Re)categorizing lexical encapsulation from an experimental approach

(Re)categorizing lexical encapsulation from an experimental approach Óscar Loureda, Mathis Teucher, Celia Hernández Pérez, Adriana Cruz Rubio (Universidad de Heidelberg) Carlos Gelormini-Lezama (Universidad de San Andrés) Abstract Anaphoric encapsulation is a discursive mechanism by which a noun phrase recovers an explicature. This eye tracking study addresses the question of whether categorizing versus recategorizing encapsulation lead to different processing patterns. Results show that (1) encapsulators are cognitively prominent areas, (2) recategorization is never less effortful than categorization, (3) the prominence and instructional asymmetry of the encapsulator with respect to the antecedent is greater in cases of recategorizing encapsulation. Overall, encapsulators initiate a complex cognitive operation due to the nature of their antecedent, which includes both encoded and inferred information. A distinctive processing pattern emerges for recategorizing encapsulators: greater local efforts, due to the introduction of new information, do not result in higher total reading times. Beyond the introductory section, the structure of this study is as follows: Section 2 discusses the properties of categorizing and recategorizing mechanisms. Section 3 reviews experimental research on nominal anaphoric encapsulation in Spanish. Section 4 outlines the key aspects of the experimental design and execution. Finally, sections 5 and 6 present the results of the experiment and offer a theoretical discussion of the findings. Keywords: encapsulation, anaphora, reference, experimental pragmatics, eye tracking. Introduction A key question in pragmatics is the issue of how discourse generates mental representations that are interconnected and modified through interaction. Nominal anaphoric encapsulation (examples 1 a-d) is a referential mechanism whereby a noun phrase (NP) retrieves an explicature, that is, a minimal schema made up of predicative linguistic material which has already been contextually enriched and from which the most appropriate inferential paths can be activated (Francis, 1986; Schmid, 2000; Carston, 2002; Wilson & Sperber, 2012; López Samaniego, 2015; Carston & Hall, 2017). Encapsulation cannot designate referents which, according to Lyons' classification (1977), comprise first-order or physical entities (people, animals, objects, etc.). Instead, it can refer to second-order entities, such as processes, events or situations, and third-order entities, which consist of concepts, propositions or ideas (Schmid, 2000: 63-66). (1a) John and Anne developed a schedule of activities. The development surprised their colleagues. (1b) John and Anne developed a schedule of activities. The preparation surprised their colleagues. (1c) John and Anne developed a schedule of activities. The initiative surprised their colleagues. (1d) John and Anne developed a schedule of activities. The feat surprised their colleagues. Coreferential anaphora differs from nominal anaphoric encapsulation in that it identifies an antecedent of a non-predicative nature (examples 2a-d) and it is considered a mechanism of reiterative cohesion, not an encapsulation mechanism (Francis, 1986; Schmid, 2000; Borreguero Zuloaga, 2018; López Samaniego, 2015: 455). (2a) John and Anne developed a schedule of activities. The schedule surprised their colleagues. (2b) John and Anne developed a schedule of activities. The calendar surprised their colleagues. (2c) John and Anne developed a schedule of activities. The project surprised their colleagues. (2d) John and Anne developed a schedule of activities. The scrap surprised their colleagues. Phoric relations can be either anaphoric or cataphoric, and, as both theoretical and experimental studies have demonstrated, the positioning of the antecedent—whether to the left or right of the encapsulator—has significant implications for processing (Kazanina et al., 2007; Kennison et al., 2009). From this point onward, we will focus exclusively on anaphoric relations. In addition to retrieving an explicature, nominal anaphoric encapsulators also serve to categorize or reify the discourse segment to which they refer (Schmid, 2000: 360-376). NPs with an encapsulating function may either introduce new information (1c-d) or not (1a-b). When they introduce new information, they recategorize the antecedent; when they do not, they categorize it. In both instances, they trigger a representational construction that selects the contextually most appropriate assumptions, thereby activating a specific inferential route. This process prioritizes information that maximizes communicative effects while minimizing cognitive effort (Wilson & Sperber, 2012). Not all anaphoric referring mechanisms have a clearly defined discourse segment that functions as a formal antecedent. Given this lack of direct correspondence, it is more appropriate to conceive of reference as a mechanism that retrieves a representation from contextually accessible units, rather than relying solely on a formally established antecedent. This approach suggests that what appears to be the antecedent of an anaphor is, in fact, an antecedent trigger (Cornish, 1996: 25). This is particularly relevant in the case of nominal anaphoric encapsulation in which the interlocutor must reconstruct a mental representation from a certain textual segment which activates the true antecedent. In the present study, the antecedent will be understood as the mental representation activated by a prominent and highly accessible linguistic element, or antecedent trigger (Ariel, 2001), as illustrated in examples (1a-d). This eye tracking study seeks to determine whether categorizing versus recategorizing encapsulation lead to different processing patterns. Categorization, recategorization and nominal anaphoric encapsulation Encapsulators function along an informativeness continuum (López Samaniego, 2013; Borreguero Zuloaga, 2018). At one end of this continuum are categorizing expressions, whose conceptualization mirrors the representation of the encapsulated fragment without introducing new information. Their minimal form includes direct nominalizations (example 1a), or nominalizations whose lexical base is close to that of the verb of the encapsulated fragment (example 1b). In these instances, only minimal inferential routes are activated. Subtle distinctions between examples (1a) and (1b) highlight different aspects of categorizing relations: in (1a), there is greater semantic (and formal) proximity to the antecedent’s representation compared to (1b), where a higher degree of generalization occurs in the representational category (Mahlberg, 2005). (1a) John and Anne developed a schedule of activities. The development surprised their colleagues. (1b) John and Anne developed a schedule of activities. The preparation surprised their colleagues. At the opposite end of the continuum, we find recategorizing encapsulators, which introduce new information (1c-d). Non-axiological recategorization only introduces new descriptive information (1c). Axiological recategorization introduces new information of an evaluative nature (1d). Recategorizing expressions replace the initial representation triggered by the antecedent and promote a comparatively more complex inferential process: they guide the interlocutor toward retrieving an assumption based on the contrast between the new category and the previously established explicature (Francis, 1986; López Samaniego, 2013). Examples (1c-d) involve recategorization, but they differ in key aspects. In example (1c), the recategorization pertains to propositional content or a lower-level explicature. In contrast, example (1d) involves the reconstruction of propositional attitudes or the illocutionary intentions with which the antecedent was framed (i.e., a higher-level explicature) (Blakemore, 1992: 62; Escandell Vidal & Leonetti, 2000: 366; Ifantidou, 2001; Carston, 2002: 119-121) (1c) John and Anne developed a schedule of activities. The initiative surprised their colleagues. (1d) John and Anne developed a schedule of activities. The feat surprised their colleagues. Encapsulators are primarily defined by their conceptual meaning, which facilitates the construction of mental representations concerning entities, events, or states of affairs (Escandell Vidal & Leonetti, 2000, 2004). However, the conceptual meaning of an encapsulator does not activate all stored encyclopedic assumptions. Instead, it activates only the most contextually accessible ones to reach informatively relevant implicatures (Wilson & Carston, 2007: §4). Through this process, the interlocutor—guided by expectations of relevance—constructs an ad hoc meaning or occasion-specific use of the concept, enriched in the explicature, which then allows for the derivation of specific implicatures (Blakemore, 2002; Carston, 2002: 15; Leonetti, 2021). In addition to conceptual meaning, encapsulators also contain procedural meaning, which provides instructions on how information should be processed within the communicative context to generate relevant effects (Fraser, 2006; Wilson, 2011). These procedural instructions take mental representations as input, apply processing rules, and generate modified representations that produce relevant contextual effects as output (Blakemore, 1992: 90-91). Thus, the relationship between procedural and conceptual meanings is asymmetrical, as conceptual meaning must adapt to procedural instructions (Escandell Vidal & Leonetti, 2004: 1730; 2011). Regarding the use -or lack thereof- of new information, we hypothesize that categorizing operations do not require more processing effort than recategorizing operations. The reason for this lies in the fact that encapsulation has the property of (re)categorizing an explicature by searching for optimal relevance based on the relation between the assumption introduced by the speaker and the co-constructed context shared between the interlocutors (Wilson & Sperber, 2012: 165). Recategorizing utterances involve an operation that necessitates additional contextual material to generate communicative effects. Hypothesis 1. Utterances with a categorizing encapsulator do not require more processing effort than utterances with a recategorizing encapsulator. The relationship between the potential new information in the encapsulator and their procedural meaning must also be considered. Anaphoric encapsulators function as an instruction to retrieve information related to an explicature. Due to this procedural instruction, these NPs are expected to assume prominent positions (Carston, 2002: 15) as they ensure that retrieved information remains integrated within the discourse (Schmid, 2000: 371). They function as conceptual reference points (van Hoek, 1997: 52-60) and play a decisive role in processing and reframing the antecedent, particularly when they introduce new information into the discourse. Although categorizing NPs involve reifications, they do not require re-adaptation to the context to activate the most relevant assumptions, as they simply retrieve previously processed information. In contrast, recategorization is arguably more complex because it introduces new information that requires contextual adaptation (Wilson & Sperber, 2012). Hypothesis 2: A recategorizing encapsulator involves a processing instruction that requires greater cognitive efforts than a categorizing encapsulator. With respect to the procedural meaning of the encapsulator, it is crucial to clarify the extent of control that the encapsulating NP exerts over its antecedent. Instructional control pertains to the NP's ability to balance local processing effort while minimizing the risk of compromising the relevance of the relationship. In this context, it can be proposed that instructional control is dependent on the informativeness of the expression, and consequently, varying (re)categorization levels introduce different degrees of asymmetry in the discourse. It is the encapsulator, along with the representations it creates, that guides the interlocutor toward optimal assumptions necessary for maintaining discourse continuity. The processing asymmetry between the encapsulator and its antecedent must remain within bounds to avoid jeopardizing the relevance of the relationship. Particularly in utterances involving recategorizing expressions, a specific processing pattern is expected to emerge, one that compensates for potential local efforts (due to the introduction of new information) by managing the processing of the antecedent. This ensures that the local effort does not negatively impact the overall processing at the utterance level. Hypothesis 3. A recategorizing encapsulator will show a higher relational asymmetry with respect to its antecedent than a categorizing encapsulator. Nominal anaphoric encapsulation from an experimental point of view Previous research on phoric relations has shown that anaphoric expressions serve as cues for the addressee to retrieve specific pieces of given information from memory. These expressions also signal how accessible the antecedent is (Ariel, 2001: 29). Empirical studies have established a strong relationship between the processing effort required to access information and its degree of accessibility (Ariel, 2001; Wilson & Sperber, 2012): the less accessible the information, the greater the processing effort involved (Cornish, 1996; Sanders & Gernsbacher, 2004; Kennison et al., 2009; Fukumura, 2012). It has also been shown that when a piece of information is activated, the overlap of representational content required to refer to it decreases (Ariel, 2001; Almor, 1996; Kintsch & Rawson, 2005). In lexical anaphora, the degree of specificity of the information provided by the anaphoric mechanism influences the effort required to retrieve the referent: the more specific or informative the mental representation generated by the lexical anaphora is in comparison to that of its antecedent, the greater the processing effort it demands. The study of encapsulation has seen significant descriptive advancements across various languages and continues to fuel theoretical debates on several topics: The definition of nominal encapsulation mechanisms, including their comparison with other anaphoric devices (Schmid, 2000; López Samaniego, 2015; Borreguero Zuloaga, 2018). The selection of different encapsulators in relation to the informative structure of the discourse, the distance of the encapsulated segment, or the semantic complexity of the referent (Pecorari, 2016; Parodi & Burdiles, 2019). Preferences for different encapsulators in relation to various domains of language use, as specialized languages, specific discursive genres or oral/written text (Bertucci, 2006; Lala 2010; Izquierdo Alegría & González Ruiz, 2013; Pecorari, 2016; Parodi & Burdiles, 2019). The contrast between encapsulation and argumentative relations (Montolío, 2013; García Pérez, 2022). Contrastive studies exploring the behavior of encapsulation across different languages or their role in translation (Magris 2000; Goethals, 2010; Abad Serna, 2016). The use of an encapsulator in relation to specific speaker characteristics, such as proficiency level or language acquisition method (Flowerdew, 2010; Duque, 2021). The productivity of an encapsulator from a diachronic perspective (Borreguero Zuloaga & Octavio De Toledo Y Huerta, 2007; Granvic, 2019). Several experimental studies have shown different factors that interact in the processing of coreferential pronouns: the syntactic hierarchy of the antecedent, antecedent competition, referent salience, among others (Almor, 1996; Brown-Schmidt et al., 2005; Almor & Eimas, 2008; Çokal et al., 2016; Gelormini-Lezama, 2018; Çokal & von Heusinger, 2024). In contrast, nominal anaphoric encapsulation has received considerably less attention. Loureda et al. (2024), using eye tracking methodology, compared the processing efforts of non-recategorizing pronominal and lexical encapsulators. Their results showed that the processing effort of utterances containing encapsulators is not greater than that of utterances containing coreferential expressions. Experimental study The aim of this study was to test Hypotheses 1-3 (§ 2) in order to examine and better understand the cognitive processing patterns associated with different levels of (re)categorization in anaphoric encapsulation in Spanish. 4.1. Materials and AOI The independent variable was defined by the amount of new information introduced by the encapsulator rendering the following four conditions: categorizing nominal anaphoric encapsulation through nominalization (CEN - 1a), categorizing nominal anaphoric encapsulation through generalization (CEG - 1b), non-axiological recategorizing nominal anaphoric encapsulation (NARE - 1c) and axiological recategorizing nominal anaphoric encapsulation (ARE - 1d) (see § 2). Juan and Ana developed a schedule of activities. (1a) [CEN] The development surprised their colleagues. (1b) [CEG] The preparation (1c) [NARE] The initiative (1d) [ARE] The feat Figure 1. Example of experimental set according to the encapsulator analyzed The antecedent of the anaphoric relation in each condition is highly accessible with no ambiguity or competition between antecedents. The encapsulator is an NP accompanied by a definite article, which provides the instruction to access a uniquely identifiable antecedent. All experimental sets were designed following a uniform structure (cf. Appendix). The critical stimulus consists of two utterances that share the same syntactic structure. The first sentence contains the textual antecedent of the encapsulator. The second utterance contains the encapsulator, which is placed in a prominent position. Processing data during reading were recorded across three areas of interest (AOI) (Brown-Schmidt et al., 2005; Çokal et al., 2016; Çokal & von Heusinger, 2024): – The entire utterance or critical stimulus: this area consists of two parts, the first part contains the antecedent (e.g., Juan y Ana elaboraron un cronograma de actividades [John and Anne developed a schedule of activities]), while the second part includes the encapsulator (e.g., La elaboración/La preparación/La iniciativa/La proeza sorprendió a sus compañeros [The development/The preparation/The initiative/The feat surprised their colleagues]). – The encapsulator: this corresponds to the NP containing the anaphoric expression (e.g., La elaboración/La preparación/La iniciativa/La proeza [The development/The preparation/The initiative/The feat]). – The discourse segment containing the antecedent: this refers to the portion of the discourse in which the antecedent of the encapsulator appears (e.g., Juan y Ana elaboraron un cronograma de actividades [John and Anne developed a schedule of activities]). By analyzing both the utterance as a whole and its individual components, we can draw conclusions about the effect of the procedural and conceptual meaning conveyed by the referring expression at the global level of the utterance and on the specific local areas that comprise it. 4.3. Design and measures The experiment follows a 1x4 factorial design. Sets of critical stimuli were arranged in counterbalanced lists according to condition, employing a Latin square design. Word frequency and length were controlled within the critical stimuli to account for potential confounding variables. Critical and filler stimuli were presented in a 1:2 ratio, and participants viewed them in a pseudo-random order. Processing time, measured in milliseconds (ms), was calculated based on one aggregated dependent variable, which served as an indicator of the quantitative differences between the conditions within the independent variable (Lowie & Seton, 2012; Gries, 2013). Total reading time (TRT) was used to assess the total number of fixations within a given stimulus, acting as a measure of the global effects of processing (Rayner, 1998; Holmqvist et al., 2011). TRT captures the processing of the utterance, reflecting the construction of an initial assumption hypothesis, as well as the inferential calculations needed to verify and, if necessary, adjust that hypothesis (Loureda et al., 2021). 4.2. Participants The sample is made up of a total of 340 participants (n ≥ 20 per experimental list; 106 men and 234 women; 31% men and 69% women) who were native Spanish speakers between 18 and 40 years old and had a high educational level (undergraduates and graduates). They did not have visual disorders that could interfere with the eye tracking methodology. 4.4. Apparatus and procedure Data were collected using the EyeLink Portable Duo eye tracking system and recorded with a sampling rate of 500 Hz and a precision of less than 0.5°. The experimental test was programmed with the Experiment Builder software. Participants read the stimuli silently and at their own pace. It took 10 to 15 minutes to complete the experimental test. The stimuli were presented on a computer screen and participants used a chin rest to minimize head movements. The participants received instructions about the dynamics of the experiment and signed an informed consent before beginning the test. After two successful calibration trials, participants read a block of practice sequences and then thirteen sequence blocks that made up the experimental trial, in which the critical stimulus (cf. Appendix) was contextualized. A calibration validation was always carried out between blocks. Once the data of the entire sample was taken, cleaning and data export was carried out with the Data Viewer program. 4.5. Statistical Treatment For data preparation, extreme values of an AOI within a critical stimulus were removed if any of the following conditions were met (Reichle et al., 2003): Track loss. Total reading time and first-pass reading time by aggregated AOI = 0; Any fast reader (extremely low values). First-pass and re-reading time per aggregated AOI < 80 ms or first-pass, re-reading or total reading time per aggregated AOI > 3x standard deviation per aggregated AOI. Any slow reader (extremely high values). Total reading time per aggregated AOI > 800 ms or first-pass, re-reading or total reading time per word < 3x standard deviation per aggregated AOI. The data obtained were treated with generalized additive mixed models (Fahrmeir et al., 2013; Wood, 2017). The areas of interest were part of the fixed effects of the model, since the comparison of reading times within the areas in the different conditions was sought. Random effects were included (Diggle et al., 2002), as the reading speed of each participant or possible topic effects of the conditions. Nonlinear effects (Wood, 2017) allowed us to estimate flexible patterns, such as word length of each AOI, since the assumption that reading time grows progressively with average word length is too restrictive. In this way, a fixed average number of characters per word was allocated to ensure comparability between AOI. The models provide estimated and predicted mean values per word and standard deviations of reading times. The values obtained statistically can be related to a scale of specific effects given in percentages. These effects are ordered as follows: differences ± 4% are considered irrelevant or unstable effects; differences between ±4% and ±10% are considered sufficiently stable; and differences greater than ±10% reveal large or very large effects (Loureda et al., 2021). The standard error (SE) of the mean was calculated to assess the precision of the estimated mean and ensure its stability for further analyses. The SE provides an indication of how much the sample mean is expected to fluctuate if repeated samples were drawn from the population. A smaller SE suggests that the calculated mean is more reliable and closer to the true population mean. This measure was essential to confirm that the mean values obtained in this study are stable and adequate for subsequent comparisons and hypothesis testing. Results 5.1. Overall, the experiment did not reveal stable differences in the aggregate reading time (total reading time) among conditions, as the comparisons consistently showed differences of less than 4% (Table 1). No statistically significant differences were found within categorizing encapsulation (1a vs. 1b) or within recategorizing encapsulation (1c vs. 1d). Nominal Anaphoric Encapsulation Categorizing Recategorizing Through Nominalization (CEN) Through Generalization (CEG) Non-Axiological (NARE) Axiological (ARE) TRT: 262.69 ms TRT: 265.77 ms TRT: 267.70 ms TRT: 273.37 ms SE: 8.73 ms SE: 8.98 ms SE: 8.84 ms SE: 8.60 ms CEN TRT: 262.69 ms - -1.06 % -1.77 % -3.81 % SE: 8.73 ms CEG TRT: 265.77 ms 1.07 % - -0.72 % -2.78 % SE: 8.98 ms NARE TRT: 267.70 ms 1.80 % 0.73 % - -2.07 % SE: 8.84 ms ARE TRT: 273.37 ms 3.69 % 2.86 % 2.12 % - SE: 8.60 ms Table 1. Total reading time of the complete utterance 5.2. The data indicate that the processing of encaspulators varies depending on the type of category introduced in the relation. These differences can be explained by three key patterns (Table 2): – Encapsulators that introduce axiological recategorization require greater processing effort compared to expressions that only reconstruct lower-level explicature content (greater than 4%). – Recategorizing encapsulators demand higher processing effort than categorizing encapsulators, with highly stable differences ranging between 9% and 19%. – Categorizing encapsulators do not exhibit significant or stable differences among themselves (less than 4%). Nominal Anaphoric Encapsulation Categorizing Recategorizing Through Nominalization (CEN) Through Generalization (CEG) Non-Axiological (NARE) Axiological (ARE) TRT: 262.11 ms TRT: 272.27 ms TRT: 298.17 ms TRT: 312.12 ms SE: 8.74 ms SE: 8.99 ms SE: 9.01 ms SE: 8.65 ms CEN TRT: 262.11 ms - -3.73 % -12.09 % -16.02 % SE: 8.74 ms CEG TRT: 272.27 ms 3.88 % - -8.69 % -12.77 % SE: 8.99 ms NARE TRT: 298.17 ms 13.76 % 9.51 % - -4.47 % SE: 9.01 ms ARE TRT: 312.12 ms 19.08 % 14.64 % 4.68 % - SE: 8.65 ms Table 2. Total reading time of the encapsulator 5.3. Another result is the difference in processing time between the encapsulator and their antecedents (Table 3). The findings reveal a consistent pattern: when an axiological evaluation is introduced, the processing time for the encapsulator increases by more than 39% compared to the antecedent. When a recategorization affecting lower-level explicature is involved, the increase exceeds 30%. In contrast, categorizing relations show minimal differences in processing between the encapsulator and the antecedent (less than 2.3%). In other words, the asymmetry in the processing of NPs relative to their antecedents is at least 13 times greater when a recategorizing operation is involved. Nominal Anaphoric Encapsulation Categorizing Recategorizing Through Nominalization (CEN) Through Generalization (CEG) Non-axiological (NARE) Axiological (ARE) AOI Antecedent (ms) 265.67 1.36 % 266.01 -2.30 % 265.38 30.82 % 266.19 39.45 % AOI Ref. Expression (ms) 262.11 272.27 383.62 439.61 Table 3. Total reading time of the textual antecedent and the encapsulator Discussion 6.1. The data presented in § 5.1 confirm Hypothesis 1, which posits that utterances linked by a categorizing nominal encapsulator do not require more processing effort than those linked by a recategorizing nominal encapsulator. This aligns with the understanding that recategorization involves more complex reifications that modify the representation of the antecedent, guiding the interlocutor to retrieve an assumption based on the contrast between the new category and the explicature to which it refers (Schmid, 2000). The results are consistent with studies on basic-level categorization (Rosch, 1978; Zhuang & Lingnau, 2022, among others). Interestingly, the data also indicate that recategorizing utterances do not require significantly more cognitive effort than categorizing ones. This suggests that encapsulating relations can retrieve complex representations with relatively low processing effort, demonstrating that they are a highly efficient referential mechanism in terms of relevance (Loureda et al., 2024). In particular, the additional effort required for recategorization is mitigated by the fact that the encapsulated element already constitutes a contextually enriched assumption. This facilitates the limitation of anaphoric retrieval efforts, the control of inferential transformation, and the integration of the encapsulated content as a topic in subsequent discourse segments (Loureda et al., 2024). In summary, the processing of recategorizing nominal anaphoric encapsulation—particularly when the recategorization affects a lower-level explicature—does not result in increased cumulative processing time compared to similar categorizing structures. This finding suggests that the distinctions between categorizing and recategorizing relations should primarily be examined within the specific local areas that govern the functioning of the relational mechanism and the redistribution of cognitive effort. 6.2. The results presented in § 5.2, regarding the encapsulator, support Hypothesis 2: a recategorizing encapsulator demands more cognitive effort than a categorizing encapsulator. This outcome is consistent with the local processing efforts required to handle new information and its contextual evaluation. Recategorizing encapsulators introduce additional dimensions to the mental representation, necessitating more complex integration, which leads to higher local processing efforts. Although expressions with lower informativeness do not require greater processing efforts compared to those introducing substantial new information, there remains a distinction in the nature of the information conveyed. Representations associated with lower-level explicatures (1c) are processed with greater ease than those tied to higher-level explicatures (1d), which pertain to the verbal action performed by the utterance (grammatical mood) or the speaker's attitude towards it. This finding aligns with previous observations on the increased cognitive demands required for constructing higher-level explicatures, as opposed to lower-level ones (see Wilson & Sperber, 2012; Escandell Vidal & Leonetti, 2000: 375). However, despite the greater cognitive load, these processing efforts remain relatively controlled because they are localized and thus, they do not significantly hinder the processing of the entire utterance. Categorizing encapsulators do not exhibit sufficiently stable differences among themselves (less than 4%), suggesting that the information contained in the antecedent, which does not explicitly appear in the encapsulating NP, is not entirely erased from working memory. The encapsulator integrates this information into various conceptual entities at different levels, thereby creating accessible representations that can be retrieved with relatively lower cognitive effort (Schmid, 2000; Escandell Vidal & Leonetti, 2004; Wilson, 2011; Leonetti, 2020, 2021). In this respect, minimally superordinate concepts, as in (1b), are particularly effective in terms of relevance. In addition to conveying new information, encapsulation carries an instruction aimed at constructing discourse, which overlaps with conceptual meaning to contextually solidify an explicature. Due to this functional interplay between these two types of meaning, encapsulators exhibit more flexibility compared to other prototypically procedural connectors, such as discourse particles (Blakemore, 2002; Loureda et al., 2021). When instruction and (re)categorization align and adjust to the context, relevant communicative effects are achieved with highly controlled cognitive effort (cf. 6.1). 6.3. The experimental data presented in §5.3 confirm Hypothesis 3, which posits that an encapsulating NP introducing recategorizing information exhibits greater processing asymmetry with respect to its antecedent than an NP introducing non-recategorizing information. This suggests that encapsulators containing new information exhibit a high degree of conceptual dependence on the context (von Heusinger, 2007: 127), establishing an asymmetrical relationship with their antecedents. Such expressions prompt the inferences needed to integrate the new information. In this inferential process, not only is the retrieval of contextual implicatures activated, but cognitive assumptions are also constructed, facilitating optimized processing (Wilson and Sperber, 2012). The encapsulating NP, along with the representations it generates, leads the interlocutor toward optimal assumptions for continued processing of the communicated content (Thompson & Hunston, 2000). The data also indicate that encapsulators function as elements that determine the interlocutor’s cognitive efforts to access the co-text. These local processing efforts are managed effectively, balancing out at the level of the entire utterance. While the encapsulator may require increased processing effort, this effort remains within controlled parameters. The redistribution of cognitive load mitigates excessive demands and facilitates the integration of new information into the interlocutors' working memory (van Dijk & Kintsch, 1983: 349). This behavior aligns with the instructional nature of such expressions (Leonetti, 2021). Conclusions In the present research, experimental evidence has shown how the level of informativeness in encapsulators influences the processing of nominal anaphoric encapsulation. The eye tracking experiment reveals three key differences between operations involving new information (recategorizing) and those involving given information (categorizing): – Categorizing operations do not require more processing effort than recategorizing operations. – Encapsulators represent cognitively prominent areas. When the relation they activate is recategorizing, they never require less processing effort than when activating categorizing relations. – The prominence and instructional asymmetry of the encapsulating NP in relation to its antecedent is greater when recategorizing relations are created, particularly when these relations carry an axiological nature. The differences in the integration of new information in nominal anaphoric encapsulation are not primarily reflected in processing times but rather in how the information is processed. Instead of significant variations in total processing time, what is observed is a redistribution of cognitive efforts throughout the utterance, depending on the degree of informativeness in the relational expression. The encapsulator activates a complex cognitive operation due to the nature of the antecedent, which includes both encoded and inferred information -an explicature-. Through its conceptual meaning, it also (re)categorizes this representation. When this instruction can be easily combined with the triggered category and both are optimally adjusted to the context, the local processing efforts required by the recategorizing encapsulator do not lead to additional cognitive demands at the global level of the utterance. This serves as a control mechanism in discourse construction, as the encapsulating instruction compensates for local processing efforts by managing the relevance of the relation. Beyond the potential integration of new information in encapsulating reification, it is evident that encapsulation functions as an instruction aimed at discourse construction based on conceptual meaning. Thus, its behavior is more flexible than other operations grounded in procedural meaning, such as those involving discourse particles (Loureda et al., 2021). This flexibility aligns with the properties of conceptual meaning and its adaptability to context (Escandell Vidal & Leonetti, 2000). Acknowledgements Authors contribution All authors contributed to the article and approved the submitted version. Declarations Data: The data used for this study were collected in the Language and Cognition Lab from the Pontificia Universidad Católica de Valparaíso, Chile. The statistical analysis of the experimental study was carried out in collaboration with the StabLab advisory team of the Institute of Statistics, University of Munich. Conflict of interest: The authors declare that they have no conflict of interest. Ethical approval: All procedures performed in the experimental study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee. Informed consent: Informed consents were obtained from all individual participants included in the experimental study. References Abad Serna, S. (2016). Funcionamiento semántico de los encapsuladores nominales en la prensa informática: Un análisis contrastivo. In R. González Ruiz, A. Jimeno Zuazu, & C. Llamas Saíz (Eds.), Lingüística y pragmática: Estudios sobre gramática discursiva (pp. 21–37). Síntesis. Almor, A. (1996). Is deontic reasoning special? Psychological Review, 103(2), 374–380. Almor, A., & Eimas, P. D. (2008). Focus and noun phrase anaphors in spoken language comprehension. Language and Cognitive Processes, 23(2), 201–225. Ariel, M. (2001). Accessibility theory: an overview. In T. Sanders, J. Schilperoord, & W. Spooren (Eds.), Human Cognitive Processing (Vol. 8, pp. 29–87). John Benjamins Publishing Company. https://doi.org/10.1075/hcp.8.04ari Bertucci, R. A. (2006). Anáforas encapsuladoras: una análise em textos de opinião. Revista Letras, 70, 207–221. Blakemore, D. (1992). Understanding utterances. Blackwell. Blakemore, D. (2002). Relevance and linguistic meaning: the semantics and pragmatics of discourse markers. Cambridge University Press. Borreguero Zuloaga, M. (2018). Los encapsuladores anafóricos. Una propuesta de clasificación. Caplletra. Revista Internacional de Filologia, 64, 179–203. https://doi.org/10.7203/caplletra.64.11380 Borreguero Zuloaga, M., & Octavio De Toledo Y Huerta, Á. S. (2007). Presencia y función de los encapsuladores en las crónicas periodísticas del s. XVII. Philologia Hispalensis, 21, 119–153. https://doi.org/10.12795/PH.2007.v21.i01.07 Brown-Schmidt, S., Byron, D. K., & Tanenhaus, M. K. (2005). Beyond salience. Interpretation of personal and demonstrative pronouns. Journal of Memory and Language, 53(2), 292–313. https://doi.org/10.1016/j.jml.2005.03.003 Carston, R. (2002). Thoughts and utterances. The pragmatics of explicit communication. Blackwell Publishing. Carston, R., & Hall, A. (2017). Contextual effects on explicature. International Review of Pragmatics, 9(1), 51–81. https://doi.org/10.1163/18773109-00901002 Çokal, D., & von Heusinger, K. (2024). The role of alternatives in the cognitive processing of German demonstratives. Insights from online and offline processing. Frontiers. https://doi.org/10.3389/flang.2024.1433482 Çokal, D., Sturt, P., & Ferreira, F. (2016). Processing of it and this in written narrative discourse. Discourse Processes, 55(3), 272–289. https://doi.org/10.1080/0163853X.2016.1236231 Cornish, F. (1996). Antecedentless’ anaphors. Deixis, anaphora, or what? Some evidence from English and French. Journal of Linguistics, 32, 19–41. Diggle, P., Heagerty, P., Liang, K.-Y., & Zeger, S. (2002). The analysis of longitudinal data. In Journal of the American Statistical Association (2nd ed., Vol. 90). Oxford University Press. https://doi.org/10.2307/2291352 Duque, E. (2021). Las etiquetas de discurso en la enseñanza del español como lengua extranjera. In M. Sáracho-Arnáiz & H. Otero-Duval (Eds.), Internacionalización y enseñanza del español como LE/L2 (pp. 854–873). ASELE. Escandell Vidal, M. V., & Leonetti, M. (2000). Categorías funcionales y semántica procedimental. Cien años de investigación semántica. De Michel Breal a la actualidad, 363–378. Escandell Vidal, M. V., & Leonetti, M. (2004). Semántica conceptual / semántica procedimental. In M. Villayandre Llamazares (Ed.), Actas del V Congreso de Lingüística General (Vol. 2, pp. 1727–1738). Fahrmeir, L., Kneib, T., Lang, S., & Marx, B. (2013). Regression. Models, methods and applications. In Regression. Models, methods and applications. Springer. https://doi.org/10.1007/978-3-642-34333-9 Flowerdew, J. (2010). Use of signalling nouns across L1 and L2 writer corpora. International Journal of Corpus Linguistics, 15(1), 36–55. Fraser, B. (2006). Towards a theory of discourse markers. In K. Fischer (Ed.), Approaches to discourse particles (pp. 189–204). Elsevier. Fukumura, K. (2012). Producing pronouns and definite noun phrases. Do speakers use the addressee’s discourse model? Cognitive Science, 36, 1289–1311. García Pérez, J. (2022). Relaciones macroestructurales y sus fines persuasivos. El caso de las anáforas con valor proposicional. Boletín de Filología, 57(2), 199–223. Gelormini-Lezama, C. (2018). Exploring the repeated name penalty and the overt pronoun penalty in Spanish. Journal of Psycholinguistic Research, 47(2), 377–389. Goethals, P. (2010). Using translation corpora as a discovery procedure. The case of discourse deictic retrospective labelling. In I. Moskowich, B. Crespo, I. Lareo, & P. Lojo Sandino (Eds.), Language windowing through corpora. Visualización del lenguaje a través de corpus (pp. 335–346). Universidad da Coruña. Granvic, A. (2019). Sobre los orígenes de la construcción encapsuladora en español. In M. Blanco, H. Olbertz, & V. Vázquez Rozas (Eds.), Corpus y construcciones. Perspectivas hispánicas (Verba, Anexo, 79) (pp. 41–79). Universidad de Santiago de Compostela. Gries, S. T. (2013). Statistics for linguistics with R. A practical introduction. De Gruyter Mouton. https://doi.org/10.1515/9783110307474 Holmqvist, K., Nyström, M., Andersson, R., Dewhurst, R., Jarodzka, H., & van de Weijer, J. (2011). Eye-tracking. A comprehensive guide to methods and measures. Oxford University Press. Ifantidou, E. (2001). Evidentials and relevance. John Benjamins Publishing Company. Izquierdo Alegría, D., & González Ruiz, R. (2013). La evaluación a través de un mecanismo cohesivo. Las etiquetas discursivas en un corpus de editoriales. In C. Llamas Sanz, C. Martínez Pasamar, & M. Casado Velarde (Eds.), Léxico y argumentación en el discurso público actual (p. 149-185). Peter Lang. Kazanina, N., Lau, E. F., Lieberman, M., Yoshida, M., & Phillips, C. (2007). The effect of syntactic constraints on the processing of backwards anaphora. Journal of Memory and Language, 56(3), 384–409. https://doi.org/10.1016/j.jml.2006.09.003 Kennison, S. M., Fernandez, E. C., & Bowers, J. M. (2009). Processing differences for anaphoric and cataphoric pronouns. Implications for theories of discourse processing. Discourse Processes, 46(1), 25–45. https://doi.org/10.1080/01638530802359145 Kintsch, W., & Rawson, K. (2005). Comprehension. In M. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 209–226). Lala, L. (2010). L’incapsulatore anaforico cosa nell’orale e nello scritto. In A. Ferrari & A.-M. De Cesare (Eds.), Il parlato nella scrittura italiana odierna: riflessioni in prospettiva testuale (pp. 57–78). Peter Lang. Langacker, R. W. (1991). Foundations of cognitive grammar, Vol. 2: Descriptive Application. Stanford University Press. Leonetti, M. (2020). Referencia nominal y anáfora discursiva. In M. V. Escandell Vidal, J. Amenós Pons, & A. K. Ahern (Eds.), Pragmática (pp. 145–165). Akal. Leonetti, M. (2021). El lugar de la estructura informativa en la organización del sistema gramatical. In M. Leonetti & M. V. Escandell Vidal (Eds.), La estructura informativa (pp. 393–426). Visor. López Samaniego, A. (2013). Las etiquetas discursivas. Del mantenimiento a la construcción del referente. ELUA, 27, 167–197. https://doi.org/10.14198/ELUA2013.27.07 López Samaniego, A. (2015). Etiquetas discursivas, hiperónimos y encapsuladores: Una propuesta de clasificación de las relaciones de cohesión referencial. Rilce. Revista de Filología Hispánica, 31(2), 435–462. https://doi.org/10.15581/008.31.1551 Loureda, Ó., Cruz, A., Recio Fernández, I., & Rudka, M. (2021). Comunicación, partículas discursivas y pragmática experimental. Arco Libros. Loureda, Ó., Gelormini, C., Izquierdo, D., Hernández, C., Teucher, M., Julio, C., & Cruz, A. (2024). La encapsulación y la correferencia en español mediante pronombres demostrativos y repeticiones léxicas. Una aproximación experimental con eye tracking. Revista Signos, 57(114), 238–276. Lowie, W., & Seton, B. (2012). Essential statistics for applied linguistics. Macmillan Education UK. Lyons, J. (1977). Semantics. Cambridge University Press. Magris, M. (2000). I nominali incapsulatori nel diritto. Contributo alla competenza testuale e lessicale del traduttore. In D. Veronesi (Ed.), Linguistica giuridica italiana e tedesca / Rechtslinguistik des Deutschen und Italienischen (pp. 459–468). Unipress. Mahlberg, M. (2005). English general nouns: A corpus theoretical approach. John Benjamins. https://doi.org/10.1075/scl.20 Montolío, E. (2013). Construcciones conectivas que encapsulan. [A pesar de + SN] y la escritura experta. Cuadernos Aispi, 2, 115–132. Parodi, G., & Burdiles, G. (2019). Los pronombres neutros ‘esto’, ‘eso’ y ‘aquello’ como mecanismos Pecorari, F. (2016). L’incapsulazione anaforica nell’italiano contemporaneo: analisi di un corpus giornalistico. Bulletin suisse de linguistique appliquée, 104, 27–50. Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3), 372–422. https://doi.org/10.1037/0033-2909.124.3.372 Reichle, E. D., Rayner, K., & Pollatsek, A. (2003). The E-Z reader model of eye-movement control in reading. Comparisons to other models. Behavioral and Brain Sciences, 26(4), 445–476. https://doi.org/10.1017/S0140525X03000104 Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization (pp. 27–48). Lawrence Erlbaum. Sanders, T., & Gernsbacher, M. A. (2004). Accessibility in text and discourse processing. Discourse Processes, 37(2), 79–89. Schmid, H.-J. (2000). English abstract nouns as conceptual shells. From corpus to cognition. Mouton de Gruyter. Thompson, G., & Hunston, S. (2000). Evaluation. An introduction. In G. Thompson & S. Hunston (Eds.), Evaluation in text. Authorial stance and the construction of discourse (pp. 1–17). Oxford University Press. van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse comprehension. Academic Press. Van Hoek, K. (1997). Anaphora and conceptual structure. University of Chicago Press. von Heusinger, K. (2007). Accesibility and definite noun phrases. In M. Schwarz-Friesel, M. Consten, & M. Knees (Eds.), Anaphors in text. Cognitive, formal and applied approaches to anaphoric reference (pp. 123–144). John Benjamins. Wilson, D. (2011). The conceptual-procedural distinction: past, present and future. Emerald. Wilson, D., & Carston, R. (2007). A unitary approach to lexical pragmatics: relevance, inference and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics (pp. 230–259). Palgrave Macmillan UK. https://doi.org/10.1057/978-1-349-73908-0_12 Wilson, D., & Sperber, D. (2012). Meaning and relevance. Cambridge University Press. Wood, S. (2017). Generalized additive models. An introduction with R. Chapman and Hall/CRC. Zhuang, T., & Lingnau, A. (2022). The characterization of actions at the superordinate, basic and subordinate level. Psychological Research, 86(6), 1871–1891. https://doi.org/10.1007/s00426-021-01624-0 Appendix 1st screen Context Juan y Ana acaban de empezar a trabajar como animadores en un campamento. John and Anne have just started working as chaperones at a camp. 2nd screen Filler item 1 El director les ha pedido que planifiquen pronto el próximo verano porque van a recibir a muchos grupos de jóvenes. The director has asked them to plan early for next summer because they are going to receive many groups of young people. 3rd screen Independent variables (critical stimuli) [Each participant reads all critical items once from each of the twelve sequence blocks] Encapsulation Categorization through nominalization (CEN) Juan y Ana elaboraron un cronograma de actividades. La elaboración sorprendió a sus compañeros. John and Anne developed a schedule of activities. The development surprised their colleagues. Categorization through generalization (CEG) Juan y Ana elaboraron un cronograma de actividades. La preparación sorprendió a sus compañeros. John and Anne developed a schedule of activities. The preparation surprised their colleagues. Non-axiological recategorization (NARE) Juan y Ana elaboraron un cronograma de actividades. La iniciativa sorprendió a sus compañeros. John and Anne developed a schedule of activities. The initiative surprised their colleagues. Axiological recategorization (ARE) Juan y Ana elaboraron un cronograma de actividades. La proeza sorprendió a sus compañeros. John and Anne developed a schedule of activities. The feat surprised their colleagues. Hay previstas más de 1000 inscripciones para el campamento. More than 1000 registrations are expected for the camp. 4th screen Filler item 2 El campamento ofrece animación para todas las edades. The camp offers entertainment for all ages. Figure 2. Example of a sequence block 10