Reframing Reformulation: A Theoretical-Experimental Approach

Evidence from the Spanish Discourse Marker "o sea"

by Shima Salameh Jiménez (Author)
Thesis 214 Pages

Table Of Content

←13 | 15→


Tab. 1.Reformulation markers in Spanish (Martín Zorraquino & Portolés, 1999, pp. 4122–4139)

Tab. 2.Discourse connection established by Sp. quiero decir in exemplification contexts.

Tab. 3.From first to bridge reformulation studies. PR represents paraphrastic reformulation; NPR represents nonparaphrastic reformulation. Shaded cells are approached as subtypes of reformulation. See also Murillo (2007) and Pons (2013).

Tab. 4.Classification of reformulation operations based on Rossari (1990: 353).

Tab. 5.Del Saz’s classification of reformulation and its markers (Del Saz, 2003, p. 242).

Tab. 6.Differences between exclusive and inclusive approaches in Pons (2013, 2017) and Murillo (2016).

Tab. 7.Temporal and movement parameters selected.

Tab. 8.Relationship between dependent, independent variables, and conditions.

Tab. 9.Participant distribution in the experiment. T. 1 and T. correspond to each theme (replication) of the experiment.

Tab. 10.Contexts employed in each function and theme.

Tab. 11.Contexts employed in each function and theme.

Tab. 12.Critical sentences per function in Spanish. Themes 1 and mixed.

Tab. 13.Critical sentences per function in English. Themes 1 and mixed.

Tab. 14.Abbreviations for variables employed in manual count.

Tab. 15.Contexts and critical sentences for paraphrase in the experiment.

Tab. 16.Manual count of total eye movements in paraphrases introduced by a discourse marker.

Tab. 17.Manual count of total eye movements in paraphrases introduced without a discourse marker.

Tab. 18.Total reading time. Paraphrase.

Tab. 19.First-pass reading time. Paraphrase.

Tab. 20.Second-pass reading time. Paraphrase.

Tab. 21.Contexts and critical sentences for paraphrase in the experiment.

Tab. 22.Manual count of total eye movements for reformulation introduced by a discourse marker.

←15 | 16→

Tab. 23.Manual count of total eye movements for reformulation without a discourse marker.

Tab. 24.Total reading time. Reformulation.

Tab. 25.First-pass reading time. Reformulation.

Tab. 26.Second-pass reading time. Reformulation.

Tab. 27.Context and critical sentences for conclusion in the experiment.

Tab. 28.Manual count of total eye movements for conclusion introduced by a discourse marker.

Tab. 29.Manual count of total eye movements for conclusion without a discourse marker.

Tab. 30.Total reading time. Conclusion.

Tab. 31.First-pass reading time. Conclusion.

Tab. 32.Second-pass reading time. Conclusion.

Tab. 33.Conclusion. Overview M1–M2 in sentences with and without a discourse marker.

Tab. 34.Overview of conclusion.

Tab. 35.Contexts and critical sentences for correction in the experiment.

Tab. 36.Manual count for total eye movements in correction introduced by a discourse marker.

Tab. 37.Manual count for total eye movements in correction without a discourse marker.

Tab. 38.Total reading time. Correction.

Tab. 39.First-pass reading time. Correction.

Tab. 40.Second-pass reading time. Correction.

Tab. 41.Overview. Differences between sentences with and without a discourse marker in paraphrase, reformulation, conclusion, and correction.

←16 | 17→


Fig. 1.Form-function trap (Pons, 2017). Circles show reformulation markers’ polyfunctionality; blue arrows represent the influence of such functions on reformulation.

Fig. 2.Descriptive cycle for nonparaphrastic reformulation. Reformulation is described with its markers.

Fig. 3.Visual representation of Pons’s exclusive approach to reformulation.

Fig. 4.Visual representation of Murillo’s inclusive approach to reformulation.

Fig. 5.Reformulation. Triangulation between theoretical studies, corpus-based studies and experimental-studies. The branch of experimental studies for reformulation has not been yet explored.

Fig. 6.Visual scanpath of fixations, saccades and regressions during an eye-tracking experiment. Adapted from Zhan, Zhang, Mei, & Fong (2016). Regressions are represented by discontinuous lines.

Fig. 7.Static system, top; head-mounted system, bottom left; and combined system, bottom right

Fig. 8.Remote eye-tracking systems.

Fig. 9.Back-and-forth movement behind paraphrases.

Fig. 10.Forward movement behind reformulations.

Fig. 11.Bottleneck behind conclusions.

Fig. 12.Invalidation behind correction. The small circle represents the invalidated content.

Fig. 13.Randomization of target contents in Latin Square designs. Adapted from Gao (2005). Participants in experiment 1 read different content than those reading experiment 2, and so on.

Fig. 14.Scale of percentages managed in this study (StabLab, LMU).

Fig. 15.Visual pattern for paraphrase with o sea

Fig. 16.Visual pattern for paraphrase without o sea

Fig. 17.Decision tree. Paraphrases with o sea

Fig. 18.Decision tree. Paraphrases without o sea

Fig. 19.M1, o sea, and M2 in paraphrases

Fig. 20.M1 and M2 in paraphrases without o sea

Fig. 21.Reduction of SPRT in paraphrases with and without o sea. Comparison M1 with/without and M2 with/without a marker.

←17 | 18→

Fig. 22.Visual pattern. Reformulation with o sea

Fig. 23.Reformulation without o sea

Fig. 24.Eye movement values for o sea in paraphrases and reformulations.

Fig. 25.Decision tree. Reformulation with o sea

Fig. 26.Decision tree. Reformulation without o sea

Fig. 27.M1, o sea, and M2 in reformulation.

Fig. 28.M1 and M2 in reformulation.

Fig. 29.M1 Comparison between paraphrase and reformulation. TRT.

Fig. 30.SPRT reduction in reformulation with and without a discourse marker.

Fig. 31.Visual pattern. Conclusion with o sea

Fig. 32.Visual pattern. Conclusion without o sea

Fig. 33.Imbalance between M1, o sea and M2 in PF and RF for conclusions. The discourse marker is ignored.

Fig. 34.Decision tree. Conclusion with o sea

Fig. 35.Decision tree. Conclusion without o sea

Fig. 36.M1, o sea, and M2 in conclusion

Fig. 37.M1 and M2 in conclusion without o sea

Fig. 38.Comparison between o sea in paraphrase, reformulation, and conclusion. Conclusion is the easiest.

Fig. 39.Visual pattern. Correction with o sea

Fig. 40.Visual pattern. Correction without o sea

Fig. 41.Decision tree. Correction with o sea

Fig. 42.Decision tree. Correction without o sea

Fig. 43.M1, o sea, and M2 in correction.

Fig. 44.M1 and M2 in correction without o sea

Fig. 45.Representation of reformulation and neighboring categories. On the one hand, discourse; on the other, grammar.

Fig. 46.Four patterns for o sea. Polysemy can be proved experimentally.

←18 | 19→


Theoretical background

This volume aims to reframe reformulation from an onomasiological-semasiological approach so as to solve the so-called form-function trap, which has become a theoretical problem in this field over the last few years (Pons, 2013, 2017). The form-function trap arises when reformulation is defined by the features behind the forms expressing it, namely reformulation markers. Since reformulation markers tend to be highly polyfunctional, other functions could be assumed to be subtypes of reformulation. This leads to:

(i)A loss of reformulation limits, which are identified with other similar but not equivalent functions expressed by reformulation markers (summary, conclusion, correction, mathematical operation, expansion, etc.): These functions become subordinate to reformulation, and new functional subcategories are created (reformulative conclusion, reformulative correction, etc.).

(ii)A creation of new subtypes of reformulation markers: Given that reformulation markers can also summarize, correct, or expand formulations, and that their behavior in such cases is assumed to be different from reformulation contexts, new categories of reformulation markers can be proposed (e.g., detachment markers, correction markers as part of reformulation devices).

The form-function trap reveals a semasiological intrusion into onomasiology; this is shown in current papers in the field, which differ from the original works distinguishing the function (paraphrastic and nonparaphrastic reformulation) from the set of devices expressing it (discourse markers, etc.) (Gülich & Kotschi, 1983; Roulet, 1987). Some researchers defend this semasiological approach (Murillo, 2016); others argue for distinguishing functions and returning to the original work in reformulation (Pons, 2013, 2017). However, there is no agreement on which approach is the most adequate. At this point, a new method to address reformulation is needed, and so this research adopts experimental pragmatics.

←19 | 20→

The novelty of the method: A necessary justification

Experimental methods such as eye-tracking technology, are an innovation in the research field of reformulation1 (Rayner, 1977; Just, Carpenter, & Wolley, 1982; Rayner & Sereno, 1994). Eye-tracking permits the analysis of ocular reactions to different input (video, text, sentences, words, pictures, etc.) in various communicative contexts, such as reading or interaction. These studies follow the “eye-mind assumption” (Rayner, 1998), which relates ocular movements (number, direction, and especially, duration) to difficulty processing information.

Pragmatics has included experimental methods to test hypotheses and problems that cannot be solved completely from a theoretical perspective (Duchowski, 2007), such as reformulation and the form-function trap. Studying functions by addressing text comprehension based on ocular movements would seem a complex task (see Underwood, 1992 about the acceptability (or not) of the eye-mind assumption); however, this method does not differ from functional linguistics and corpus studies. In the end, they share the same aim: to support theoretical linguistic descriptions with specific data. Corpus-based studies address oral or written patterns in conversations, interviews, or real texts; eye-tracking employs reading data also obtained from texts in order to test how the different functions they show are processed and understood by people. As Godman (1996) and Nelson et al. (2008) suggest:

(…) reading is described as the process of constructing meaning from print. From this perspective reading is a transactional interaction between the individual reader and texts for the purpose of making meaning (Nelson et al., 2008, p. 294).

In other words, since texts also aim at being communicative and meaningful, comparable to conversations, interviews, or phone calls in corpus-based studies, they should also be analyzed for linguistic purposes.

←20 | 21→

Applying experimental methods to reformulation and reformulation markers obtains benefits.2 Eye-tracking experiments provide this field with quantitative and interpretable data complementing theory, which opens reformulation up to new treatments. This volume focuses on the form-function trap in Peninsular Spanish. All the examples, contexts, and sentences created for the experiments are proposed for Peninsular Spanish and focus on the reformulation marker o sea. So as to reach optimal results, all the participants are native speakers from Spain.

Research objectives

Two main onomasiological-semasiological objectives are addressed in this study. First, we want to extract cognitive processing patterns related to paraphrase, reformulation, conclusion, and correction. Depending on the similarities or differences shared by the reading data obtained in the experiments, these patterns will show to what extent neighboring functions should be included in or excluded from the category of reformulation. Second, we measure reformulation markers with the effects derived from their presence or absence to prove their polyfunctionality and which role they play in establishing the functions analyzed (Rossari, 1990, 1994; Gülich & Kotschi, 1995; Noren, 1999). These results will not substitute, but rather support, the theoretical ideas underlying reformulation and reformulation markers.

The following four additional objectives assist in reaching the two general ones:

I.To find the form-function trap trigger in the literature; in other words, to detect the moment when researchers started to define reformulation by including features from other functions. To do so, this volume presents a critical review of some of the most important works in the field of reformulation (Gülich & Kotschi, 1983; Charolles & Coltier, 1986; Roulet, 1987; Adam & Revaz, 1989; Blakemore, 1993; Fuchs, 1994; Flottum, 1994; Gülich & Kotschi, 1995; Norén, 1999; Apotheloz, 1999; Cuenca, 2001; Ciapuscio, 2003; Bach Martorell, 2009, etc.). Relevant informative excerpts have been extracted and interpreted to detect how the trap has been progressively developed.

II.To define the properties of an eye-tracking experiment using basic references in the field (Kennedy et al., 1987; O’Regan, 1980; O’Regan et al. 2000; Clifton, 1992; Ferreira, Apel, & Henderson, 2008; Pollatsek, 1993; Rayner, 1984, 1995; Juhasz & Rayner, 2003; Reingold, 2003; Reichle, Rayner, & Pollatsek, 2003).

←21 | 22→

III.To design experimental materials and develop experiments on paraphrase, reformulation, conclusion, and correction. Such experiments should reflect how these functions work structure, meanings, discourse markers, etc. To achieve this, critical sentences with and without a discourse marker are prepared. This allows distinguishing onomasiological from semasiological results, avoiding possible theoretical problems.

IV.To propose research hypotheses to be solved in this volume. These hypotheses should be based on the theoretical framework proposed in Chapters 1 and 2. Each hypothesis requires precise justifications and explanations.

While the general objectives are treated throughout this volume, the additional objectives are addressed in specific chapters. The chapters are organized as follows:

Chapters 1 and 2 address additional objective (I). Chapter 1 focuses on the form-function trap by reviewing several basic references: on the one hand, discourse markers and their polyfunctionality are described to delimit reformulation markers appropriately. On the other, important studies on reformulation are reviewed and definitions of reformulation and reformulation markers are interpreted in order to detect how and when this function begins to be confused with others. Chapter 2 details the three most influential works published in the last few years that debate the form-function trap (Pons, 2013; Murillo, 2016; Pons, 2017). This chapter defines the two theoretical approaches to reformulation that formed the basis of the reading experiments proposed in this volume. It also establishes a bridge between reformulation and experimental pragmatics.

Chapter 3 covers additional objective (II), explaining the concepts of eye-tracking research required to adequately link the functions studied to the experiments.

Chapter 4 accomplishes additional objectives (III) and (IV). This chapter is a protocol it lists all the decisions adopted to design the experiments proposed in this volume. Because of the innovative nature of this research, all the experimental parameters, research questions, and hypotheses are justified in detail. Chapter 4 also clearly summarizes the statistical treatment (decision trees and linear regression mixed models) applied to the raw experimental data obtained in order to understand the analysis.

Chapter 5 presents the eye-tracking reading results for each function. The chapter is divided into four main blocks, one per function. The results are quantitative-qualitative: their triangulation makes it possible to answer the research questions and hypotheses previously proposed.

Lastly, Chapter 6 presents general and particular conclusions derived from the analysis. Some future research guidelines are also suggested.

1All the studies on reformulation currently published are theoretical (Martín Zorra-quino & Portolés, 1999), corpus-based (Gülich & Kotschi, 1995), or contrastive (Nolke, 1994; Cuenca & Bach, 2007).

2Benefits from combining theoretical and experimental approaches have been proved in previous works with focal particles (Loureda et al., 2013; Cruz Rubio, 2019) and counter-argumentative connectives (Sanders & Spooren, 2015; Canestrelli et al., 2013; Zufferey et al., 2017; Nadal, 2019), but not with reformulation markers (except Loureda and López Serena, 2013).

←22 | 23→

Chapter 1Reformulation and reformulation discourse markers: The form-function trap


Defining reformulation exclusively by its markers can lead to descriptions based on the functions they express. Reformulation markers do not only express reformulation but also conclusion, correction, explanation, etc. When markers are the only means of organization, these other functions can be improperly classified as subtypes of reformulation and the limits between functions become blurred. This is the basis of the “form-function trap” (Pons, 2017), which is a semasiological intrusion into the onomasiological field.

This chapter focuses on the development of the form-function trap to understand what it is and how it works (see Chapter 2). Two questions need to be answered:

(a)Why the polyfunctionality of discourse markers provokes a loss of theoretical boundaries of reformulation?

(b)When did researchers begin to describe reformulation with reformulation markers?

The following sections approach these research questions:

-Section § 1.1. accounts for the polyfunctionality of discourse markers (§ 1.1.1.) as a trigger of the form-function trap, especially in reformulation markers (§ 1.1.2.).

-Section § 1.2. focuses on reformulation markers and their role in studies on reformulation via a critical review of representative publications in the field. These works show the changes in the definition of reformulation: first works offer clear definitions for this function (§ 1.2.1.); some works show small changes toward a more semasiological approach (§ 1.2.2.); and, finally, works published in the early 1990s use definitions determined by reformulation markers and the functions they cover (§ 1.2.3.).

←23 | 24→

1.2.Discourse markers and polyfunctionality: A trigger of the form-function trap

1.2.1.Approaches to polyfunctionality

Discourse markers3 are a well-established object of study in linguistics (Halliday and Hasan, 1974; Zwicky, 1985; Schourup, 1999; Bazzanella, 1986; Schiffrin, 1987; Blakemore, 1987, 2002; Fraser, 1988, 1999, 2006, 2009; Brinton, 1996; Schwenter, 1996, 2000; Aijmer, 2002; Fischer, 2006; Fuentes Rodríguez, 1987, 2009, 2012; Cortés Rodríguez, 1991; Casado, 1991, 1996; Briz Gómez, 1993a, 1993b, 2001; Portolés, 1993, 1998; Hansen, 1998; Haselow, 2011; Pons Bordería, 1994, 1998, 2006, 2008; Martín Zorraquino & Portolés, 1999; Cuenca, 2006; Briz, Pons, & Portolés, 2008; Loureda & Acín, 2010; Aijmer & Vandenbergen, 2011; Tanghe, 2016; Crible, 2018, etc.). Researchers show little consensus on a systematic set of definitions, features, classifications, and labels4 (Fischer, 2006) for discourse markers.

←24 | 25→

There is, however, a feature common to all works: Discourse markers are polyfunctional, i.e., they are forms related to several different interpretations or meanings (Hansen, 2008, p. 34) depending on the semantic-pragmatic context (Schiffrin, 2015, p. 62; Bazzanella et al., 2007, p. 10; Pons, 2006, p. 79).5 Such polyfunctionality has also been studied with different approaches: homonymic (Jucker, 1993), monosemic (Fretheim, 2000), and polysemic (Travis, 2006) (see Hansen, 1998, 2006, 2008 for a full review).6 All three approaches argue for a concrete explanation of the way the functions expressed by discourse markers are acquired and interrelated (Hummel, 2012).

Homonymy argues for a number of readings or meanings that are identifiable and listed as distinct in different defining entries (Jucker, 1993, p. 437). These meaning lists are sometimes associated with their conditions of usage. In such cases, no relationship between the readings is assumed (Fischer, 2006, p. 13).7

Monosemy defends that discourse markers cover a complete procedural meaning (“core meaning,” Fretheim, 1990) based on one basic instruction, e.g., argumentative and formulative, etc. This basic instruction is contextually enriched. As a result, various particular discursive uses are developed (Portolés, 2001). These uses, however, are secondary (Portolés, 1998, p. 85; Pons, 2004, p. 54).8

Polysemy assumes that “words may indeed have different senses which are not merely a matter of pragmatics” but “are related” (Hansen, 1998, p. 241; Lewis, 2006, p. 52). Polysemic approaches better explain the rise of the various functions that discourse markers historically develop, because it considers the possible relationships the functions share (Waltereit, 2006, 71).

←25 | 26→

Regardless the approach adopted, it is assumed that all discourse markers are always polyfunctional in both oral (Montañez Mesas, 2015, p. 26) and written discourses (Schourup, 1999, 234). One of the most polyfunctional groups is that of reformulation markers9 (Portolés, 1998; Martín Zorraquino & Portolés, 1999; Domínguez García, 2007; Briz, Pons, & Portolés, 2008; Pons, 2008; Loureda & Acín, 2010; Borreguero & López Serena, 2010; Borreguero, 2015, etc.). Reformulation markers express further functions in addition to reformulation cause-consequence, conversational modality, correction, formulation, etc. Their polyfunctionality makes them suitable items to study different functions in a general way. Such study, however, involves theoretical problems, such as the form-function trap.

1.2.2.Reformulation markers

Reformulation markers are commonly employed to reformulate across languages (Rossari, 2000, p. 110; Gülich & Kotschi, 1983, p. 315).10 Several seminal papers (Gülich & Kotschi, 1983, 1987, 1995; Charolles & Coltier, 1986; Murat & Carter-Bresson, 1987; Roulet, 1987; Adam & Revaz, 1989; Blakemore, 1993; Vázquez Veiga, 1994; Schwenter, 1999; Del Saz, 2003; Murillo, 2007; Bach, 2009) define reformulation markers11 as marks indicating that the new utterance is the best option to re express the idea(s) previously formulated (Rossari, 1994). They are guides reflecting the way discourses are construed and how formulation obstacles are solved (Gülich & Kotschi, 1995).

These fundamental papers also classify reformulation markers into various subgroups based on their semantic instructions. For example, there are four general subtypes of reformulation markers in Peninsular Spanish: explanation, rectification, detachment, and recapitulation.

←26 | 27→

These four subtypes of reformulation markers are related to the following processes. (i) Explanation markers introduce explanations for unclear ideas, especially in technical discourse, but also informally (Bach, 1996; Murillo, 2007). (ii) Rectification markers present new information to change previous ideas wrongly expressed (Fuentes, 1993; Del Saz, 2003; Garcés, 2008). (iii) Detachment markers highlight differences between inaccurate and new, accurate content (Briz, 2001; Pons, 2013). Finally, (iv) recapitulation markers signal a summary of all the ideas previously said12 (Vázquez, 1994; Murillo, 2016).

Tab. 1. Reformulation markers in Spanish (Martín Zorraquino & Portolés, 1999, pp. 4122–4139).

There are further subtypes of reformulation markers related to these four subgroups in Peninsular Spanish and other languages: denomination, reconsideration, invalidation, correction, particularization, exemplification, separation, conclusion, and definition (§ 1.2.). These subtypes have been proposed in papers published in the last thirty years as subcategories of reformulation because they depend on the speaker’s aim of changing some formulation previously uttered (Martín Zorraquino & Portolés, 1999, p. 4128). Nevertheless, most of these categories are derived from the polyfunctionality of reformulation markers and not from the description of reformulation as a function. Observe the following example:

←27 | 28→

(1)Podríamos organizarnos bien. Quiero decir, mientras tú reservas en el restaurante, yo me acerco a comprar las entradas para el teatro.13

Let’s get organized. I mean, while you book the restaurant, I’ll go buy the theater tickets.

Example (1) presents a context of exemplification in which members 1 and 2 of the utterance are related by the reformulation marker quiero decir (“I mean”). This reformulation marker introduces concrete information completing the general idea previously expressed. This concrete information is added because the speaker feels that the first part of the utterance might not be clear enough and thus it should be exemplified:

Tab. 2. Discourse connection established by Sp. quiero decir in exemplification contexts.

The fact that quiero decir (or other reformulation markers such as esto es and o sea) expresses exemplification in some contexts does not mean that exemplification markers e.g., por ejemplo (“e.g.”), digamos (“let’s say”) are a subtype of reformulation markers and, by extension, that exemplification is a type of reformulation.14

Cases such as (1) are very frequent in reformulation studies. They can be summarized as follows:

-Reformulation markers are addressed because explaining reformulation by focusing on them is easier.

←28 | 29→

-Reformulation markers are polyfunctional they express further functions additional to reformulation.


-New subcategories of reformulation markers are created explicative reformulation markers, exemplificative reformulation markers, reformulation conclusive markers, etc.

And, as a corollary,

-The features of such new subcategories are improperly related to reformulation as a function.

This methodological problem becomes theoretical when the defining features of reformulation are mixed with other functions expressed by reformulation markers e.g., conclusion, correction, recapitulation, summary, rectification, and invalidation. This is the basis of the form-function trap.

The following section addresses how and when the form-function trap developed through reviewing important works on reformulation and its markers.

Fig. 1. Form-function trap (Pons, 2017). Circles show reformulation markers’ polyfunctionality; blue arrows represent the influence of such functions on reformulation.

←29 | 30→

1.3.Reformulation studies and reformulation markers

Reformulation has been widely addressed in linguistics literature.15 Initially, studies on reformulation presented two separated aims: on the one hand, defining reformulation theoretically; on the other, describing reformulation markers. Notwithstanding, in the end, these studies became characterizations of reformulation based on how its markers work (Pons, 2017; Pons, 2013, p. 154) and, therefore, on the form-function trap.

For example, Gülich & Kotschi (1995) focus on discourse production activities, specifically on reformulation processes. Their paper starts with a clear general perspective on what reformulation is: a process “which contributes to resolve expression problems by explicating, correcting or stating more precisely the segment of talk” (pp. 30, 40). However, the function is ultimately defined using reformulation markers:

It is possible to distinguish between paraphrastic and nonparaphrastic reformulations. This suggestion offers an interesting way of subdividing the class of reformulating treatment procedures. The distinction is based on the assumption that markers like de toute façon, en somme, en un mot, tout compte fait, somme toute, après tout, en tout cas, en fait, de fait, au fond have a particular feature in common: they indicate something which would not be expressed by the discourse structure alone, namely the “change in utterance perspective”, which at the same time indicates a certain degree of distance to the perspective contained in the reference expression.

(Gülich & Kotschi, 1995, pp. 40, 44)

←30 | 31→

This definition prevails in reformulation studies, as shown in some of the most important literature, which reveals when the form-function trap developed and how reformulation markers progressively influence it (Gülich & Kotschi, 1983; Roulet, 1987; Murat & Cartier-Bresson, 1987; Fuentes, 1993; Rossari, 1994; Gülich & Kotschi, 1995; Martín Zorraquino & Portolés, 1999; Noren, 1999; Del Saz, 2003; Murillo, 2007; Garcés, 2008). These works have been selected for their representativity in this research field, especially for Peninsular Spanish, French, and English. Furthermore, they have been mentioned or detailed in Murillo and Pons’s extended discussion (2013, 2016, 2017), which constitutes the point of departure of this volume (§ 2.1.). The information is presented in three sub-sections: first studies (§ 1.2.1.), bridge studies (§ 1.2.2.), and subsequent studies (§ 1.2.3.).

1.3.1.First studies on reformulation

Gülich & Kotschi (1983a, 1983b) and Roulet (1987) are the first seminal works published in the field. These studies provide a starting point16 because they:

(a)Established the two basic categories of reformulation: paraphrastic reformulation (Gülich & Kotschi, 1983, 1987) and nonparaphrastic reformulation (Roulet, 1987). These categories would be employed in later studies;

(b)Addressed reformulation using different theoretical frameworks within pragmatics and discourse analysis, especially textual analysis and conversation analysis, and others such as argumentation, relevance, or polyphony;

(c)Focused on different structural features to determine with which parameters paraphrastic and nonparaphrastic reformulations are produced in written and oral discourses;

(d)Described the discourse markers reformulation markers usually employed by the speakers in reformulation contexts.

These works are highly influenced by the theory of formulation (Antos, 1982), by the concept of repair developed in American conversation analysis (Schegloff, Jefferson, & Sacks, 1987), and by the Geneva School’s studies on dialogue and discourse analysis (Roulet, 1981; Moeschler, 1985; Roulet et al., 1985; Murillo, 2007, p. 47).ülich & Kotschi (1983)

←31 | 32→

Gülich & Kotschi (1983) present a pioneering account17 for reformulation based on the concepts of paraphrase (coined as “paraphrastic reformulation”) and predication of identity (Mortureux, 1982). Paraphrastic reformulation is produced as a communicative strategy when an idea has not been clearly uttered (Gülich & Kotschi, 1983, pp. 305–308). A second piece of content is introduced and both fragments are assumed as “identiques” or as “discursive duplications” (Gülich & Kotschi, 1983, p. 341).

The new content (Member 1) is semantically similar to the previous one (Member 2).18 They are not required to be identical; however, they share semantic-contextual features accepted by the speakers and predicating their identity (Gülich & Kotschi, 1983, pp. 307–308). Paraphrase is not determined by reformulation markers, which are defined separately as “marks showing how speakers organize their discourse production.” The separation is clearly stated:

This study19 addresses these functions [paraphrase] defining separately: (a) the general function behind paraphrastic reformulation markers (PRM) and (b) the specific different functions which some PRM cover apart from reformulation.

(Gülich & Kotschi, 1983, p. 324; our translation).

Reformulation markers are key in paraphrase: their presence is optional,20 but they help to establish semantic similarities shared by M1 and M2 even when a complete semantic correspondence between contents does not exist.21

←32 | 33→

Reformulation markers cover three specific functions: expansion, reduction, and variation.22 Such functions are ways to establish similarities between M1 and M2. In expansion, M2 shows semantic-contextual features not explicitly expressed in M1. In reduction, M2 is simpler than M1 and condenses all the previous contents. Variations are not expansions or reductions (Gülich & Kotschi, 1983, pp. 328–330; Gülich & Kotschi, 1987b, pp. 256–258). Examples (2) to (4) show such specific uses:

(2)Mary is a linguist; i.e., a person who analyzes how language is produced and used [expansion]

(3)Mary works hard and passes all her exams; i.e., she is a good student. [reduction]

(4)Mary is an ecologist; i.e., she has adopted a certain way of life. [variation]

This suggests that Gülich & Kotschi (1983a) understand reformulation and its markers as two separate objects of study in their first works. (1987)

Roulet (1987) proposed nonparaphrastic reformulation to distinguish it from paraphrastic reformulation. While paraphrase allows speakers to introduce clarifications or explanations to make M1 more understandable, nonparaphrastic reformulation reflects a change in the discourse orientation to modify the illo-cutionary force expressed in M1. This modification is based on the concept of distance (Roulet 1987: 115).23

←33 | 34→

Nonparaphrastic reformulation is usually expressed by reformulation markers (“connectives” in Roulet’s terms):

(…) this study24 has three main objectives: to propose a global definition for reformulation connectives, to distinguish different subtypes of reformulation connectives depending on how the discourse perspective changes, and to describe how reformulation connectives work (…)

(Roulet, 1987, p. 117; our translation).

Reformulation markers25 work differently than argumentative, counterargumen-tative or consecutive markers (Roulet, 1987, p. 119). They highlight changes of discourse perspective at both the monological and the interactive level26 (Roulet, 1987, p. 135) in three different ways (Roulet, 1987, pp. 120–121):

-By invalidating the enunciative perspective offered in M1;

-By highlighting the new perspective adopted by the speaker in M2;

-By focusing on the type of change executed from M1 to M2

These three processes, however, are not presented as subcategories of reformulation markers or as subtypes of reformulation: form and function are again distinguished.

←34 | 35→

To sum up, these first works characterize paraphrastic and nonparaphras-tic reformulation in a clear way. Reformulation markers are marks to establish reformulation processes, but they are not the basis of reformulation. That is why Gülich & Kotschi (1983a, 1983b) and Roulet (1987) do not define reformulation exclusively by analyzing reformulation markers but use them as an additional explicative device.

1.3.2.Bridge studies

Other studies are in line with the first works (Charolles & Coltier, 1986; Murat & Cartier-Bresson, 1987; Gülich & Kotschi, 1987). These studies, however, begin to open reformulation up to further functions. They function as a kind of bridge: they were published before Roulet’s study (1987), and thus do not address nonparaphrastic contexts, and introduce definitions for reformulation based on some reformulation markers. In addition, such works incorporate new subcategories closer or related to reformulation. These subcategories are not definitive but, in the end, could trigger the form-function trap and blur the functional limits of reformulation. & Coltier (1986)

Charolles & Coltier (1986) take a contextual approach to reformulation27: para-phrastic reformulations are produced when writers aim to make their discourses easily understandable for readers by providing them with coherence28 (Charolles & Coltier, 1986, p. 56). This definition shares features with Gülich & Kotschi’s (1983a) proposal and their view of reformulation as a communicative strategy. However, Charolles and Coltier introduce further categories closer to reformulation that could be interpreted as subcategories of this function: denomination, conclusion, and correction. These subcategories are derived from the polyfunctionality of the reformulation markers they address, c’est-à-dire (“that is to say”) and autrement dit (“in other words”):

(5)Je veux un cadeau pour noël; c’est-à-dire une encyclopédie.

(6)Il étude beaucoup et fait ses devoirs chaque jour; c’est-à-dire, il est un garçon appliqué.

(7)J’aime la douce; or autrement dit, j’aime le fruit sucré.

←35 | 36→

The fact that c’est-à-dire or autrement dit can denominate (5), conclude (6), and correct (7) does not imply that denomination, conclusion, and correction29 are subcategories of reformulation. This work notes that such functions are closer to reformulation (and not subtypes of it), but this proximity could propagate incorrect associations between the main function analyzed (reformulation) and the other functions covered by reformulation markers. & Cartier-Bresson (1987)

Murat & Cartier-Bresson (1987) also study c’est-à-dire and its different functions to account for reformulation, which is described as a (re)interpretation of content (Murat & Cartier-Bresson 1987: 6). They distinguish between paraphrase, correction, argumentation values, and interpretative retakes. Their study notes that such functions should not be confused:

Paraphrase30 and interpretative retakes should not be confused: although they could lead to the same result, equivalence does not play the same role. Interpretative retakes show profound differences from reformulation; reformulation is not a rectification: there is not a subordination between two formulations, they are placed at the same level

(Murat & Cartier-Bresson, 1987, 6–10; our translation).

←36 | 37→

However, after distinguishing these four functions, various subcategories are introduced: interpretation in intention (the meaning of the word) and interpretation in extension (the identity of what is being referred to). Intention is subdivided into réprise definitionelle; extension into coreference, quantification, or specification (Murat & Cartier-Bresson, 1987, p. 11). These subcategories are again similar to paraphrase and could be identified with this function because they are all expressed by the same reformulation marker(s)31; such associations between similar functions contribute to the development of the trap.ülich & Kotschi (1987a)

Gülich and Kotschi (1987a) update their first classification of reformulation (1983) by including further subdivisions. Two new categories are added and labeled as reformulation acts: repetition and correction (Gülich & Kotschi, 1987a, p. 30).32 Expansion and reduction (old categories established in Gülich & Kotschi, 1983) are also subdivided into defining explanation, exemplification, denomination, summary, and variation:

In paraphrases33 in oral texts, equivalence (regardless of its degree of force) can be employed as an expansion, a reduction or a variation (…) Considering other semantic features, five subcategories of paraphrase are distinguished: explication définitoire, exemplification, denomination, summary, variation.

(Gülich & Kotschi, 1987a, p. 40; our translation).

←37 | 38→

It is noted that such subcategories are not determined only by the semantic relationship between the contents involved but also by the discourse markers introducing them c’est-à-dire, c’est ce qu’on appelle, en d’autres termes, cela veut dire, je m’explique, to name but a few (Gülich & Kotschi, 1987a, p. 45).34 Given that they are prototypical reformulation markers, but that they can exemplify or summarize, it is assumed that exemplification and summary are subtypes of reformulation. The same occurs in the definition of explanation, denomination, and correction.

In summary, Gülich & Kotschi’s (1987a) examples are not always based on discourse markers, and this is the reason why their work also sits amidst fundational works and amplified works; however, most of the examples they study are determined by reformulation markers.

These bridge studies show how the polyfunctionality of reformulation markers prompts further interpretations of reformulation and adds new subtypes or subcategories for this function, whose functional limits could be blurred with other, closer functions (see Tab. 3).

After bridge publications, reformulation adds many subcategories and increases in complexity. There are no guidelines separating the subcategories introduced from paraphrastic and nonparaphrastic reformulation. Subsequent studies reveal a clear consolidation of the form-function trap.

1.3.3.Subsequent studies

A series of studies amplifies the first definitions of reformulation proposed by Gülich & Kotschi (1983) or Roulet (1987), Fuentes (1993), Rossari (1994), Gülich and Kotschi (1995), Martín Zorraquino and Portolés (1999), Noren (1999), Del Saz (2003), Murillo (2007), Garcés (2008), etc. These subsequent studies focus on reformulation markers and explain every reformulation process by analyzing it: in comparison with bridge studies, they show a complete intrusion of semasiology into onomasiology. (1993)

Fuentes (1993) follows the paraphrastic/nonparaphrastic distinction. Such distinction presents further subdivisions into new subcategories, e.g., explanation, correction, conclusion, recapitulation, and so on. Several new labels are

←38 | 39→

Tab. 3. From first to bridge reformulation studies. PR represents paraphrastic reformulation; NPR represents nonparaphrastic reformulation. Shaded cells are approached as subtypes of reformulation. See also Murillo (2007) and Pons (2013).

employed: paraphrastic reformulation is subdivided into explication and denomination; correction covers generalization label, conclusion, and recapitulation and expansion enumeration and particularization/exemplification (Fuentes, 1993, p. 177). These categories present a different distribution in comparison with previous studies: correction and conclusion had not been yet explicitly subordinated to reformulation, and exemplification was always related to para-phrastic reformulation, but not to nonparaphrastic reformulations. All the categories proposed include various examples based on discourse markers:

←39 | 40→

(…) we will focus on analyzing devices explaining or reformulating with summary or conclusions. There are some specific markers: en una palabra, brevemente, en pocas palabras, en fin, por fin, finalmente, al fin, a fin de cuentas, al fin y al cabo, al fin y a la postre, en definitiva, bueno, bien, pues bien, and total.

(Fuentes 1993, 177–193; our translation).35

Fuentes argues that something previously said can be reformulated through a summary or a conclusion because the discourse markers analyzed in such a study are not employed only to reformulate but also to conclude, correct, exemplify, denominate, and so on. Limits between reformulation, conclusion, and correction are blurred; studying reformulation becomes a detailed description of discourse markers, and the results obtained are considered general features of reformulation as a function. (1990, 1994)

Rossari (1990, 1994) intends to return to the simplicity of paraphrastic and nonparaphrastic reformulation. These works aim to distinguish both types of reformulation established in Gülich & Kotschi (1983) and Roulet (1987), but reformulation markers are studied again. Paraphrastic reformulation does not require reformulation markers to be produced or highlighted: equivalent content is enough to detect this operation (Gülich & Kotschi, 1983). However, non-paraphrastic reformulation needs the presence of reformulation markers to be established (Rossari, 1990, p. 348).36 Rossari states explicitly that reformulation should be defined by analyzing the discourse markers expressing it:

(…) describing reformulation markers allows defining with precision what reformulation is and how this function works. For this reason, I propose a classification summarizing the different operations related to reformulation depending on the reformulation markers addressed (…)

(Rossari, 1990, p. 349; our translation).37

←40 | 41→

On the one hand, reformulation markers easily establish the change of perspective behind the process; on the other, they highlight different degrees of nonpara-phrastic reformulation (Rossari, 1990, p. 349). These degrees of reformulation38 are recapitulation, invalidation, detachment, and reconsideration (Rossari, 1990, pp. 349–352) and are classified as subtypes of nonparaphrastic reformulation39 as a function and not as operations related to reformulation markers (Rossari, 1990, p. 353; 1994, p. 24):

Tab. 4. Classification of reformulation operations based on Rossari (1990: 353).

According to Rossari,40 given that nonparaphrastic reformulation is usually expressed with reformulation markers, definitions based on such markers should contribute to better describing this function:

←41 | 42→

Reformulation markers highlight changes in speakers’ first formulations; depending on the reformulation marker employed, the function expressed will be different: reformulation can be introduced by connectives, such as when it shows condensation meanings, or can be marked when it shows great distance between formulations

(Rossari, 1990, p. 348; 1994; our translation).41

Therefore, polyfunctionality involves more precision, so this feature should be incorporated into definitions of reformulation.ülich & Kotschi (1995)

Gülich and Kotschi (1995) established new subcategories for reformulation. Following their study published in 1987, this proposal adopts the paraphrastic/nonparaphrastic division. Reformulation is part of the so-called discourse production activities (Gülich & Kotschi, 1995, p. 34); specifically, it is a “treatment procedure” which “allows to modify, to state more precisely, to explicate, or to correct the segment of talk which is thereby specified as a reference expression.” They also claim that the “typical structure of reformulations contains one or more expression functioning specifically as (a) marker(s)” (Gülich & Kotschi, 1995, p. 40).

This definition shows Gülich & Kotschi (1995) opening up a broader approach to reformulation. On the one hand, they focus on reformulation markers and their importance as prototypical reformulation structures; on the other, they approach explanation, correction, or subtle modification as reformulations. Reformulation is subdivided in further complex processes determined by reformulation markers: paraphrastic reformulations can be repetitions (retakes, réphrasages) or paraphrases (equivalent content) (Gülich & Kotschi, 1983, 1987). Nonparaphrastic reformulation are dissociations, highlighting some distance between content with recapitulations, reconsiderations, detachment (see also Rossari, 1990, 1994), and corrections, concerning the form, the formulation procedure itself, or the content (in Holker’s 1981 terms) (Gülich & Kotschi, 1995, pp. 46–51).

←42 | 43→

All the (sub)types proposed for reformulation are explained by reformulation markers; in line with Rossari, reformulation cannot be produced without such markers (Gülich & Kotschi 1995: 48–50).42 These markers express further functions besides reformulations, which is why they are employed in reconsiderations and recapitulations (similar to summary or conclusion) and correction (similar to invalidation). However, it does not follow that these other functions should be considered types of reformulation.ín Zorraquino & Portolés (1999)

Martín Zorraquino and Portolés (1999) define reformulation through its markers in Peninsular Spanish. Previous studies influenced this work, as shown by the four subcategories of reformulation markers proposed: explanation, recapitulation, detachment, and rectification markers (Gülich & Kotschi, 1987; Roulet, 1987; Rossari, 1990, 1994; Gülich & Kotschi, 1995).

Their study describes four such subgroups of reformulation markers by the type of reformulative meaning they express. Explicative markers introduce the new content as an explanation of the previous content. Rectifications are associated with nonparaphrastic reformulation because they also involve a change of discourse perspective (such change, however, is extreme since M1 is canceled and substituted by M2). Detachments reflect initial canonical definitions of reformulation (Roulet, 1987). Finally, recapitulations are related to nonparaphrastic reformulation because the conclusion introduced is different from the previous formulation.

←43 | 44→

Their definition goes from semasiology to onomasiology, in which reformulation markers introduce new content, reformulating the preceding information (Martín Zorraquino & Portolés, 1999, pp. 4121–4122).43 Considering the subtypes of reformulation and functions they describe, this is a broad definition: reformulation comprises everything from explanations to rectifications. This gradual view avoids the paraphrastic/nonparaphrastic distinction (Martín Zorraquino & Portolés, 1999, p. 4121; our translation)44:

(8)Justo Redondo, the baker, o sea (“that is”), Emilio’s son, wakes up early every morning (…) rectification

(9)Marcos lives here, es decir (“I mean”), in Spain (…) explicitness

Again, the inclusion of further functions, such as conclusion, recapitulation, or detachment in a broad definition results from defining reformulation from the reformulation markers expressing it, as shown in examples (8) and (9): given that o sea and es decir are polyfunctional and can explain or rectify in different contexts, these functions (explanation and rectification, among others) are considered part of this gradual continuum of reformulation. (1999)

Noren (1999) does not follow the paraphrastic/nonparaphrastic distinction of the possibility of affecting the characterization of reformulation markers. This approach also accounts for the polyfunctionality of reformulation markers appropriately45:

←44 | 45→

Paraphrastic/nonparaphrastic distinction leads to the idea that reformulation connectives only pertain to one or another category, and that they always show the same meaning despite the nature of X and Y contents.

(Noren, 1999, p. 58; our translation)

Reformulation and its markers are clearly distinguished: these discourse markers prototypically employed to express equivalence or distance between content do not always reformulate, e.g., c’est-à-dire, bref or finalement. This work approaches reformulation from an argumentative-polyphonic framework (Noren, 1999, pp. 49–57, 96–119) and proposes a definition based on the concept of similarity between content46 (ressemblance semantique) (Noren, 1999, pp. 49–53)47:

Contents X and Y in reformulation should be based on the same topos (…) the degree of semantic similarity between X and Y can vary. Cases of very weak semantic similarity should be suppressed since they are not reformulation: reformulation can only be produced if X and Y can be exchanged in the contexts where they are employed without changing the global meaning of the discourse.

(Noren, 1999, pp. 49–53; our translation).

←45 | 46→

Noren’s approach to reformulation is based on the actualization of topoi (i.e., conversational topics) determining the semantics of the discourse. This leads to three subtypes of reformulation: repetition, repetition with strong semantic similarity, and repetition with weak semantic similarity (Noren, 1999, p. 52). Strong semantic similarity approaches what has been defined here as paraphrastic reformulation; weak semantic similarity refers to subtle variations in M2. This definition of reformulation does not contemplate cases such as recapitulation or summary, addressed as subcategories of reformulation in previous studies (Noren, 1999, p. 36).48

In summary, Noren (1999) proposes a clearer model for reformulation concerning the terminology employed: there are no subcategories of reformulation based on strong or weak similarity. However, using this definition creates difficulties in distinguishing weak semantic similarity cases from distance cases, such as those characterized as nonparaphrastic reformulation (see Pons, 2013, for further details). Saz (2003)

Del Saz (2003) proposes another model, blending the characterization of reformulation and its markers in English. Reformulation is first defined as follows (Del Saz, 2003, pp. 211–212):

A reformulation takes place between a source discourse segment S1 (Gülich and Kotschi, 1983) or any of its constituents, and a reformulated segment, or S2, along with the presence of a marker or reformulator, which displays the type of relationship accomplished between the two linked discourse segments and indicates that a recharacterization of the previous discourse segment (S1) has been carried out, so that a new formulation or reformulation is “on the way.”

←46 | 47→

Contrary to Noren (1999), Del Saz argues for the importance of markers in displaying reformulation processes and even establishing them. Reformulation markers are classified into four main groups (Del Saz, 2003, p. 233): explanation: the whole previous discourse segment S1, or one of its constituents, takes the form of an explanation in the reformulated S2 (Del Saz, 2003, p. 234); rectification: S1, or one of its constituents, is a rectification in the newly reformulated S2 (Del Saz, 2003, p. 237); conclusion: S1, or one of its constituents, is a concluding restatement drawn from the S1 (Del Saz, 2003, p. 238); and summary: S1, or one of its constituents, is a shorter version of it (Del Saz, 2003, p. 239). Apparently, her proposal distinguishes between the semasiological level of discourse markers and the onomasiological level of reformulation. However, some parts of the research show how the features of reformulation markers are generalized to reformulation as a function (see Tab. 9, taken from Del Saz, 2003, p. 242):

Clarification: A reformulator from this group carries out a recharacterization of the message conveyed by the whole previous discourse segment S1, or one of its constituents, the rendition of which is a newly reformulated S2 that clarifies some aspect of S1 (…)

Identification: (…) the rendition of which is a newly reformulated S2 that identifies or assigns reference to it.

Illustration: (…) the rendition of which is a newly reformulated S2 that illustrates some aspect of S1.

Rectification: (…) as shown above, or rather and on second thought would mark the cases of Neutral Rectification. Then a first sub-group of Rectification and Improvement (…) is distinguished (…) The specific meanings that these reformulators convey entail a rectification of the message.

Summary: (…) under the general label of Summary, I have opted to differentiate two main sub-groups, DMs of Recapitulation, namely, to recap, and its longer version to recapitulate, along with DMs of Summary, with varied units such

Tab. 5. Del Saz’s classification of reformulation and its markers (Del Saz, 2003, p. 242).

←47 | 48→

as in sum, to summarize, the shorter form to sum up, in short, and its synonymous counterpart in brief, and the most informal in a nutshell, along with in a word, and in a few words (…).

Such subgroups of reformulation markers are the same as the subtypes of reformulation proposed in previous studies on the subject. They express different functions concerning their meaning and are directly extrapolated to reformulation, which becomes a function that not only changes the discourse perspective but summarizes content, identifies, or exemplifies. Del Saz’s proposal thus maintains the trend of broad approaches to reformulation, distorting the first, clear definitions for this function. (2007)

Like Del Saz, Murillo (2007) identifies several discursive instructions instead of distinguishing between paraphrastic and nonparaphrastic reformulation. Up to eleven types of procedure are expressed by es decir, “that is”: (i) identification, (ii) specification, (iii) orientation, (iv) explanation, (v) introduction of restrictions, (vi) correction, (vii) definition, (viii) denomination, (ix) conclusion, (x) mathematical operation, and (xi) consequence. These categories are based on all the previous studies published in the field49 (Murillo 2007: 98) and in turn, are related to different concepts from relevance theory. Identification, specification, and orientation are associated with the recovery of the logical form of utterances (Murillo, 2007, p. 212):

a.(…) identification can be performed by operations that resemble those of conclusion or mathematical operation (Murillo 2007: 213; our translation) in other words, such functions are blended.

Murillo relates explanation, introduction of restrictions, and correction to the recovery of explicatures. Sometimes explanation can be interpreted as an illustration, and again, limits between functions are blurred:

b.(…) an explanation can also be an illustration when an example is provided to explain something. These examples reflect the proximity of reformulation and exemplification markers (Murillo, 2007, p. 218; our translation).

Corrections and repairs are also put on the same level in that they can be identified:

←48 | 49→

c.Finally, correction is far more frequent in oral language (…) These type[s] of examples, which are also repairs (i.e., they are related to unplanned discourse), were found in the Spanish sub-corpus as well (Murillo, 2007, p. 220; our translation).

Definition and denomination are related to the recovery of implicated premises (Murillo, 2007, p. 223). Again, such processes are identified with other functions:

d.Definition examples have a similar structure to that of lexicographical definition. The main patterns used are paraphrases and synonyms (cf. Alvar, 1982; Werner, 1982; Murillo, 2007, p. 223; our translation).

Lastly, conclusion, mathematical operation, and consequence are the basis of the recovery of implicated conclusions (Murillo, 2007, p. 227). Such operations behave like the previous ones:

e.Like the other discourse processes (specification and explanation), a conclusion may be supported by an illustration (…) Recapitulation is a particular case of conclusion (…) The other main type of conclusion is label (cf. Fuentes, 1993). S sometimes a conclusion takes the form of a term that condenses and at the same time qualifies the previous concepts (Murillo, 2007, p. 228; our translation).

f.In the same way, mathematical operations are related to the process of recovery of implicated conclusions (…) A second level of this discourse process includes those in which a result is provided in the second member of a reformulation but is not possible to make a calculation with the information provided (Murillo, 2007, p. 230; our translation).

g.Finally, when the reformulation introduces a fact or event that follows or results from another fact or event; in other words, when it is factual, the corresponding discourse process is consequence (Murillo, 2007, p. 230; our translation).

Identification, explanation, correction and definition are related to reformulation (paraphrastic or nonparaphrastic). Conclusion is considered a reformulative operation related to illustration, recapitulation, or label. All these functions modify formulations, condensing what has been previously said. Murillo’s proposal thus establishes explicit relationships between reformulation and conclusion (in line with Fuentes, 1993). These processes are defined as subtypes of reformulation since the same prototypical reformulation marker (“that is”) expresses all of them. This idea will be the basis of Murillo’s approach to the form-function trap (Murillo, 2016).

←49 | 50→és (2006, 2008)

Garcés intends to separate reformulation from its markers. His studies differentiate various operations of reformulation beyond paraphrastic and nonparaphrastic50:

There are two types of reformulation: paraphrastic and nonparaphrastic. However, this classification should be reviewed given that reformulation is a complex category that includes different meanings such as identification, specification, clarification, correction, conclusion or consequence, recapitulation, reconsideration, or detachment (complete or partial).

(Garcés, 2008, p. 71; our translation)

According to this idea, the paraphrastic/nonparaphrastic distinction should be abandoned because reformulation is a broad relationship involving other meanings related to the intention of modifying the discourse; in other words, reformulation is not only about equivalence or distance, it comprises further degrees of meaning related to additional functions. Therefore, reformulation can be produced through specifications or conclusions, among others.

This proposal considers that the presence of discourse markers is not required for reformulation but that their use shows the type of discourse operation established between M1 and M2 (Garcés, 2008, p. 75). It seems that both objects of study are addressed separately51:

←50 | 51→

Reformulation is defined as a return to a previous formulation (…) in order to reinterpret it. After reformulation, the content behind this formulation is explained, corrected, recapitulated, reconsidered (…). Five groups of reformulation markers can be proposed from this characterization depending on the meanings they cover: explanation, rectification, recapitulation, reconsideration or separation.

(Garcés, 2008, p. 75; our translation)

However, all the operations of reformulation distinguished are exemplified with discourse markers, which suggests that all the operations proposed are based on the meanings codified by reformulation markers and not on reformulation as a function. For instance, Garcés defines explanation as a subtype of reformulation by addressing two prototypical reformulation markers in Spanish, o sea and es decir (“I mean”)52:

(…) The discourse markers developing such operations are es decir and o sea: o sea presents an alternative explanation between different options; esto es identifies or specifies some content within the reformulated content or a clarifies the content previously expressed, and in some contexts, it also highlights the conclusion derived from what has been said or written.

(Garcés, 2008, p. 87; our translation)

This way of defining reformulation shows the semasiological intrusion into on-omasiology: (1) explanation is considered a subtype of reformulation; (2) o sea and es decir express identification, specification, and cause-consequence in some contexts related to explanation— again, many of these functions are related to reformulation; (3) since these prototypical reformulation markers in Spanish express these other functions, they can be associated with reformulation, and therefore, reformulation can be defined by describing the meanings of these markers. This is also true for rectification, recapitulation, consideration, and separation (Garcés, 2008, pp. 104–112, 115–125, 126–141, 142–154).


ISBN (Book)
Publication date
2021 (May)
Berlin, Bern, Bruxelles, New York, Oxford, Warszawa, Wien, 2021. 214 pp., 9 fig. col., 37 fig. b/w, 41 tables.

Biographical notes

Shima Salameh Jiménez (Author)

Shima Salameh teaches and researches in the areas of linguistics and pragmatics. Her research interests include experimental pragmatics, discourse segmentation and discourse markers synchronically and diachronically addressed.


Title: Reframing Reformulation: A Theoretical-Experimental Approach