experience narratives and 10 opinion reports,¹⁷ chosen to cover the diversity of informant types represented in IBORUNA.¹⁸ We decided to examine descriptions, experience narratives and opinion reports because these texts can be considered representative of predominantly descriptive, narrative and argumentative textual genres, respectively, which enables us to make reflections based on the investigation of texts of a varied typology.

This typological diversity includes a considerable part of the traditionally recognized textual types (narrative, descriptive, argumentative, expository and injunctive – cf. Koch and Fávero, 1987) and, thus, can demonstrate different forms of manifesting the referent with topic status in the text, subsidizing, in a more complete way, our task of identifying a list of procedures for naming topics and thus cooperating so that this list is able to account for diverse situations of textual construction.

Regarding our analysis methodology, we used the topic analysis method (Jubran, 2006b, 2015c), which, based on the particularizing properties of the discourse topic explained previously – centering and organicity –, makes it possible to identify the discourse topics of a given text. Thus, the property of centering, characterized by the features of semantic interdependency, relevance and delimitation, supports the analyst in identifying the topic segments of a text and its discourse topics. On the other hand, the topic property of organicity guides the verification of interdependence relations between the discourse topics of a given text. The examples presented in (1) and (2) allow us to glimpse how we applied the topical analysis method in our study.

Having made these brief explanations about our analytical procedures, we discuss the recognition of meta-discourse topics based on the example in (12), a description.

As we see in (12), lines 1-4 constitute a self-reflexive excerpt. In this segment, the particular focus is precisely on the definition, between documenter and informant, of the central discourse topic to be established in the text, the topic “Casa da informante/informant’s house”, which is evidence that, at this point in the description, the discourse unfolds itself, being a reference of itself, a fact that signals self-reflexivity, a fundamental property of meta-discourse, as Risso (1999) points out. In other words, the excerpt in 1-4 is a meta-discourse topic segment not only because it indicates the central discourse topic to be addressed, which is explicitly announced in “eu vô(u) falá(r) da minha casa/I will talk about my house” (L. 4), but also because, by indicating the central topic, it provides clues about the topic structure of the text and, thus, about the textual construction itself, considering that the organization of the text into topics is seen in TIG, as we affirm, as a basic process of textual construction.¹⁹

The meta-discourse status of the topic segment in lines 1-4 can also be attested by the relevance that, in this excerpt, the expressions “descrevesse/to describe” (L. 1), “descrevê(r)/to describe” (L. 3) and “falá(r)/talk” (L. 4) assume, which can be understood, from Risso (1999), as signaling the meta-discourse, since they refer to the verbal act. In addition, in this first topic segment of the description, the central topic “Casa da informante/the informant’s house” is not yet properly developed, which begins to be done, with focus, on line 5, when the informant starts to build centering on the lower floor of the house (L. 5-10) and, subsequently, when she focuses on the upper floor of the residence (L. 11-15). The fact that the topic segment in lines 1-4 does not focus on the development of the topic that runs through the entire text may highlight the meta-discourse character of this topic segment, given that, as can be seen from Risso (1999) and Jubran (2022), in the meta-discourse the referents do not act directly on the topic construction.

Regarding this direct non-participation of the meta-discourse topic in the topic construction, it should be recalled that Risso (1999), as discussed previously, highlights that meta-discursivity is one of the different verbal procedures that signal the instance of enunciation in the textual structure, being an important factor in the pragmatic anchoring of informational content. In this same direction, Jubran (2022) points out that meta-discourse expressions such as the anaphora “a afirmação/the statement”, in (6), leave their functionality in the informational flow of the text in the background, shifting the focus to the plane of enunciation. Therefore, as the gradient principle of TIG predicts that a textual construction procedure always operates simultaneously in the informational organization of a text and in the enunciative activity, and one of these planes can often be dominant, we understand that the meta-discourse topic in (12) can be seen as a topic in which the textual function decreases and the interactional function increases.²⁰ It is based on this increase in the interactive function in the meta-discourse topic that we interpret that topics of this nature do not act directly in the topic construction (although they do not stop acting), an interpretation that reinforces the intrinsically textual-interactive character of textual construction as postulated by TIG.

In (12), this somewhat enhanced presence of the interactive instance in the textual formulation can still be verified by observing the joint definition, between documenter and informant, of the referent that will assume the status of a discourse topic. As we noted, the documenter’s speech, on lines 1-3, suggests to the informant the topic to be developed in the description, through the activation of the referent “algum lugar em que a informante já foi/somewhere the informant has already been”, specified shortly after by the interviewer himself with the use of “sua casa/your home”. Given this suggestion, the speaker announces, on line 4, that she will, in fact, focus on her house, as we read in “bom... eu vô(u) falá(r) da minha casa/ well... I will talk about my house.” This collaborative definition of the central topic addressed is so textually marked that, after this initial topic segment, in which the documenter directs the topic and the informant accepts it, she develops her text and, to finish it, formulates an utterance that practically repeats the passage in line 4, as we see in “bom eu falei um po(u)co da minha casa/well I talked a little bit about my house” (L. 15), explicitly interacting with the documenter’s proposal about the focus of the description to be the informant’s house.

The topic segment in lines 1-4, in (13), is another case that concretizes a meta-discourse topic, this time in a narrative of experience.

It is acceptable to say that, in (13), the topic materialized in lines 1-4 is a meta-discourse one because, in this topic segment, we have a centered set of utterances that defines and indicates a topic, in an interactive process between documenter and informant, hierarchically comprehensive, to be addressed in the narrative of experience. Therefore, in this topic segment, there is a focus on the activity of textual construction, unfolding the discourse on itself (self-reflexibility), even, in this case, without explicit marks of meta-discourse on the textual surface, which may have an ellipsis and be filled by the recovery of the omitted declarative terms, as Risso (1999) recalls. This can be seen in the documenter’s speech “C. tem algum fa::to que aconteceu [...] que você tenha vivenciado/C. is there any event that happened in your family or with you that you remember... that you have been a part of that story or you have experienced it?” (L. 1-3), which can be interpreted as “C. I would like to ask if there is any event that happened [...] that you have experienced”.

We also understand that this topic is meta-discursive because, as occurs in (12), the topic that runs through the entire excerpt in (13), the topic “Separação dos pais/parents’ divorce”, is not yet directly developed in lines 1-4, and its definition is focused on, so that this development takes place from line 5 onwards, which corroborates the considerations of Risso (1999) and Jubran (2022) that the meta-discourse does not work directly in the topic elaboration.

Once we assume that the topic “Definition of the theme ‘parents’ divorce’” does not act directly in the topic construction, we can, guided by the gradience principle, consider that the interactional function in this topic grows, reducing its role in the informational organization of the text. In this way, a point is constructed in the narrative of experience in which the interactional function is intensified, a fact that can be confirmed by verifying the explicit interaction between documenter and informant, which jointly define the topic to be addressed.

In our view, in order to improve the methodological tools of TIG, it is relevant to recognize, as proposed in this paper, that the construction of texts sometimes involves the establishment of meta-discourse topics, particularly topics centered on the topic elaboration of the text itself. First, this is a finding that has not yet been attested and discussed in the set of works carried out by the approach so far. In addition, the finding seems significant, above all, because it provides TIG with a more complete and accurate analytical enterprise.

In a textual/topical analysis, when the possibility of textual construction involving such meta-discourse topics is not considered, one tends to incur one of two descriptive limitations, which, incidentally, can be observed in works already carried out in TIG, such as Penhavel (2010, 2020): either the analysis does not include the description of excerpts such as those in lines 1-4 in (12) and (13) above, or excerpts such as these are taken as parts of the textually adjacent topic segment. The first option means a certain descriptive incompleteness, since the topic structure of a text covers the entire text, from the first to the last word, forming a network of relations, a unity, in which all the utterances are structurally interconnected with each other, composing the whole of the text – thus, in the analysis of a text, a description that fails to cover the entire text is not completed. The second alternative, in turn, fails to recognize that excerpts such as those in question, as we sought to demonstrate, establish, in themselves, a proper centering, distinct from the centering of adjacent excerpts, which gives them the aspect of autonomous topic segment, and not exactly the aspect of being part of another topic segment. Hence the relevance of considering these meta-discourse topics in textual/topical analysis.

Equally decisive for this type of analysis is the methodology for naming the referents that assume topic status in a text (the target of the second objective of this work). The importance of this issue lies not only in the fact that, once again, it is a matter not yet systematized in TIG, but also in the centrality that this identification of topics presents in text studies based on discourse topic. For example, as can be seen in Penhavel (2010, 2020), in several textual genres the internal structuring of minimum topic segments (and, by extension, entire texts of these genres) is based on the combination of units (groups of utterances) that differ from each other by expressing, in different ways, the topic of the topic segment. In other words, the identification/naming of the topics is fundamental for the understanding of the entire structure of texts in these genres. In the same sense, by the identification of topics in textual analysis, it is possible to observe the relationship between the discourse topics of a whole text, such as those illustrated in Figure 1. The recognition of the different referents that are topics, including the referent that is the central topic of a text, makes it possible to verify the entire relationship of topical interdependence because it makes it possible to investigate, for example, the way the connection between the central topic and each of the topics hierarchically subordinate to it occurs, which once again makes evident the centrality of the identification/naming of topics in the analysis of the entire organization of a text within the scope of topically oriented textual studies.

In this regard, according to our proposal, there are three basic procedures that can be adopted in the recognition of the discourse objects that, in the course of a text, assume topic status: (i) the identification of a referential expression that makes explicit, in the materiality of the topic segment, the referent with topic status; (ii) nominalization; (iii) the inference of a referent that is topical, but that is not explicit in the topic segment.

The example in (14) illustrates a topic segment in which the identification and naming of the topical referent may be carried out using procedure (i).

The topic segment in (14) integrates a description in which the informant describes different places in her city, such as the municipality’s health center, the topic on which the interaction in the topic segment in question is centered. Within this topic segment, we can see the explicit activation and maintenance of the speech object “postinho de saúde/small health center”, for example, through the reference expressions “o Posto de Saúde/the health center” and “um postinho/a small health center”, highlighted in bold on lines 2, 3 and 4. This is how the example in (14) demonstrates the procedure (i) of recognizing the referent that is topical, since the object of discourse around which the textual-interactive process is centered is formally represented in the topic segment, being explained by referential expressions that name it.

In (15), the identification of the referent that is the topic can also be obtained by observing the explicit activation of this object of discourse in the topic segment.

The topic segment in (15) was taken from a description in which the informant describes his two rural properties, focusing, in this topic segment, on the changes he has made to one of these properties since he took over. As can be seen, until line 7 the referent “mudanças/changes”, the core of the phrase that names the topic, is not formally introduced into the text through a referential expression, which only occurs in line 8, being reactivated in line 9, as indicated by the bold highlights in the example. Thus, in this topic segment, we indeed have a referent with topical status that appears explicitly in the text. However, this is a peculiar case in relation to that in (14). In the example in (15), the expression “mudanças/changes”, which formalizes such a referent, operates as an encapsulating anaphora, summarizing a series of utterances that also constitute the topic, which, however, had not been presented as entities, such as the utterance “eu a transformei... num:: num num... num plantio de laranja manga café/I turned it... into a coffee mango orange plantation ” (L. 4).

The topic segment in (16) allows us to discuss the second topic identification procedure, the strategy of nominalization.

In the topic segment in (16), the centering is on the paternal registration of the child by the parents hidden from the informant’s family, who was a teenager mother. We notice that there are two very similar excerpts that directly indicate the discourse topic – the first on lines 1-2 and the second on line 6, both highlighted in bold. These utterances, at first, could be used to name the discourse topic, causing the naming of the topic to directly portray the text produced by the speaker. Another alternative, however, for naming the topic is to convert the predications expressed in these passages into a name, through a nominalization, based on the types of formal and semantic adjustments demonstrated with the examples (11a) and (11b), which would result in a name such as “paternal registration of the child by the parents hidden from the mother’s family” and would transform these predications into referents (referential expressions), based on a relation with the cotext.

These two ways of naming the discourse topic could express the theme of this topic segment. However, being an analyst’s job, the naming of topics may not exactly employ the speaker’s own words and still represent the basic content of the topic segment. Assuming this principle, it must be admitted that naming a topic using a noun phrase, as has been done in TIG (Jubran et al., 2002; Jubran, 2006b, 2015c), is a more technical way of representing the content of what the speaker says. In addition to standardizing the way topics are named, naming by noun phrase has the theoretical and methodological advantage of capturing and highlighting the referent status of the topic, as we have argued (Garcia, 2022).

If the topic in (16) were named as The father registered the child hidden from the mother’s family, this discourse topic would represent a semantic category, the one of states-of-affairs, and, being named as Paternal registration of the child by the parents hidden from the mother’s family, it represents this same semantic class. Therefore, from the semantic point of view, there is a correspondence between the sentences uttered by the speaker, such as those in lines 1-2 and 6, and the name – in the form of a noun phrase – that we have attributed to the discourse topic, proving that the method of naming topics by means of a noun phrase, the result of a nominalization, can indeed represent the basic semantic content of the topic segment, being an effective procedure to give names to discourse topics.

The use of nominalization to identify a referent that is topic can also be illustrated from the topic segment in (17).

In (17), extracted from an opinion report in which the informant talks about politics, the centering is focused on the disappointment that the informant claims to have had with Lula. The identification of this referent with topical status occurs, firstly, from the fact that there is an utterance in line 6 which indicates, more directly, the discourse topic of the topic segment, and, secondly, from the conversion of the content of this utterance into an entity, considering formal and semantic adjustments involved in the transformation of verbal predication into nominal term, as discussed by Camacho (2011). It is therefore because we refer to the informant’s experience encoded in the excerpt in bold, which did not have the status of a referent, by the noun phrase nucleated by “decepção/disappointment”, that we say that the identification of the referent that is topical occurs by nominalization.

The case in (18), in turn, makes it possible to discuss how the objects of discourse that are topics in a text can be recognized by an analyst’s inference.

The example in (18) was extracted from an experience narrative in which the informant tells the story of his candidacy for mayor in the city of Ipiguá, focusing, in the topic segment in question, on the commitments that he and his candidate for vice had during the campaign period. Throughout the reading of the example, we verified that the referent with topic status does not explicitly appear within the topic segment through a referential expression, as in (14) and (15), nor can it be apprehended by the nominalization procedure, as in (16) and (17). However, there are other elements that indicate the object of discourse on which centering is built, such as semantic interdependency, especially between the excerpts in bold in (18). Passages such as “foram seis meses acompanhando amigos/ for six months we were with friends”, “visitando aqui visitando ali/ visiting here visiting there”, “montá(r)... duas horas por dia do teu rote(i)ro/ you had to schedule... two hours a day of your time”, “marcá(r)... os comícios/ to schedule... the rallies” do not contain textually explicit references to the same object of discourse. However, they can be recognized as having a semantic relationship, since all can be interpreted as particular forms of “campaign commitments”, an idea that, even implicit, gives topic unity to the topic segment. Thus, even if there has been no explicit mention of the referent that is topical, the analyst can infer it, observing traces of the topical property of the centering, such as semantic interdependency– and even assuming that the emergence of this referent-topic also occurs in the processing of the text by the interlocutors, through an implicit referencing process, since the interlocutors need to understand a topic common to the entire topic segment, even if this topic is not expressed in the textual materiality.

The topic segment in (19) also illustrates the identification, by inference, of a referent that is a topic.

In (19), taken from an opinion report in which the informant addresses drug consumption, a centering is established on the reactions of those who use drugs, the discourse topic of the topic segment. Similar to what occurs in (18), in the example under analysis, the referent which is also the topic is not formally represented by a nominal expression in the text, nor is there an utterance that indicates it more directly and which could go through a nominalization process. Thus, there are other elements that constitute clues about the referent around which centering is built, such as the semantic interdependence between the statements in bold, which collaborate to circumscribe the object of discourse “reações de quem usa drogas/reactions of those who use drugs”, which can be taken as the discourse topic of the topic segment, by an implicit referential activity of the interlocutors and, in the same sense, by an inference of the analyst.

Conclusions

Inserting ourselves in a context of establishing methodological procedures for textual analysis based on the category of the discourse topic, we brought to the center of the discussion two issues involved in the methodology of textual/topic analysis. The first concerns the elaboration of meta-discourse topics in the construction of texts, and the second one comprises the methodology for identifying referents with topic status, involving three basic procedures.

As we sought to emphasize, the recognition of meta-discourse topics, particularly aimed at defining and indicating a topic addressed in a text, is a sign of the textual-interactive aspect of textual processing defended in TIG, considering that interactional factors are inherent to linguistic expression and that textual elements always correlate textual and interactive functions, according to a gradient of functions.

Regarding the three procedures for identifying referents that are topics, we recall that Jubran (2006b, 2015c), when defining the topic property of centering, clarifies that the integrated set of objects of discourse that constitutes the semantic interdependency can be composed of explicit or inferable referents, targets of the interaction. In line with this view, according to the methodology proposed in this work, a referent that, in a text, functions as a topic can be explicitly represented on the textual surface, can be captured by nominalization or can even be inferred. Also, on this second issue discussed in the work, it is worth mentioning our understanding that systematizing methods of identifying referents that acquire topic centering corroborates a fundamental property of the topic, which we have defended in TIG (Garcia, 2022), its referent status – the three proposed procedures are compatible with the inference of topics that are precisely configured as referents.

Finally, while recognizing the limits of this work, which are natural due to the cut required of all scientific research, it is important to highlight our proposal to contribute to the development of the methodological apparatus of TIG, a Brazilian theory of textual analysis. Thus, we join other works that also act in the definition of methodological procedures within the theory, such as those that demonstrate the need to operate with topic segments in textual-interactive description, as the studies by Jubran (2015c) on the parenthetical insertion process and Risso, Silva and Urbano (2015) on DMs.

Acknowledgments

The translation of this paper from Portuguese into English was financed by the Postgraduate Program in Linguistics Studies (PPGEL), São Paulo State University (UNESP), via resources of the Program of Academic Excellence (PROEX) from the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES) (2023, code n. 021/2021). The authors would like to thank PPGEL/UNESP and PROEX/CAPES for the support.

References

Publication Dates

History