CC BY
0Volume 8 Issue 2, 2024, pp. 52-65
doi: 10.22363/2521-442X-2024-8-2-52-65
Original Research
Empirical application of sentiment analysis and emotions in Spanish: A post-cognitivist approach
by Juan Antonio Dip and María Inés Silenzi
Juan Antonio Dip
ORCID 0000-0003-3714-2478 h juan.dip@fce.unam.edu.ar Universidad Nacional de Misiones, Argentina
María Inés Silenzi
ORCID 0000-0002-3003-6261 h misilenzi@uns.edu.ar Universidad Nacional del Sur, Argentina
Article history Received March 26, 2024 | Revised May 11, 2024 | Accepted June 5, 2024 Conflicts of interest The authors declared no conflicts of interest Research funding No funding was reported for this research doi 10.22363/2521-442X-2024-8-2-52-65
For citation Dip, J. A., & Silenzi, M. I. (2024). Empirical application of sentiment analysis and emotions in Spanish: A post-cognitivist approach. Training, Language and Culture, 8(2), 52-65.
Text mining has led to growth in sentiment analysis (SA) across various research disciplines. The pandemic has provided a unique and special context for analysing students' written expression. We utilised comments from a survey conducted during the pandemic to create a corpus for SA. The corpus comprises 25,197 words extracted from over 600 comments in Spanish, collected during a survey that lasted around 20 days. We aim to detect sentiments and emotions from this corpus using SA. However, some essential and little-discussed issues in literature should be addressed, such as its relationship with post-cognitivist theory. This paper uses the post-cognitivist approach to analyse emotions and sentiments through SA with the Spanish lexicon in the economics of education. Literature in this area needs further development, especially in Spanish. The article shows that the emotions and sentiments of students in challenging situations can be identified through a corpus of student comments. However, specific elements should be considered while interpreting emotions and sentiments within the framework of post-cognitivism methodologies. Recognising that the human experience is a complex interaction, it is essential to consider the emotional nuances within the context in which they develop. Addressing this issue from the post-cognitivist approach is one of several ways to carry out this task. Using SA and emotions to analyse a text corpus is still helpful for researchers who follow the post-cognitivist approach. However, combining this technique with other qualitative and in-depth computational methods is essential to fully understanding the emotional experiences within their respective contexts.
KEYWORDS: sentiment analysis, post-cognitivist, lexicon, emotions, Spanish, Covid-19
This is an open access article distributed under a Creative Commons Attribution-Noncommercial 4.0 International License (CC BY-NC 4.0), which allows its unrestricted use for non-commercial purposes, subject to attribution. The material can be shared/adapted for non-commercial purposes if you give appropriate credit, provide a link to the license, and indicate if changes were made.
The field of text mining in computing has rapidly advanced, leading to significant growth in sentiment analysis (SA) and its practical application in recent years. SA is crucial in various research disciplines, including psychology, education, sociology, business, political science, medicine and economics. SA is the application of natural language processing (NPL) and computational linguistics to identify a digitised text's sentiments and emotional tone. According to Nandwani and Verma (2021), SA evaluates whether the data emanating from texts, opinions and phrases are related to positive, negative or neutral sentiments,
1. INTRODUCTION
i.e., it has the potential to identify their polarity. On the other hand, emotion recognition identifies and classifies human emotions such as anger, joy or sadness. However, there may also be opinions in texts that involve a negative sentiment associated with an emotion such as anger or annoyance. Although cognition and emotion have historically been viewed as separate, recent studies suggest they are interconnected. Emotional motivations and affective states can impact cognitive processing, while cognition can regulate emotion (Pessoa, 2013). Thus, understanding the connection between emotion and cognition in writing is essential and has widespread implications (Brand, 1987).
© Juan Antonio Dip, Maria Ines Silenzi 2024 Licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
by Juan Antonio Dip and Maria Ines Silenzi
This paper analyses students' comments from a public university in Argentina during the Covid-19 pandemic. Our objectives are twofold: first, to highlight issues that the standard method for SA and emotions in Spanish texts needs to capture, and second, to look for a relationship between this method and the post-cognitivist theory. The latest relationship analysis is a distinctive addition to the existing literature.
Students' written expression has significantly increased throughout the pandemic, creating a unique environment for better analysing their texts, sentiments and emotions. However, some essential and little-discussed issues in literature should be addressed, such as its relationship with post-cognitivist theory. Thus, the paper's critical questions revolve around (i) certain post-cognitivist theses or postulates of cognition and (ii) how these theses are presented in the transdisciplinary field, particularly between cognitive and economic research.
To bridge cognitive research and the economics of education, we will explore the scope and limitations of SA and the emotions conveyed in texts and phrases. Based on the 'fashionable struggle' between the cognitivist and post-cognitivist approaches to cognition (Gardner, 1987), we question the explanatory capacity of SA to account for what situated agents (university students) feel (emotions/sentiments) in a particular context (the pandemic), despite the available empirical evidence and our restriction to fieldwork in a public faculty.
Our key research question is whether SA and emotions can efficiently capture the nuances of the interaction between context, sentiments\emotions, and language. We aim to explore this technique's limitations and possibilities in shedding light on the multifaceted and dynamic nature of human emotions, which various contextual factors shape. We follow the line of Pluwak (2016), who suggests that more than simple SA methods and interdisciplinary research are needed. We also propose the 'per-meabilisation' of specific post-cognitivist theories to enable a better understanding of the complex interplay mentioned above and contribute towards a more accurate interpretation of the results.
The post-cognitivist (embodied) approach emphasises that the mind cannot be considered a 'disembodied' entity, separated from the body, or 'de-situated', removed from the environment. This approach includes situated teaching, anchored instruction, and others. Situated teaching emphasises the importance of activity and context in learning, recognising that education is a process of 'enculturation' in which students become integrated. Anchored instruction is centred around an 'anchor', a context, a real-life problem, or a situation. It is expected to facilitate the generation of new ideas and promote innovative and productive learning. Notably, all these concepts, approached from cognitive psychology, manifest, as the embodied approach does, those cognitive processes resulting from the reciprocal action between the subject (student) and the educational environment (situation) where the student is inserted (Silenzi, 2012). Therefore, we refer to the embodied-situated student as that student who is part of these cognitive processes.
Researchers can focus on cognition's embodied or situated aspect, but the two are intrinsically interconnected. 'Cognition becomes embodied due to its situatedness in specific contexts, such as home or school, where learning occurs' (Reggin et al., 2023, p. 9). Thus, the concept of the situated-embodied student is central to post-cognitivist approaches that emphasise the importance of context, environment, and embodiment in cognitive processes.
Our hypothesis is that, in many contexts, the fact that specific empirical evidence is relevant depends crucially on certain generally implicit conceptual assumptions, such as the context in which the agent is located. Thus, the objective of our analysis is to make these explicit assumptions (cognitivist type) and evaluate them critically (from post-cognitivism). Such questions highlight specific nuances concerning the situated, embodied, anchored student in a given situation, the product of the environment in which they find themselves, reports and 'feels'. Therefore, this study explores how these factors influence students' emotional and sentiment expressions in their written comments during the pandemic.
The paper presents two significant contributions. Firstly, it focuses on the possibilities and limitations of using the post-cog-nitivist approach to analyse sentiments and emotions through SA with Spanish lexicons, specifically in economics in education. The literature in this area needs further development, especially in Spanish. Secondly, we present empirical evidence that students' emotions and sentiments in challenging situations, such as during a pandemic, can be identified through a corpus of student comments. We establish a link between our findings and the 'situated-embodied student' concept and explore elements that should be considered while interpreting emotions and sentiments within the framework of post-cognitivism methodologies.
The article is structured as follows: We first introduce the post-cognitivist theoretical framework and SA. Then, we explain the methodological strategy used to identify sentiments and emotions in students' comments during the pandemic. Next, we discuss the post-cognitivist approach, its similarities, differences, and criticisms. Finally, we present our conclusions.
2. THEORETICAL BACKGROUND
2.1. The post-cognitivist approach to cognition
Before proceeding, it is essential to clarify that the purpose of this article is not to provide a detailed description of the post-cognitivist approach to cognition. Instead, we aim to offer a broad overview of the current landscape, focusing on how SA and emotions can be (re)approached.
To better understand the various approaches in cognitive science, we can begin by classifying them based on our conception of the mind and the historical aspect of the different theories developed within the field. This classification will provide us with a framework to formulate our key question later. For a deeper dive into these approaches, you may find the papers of Burdman (2015), Newen et al. (2018), Shapiro (2019), Varela et al. (2017), Heras Escribano (2021) and Dawson (2013), among others, to be valuable resources.
Volume 8 Issue 2, 2024, pp. 52-65 doi: 10.22363/2521-442X-2024-8-2-52-65
Concerning the first criterion, there are two significant conceptions of human cognitive architecture: the computational conception of the mind and the situated-embodied conception. From the first conception, and in a very general way, the mind is seen as a computational information processing system. From the second, however, the mind is seen as a body and an environment in isolation. Concerning the second criterion, three major research frameworks can be observed in chronological order: classical cognitivism (also called symbolic paradigm, orthodox computationalism or representations and rules approach), con-nectionism (also called subsymbolic paradigm or neurocomputa-tionalism) and a set of alternative theories to the previous proposals. The term 'post-cognitivist approach' will be used here to refer to this last set of alternative theories, corresponding to the situated-embodied conception of the mind. Thus, Gallagher (2023) argues that embodied cognition emphasises the importance of the body's neural and extraneural processes and their interaction with the environment in cognition. This perspective draws from phenomenology, pragmatism, analytic philosophy of mind, developmental and experimental psychology, neuroscience, and robotics.
Several other labels have been invented to bring together these 'new perspectives on cognition'. Dokic (2006) groups them as 'a third wave of cognitive science', Rowlands (2010) under the label 'anti-Cartesian approaches' or the acronym '4E' (Stanciu, 2023), or Marsh and Onof (2008) under the abbreviation DEEDS (dynamical, embodied, extended, distributed, situated). Through different postulates about the nature of cognition, Wilson (2002) brings together some general statements that characterise the core of what she calls a 'new vision of cognition.' Among them, according to our purposes, we highlight that which describes cognition as (i) situated, (ii) extended, (iii) enactive, and (iv) distributive.
In summary, when referring to cognition as situated, a predominant role is assigned to the context (surroundings, environment) since it is considered that mental processes have been designed to function only in combination with the environment outside the subject's brain. All processes carried out outside the elements that structure the natural environment would not be considered within the characterisation of cognition as situated (Haugeland, 1998; Clark, 1998). When referring to cognition as extended, postulates that the cognitive system 'leans' on the environment through cultural artefacts, including language and technological tools, such as the agenda and the computer, to free up limited cognitive resources (Clark & Chalmers, 1998; Hurley, 1998; Clark 1998, 2008). If we consider, for example, our limitations regarding working memory and attention capacity, human beings exploit the environment to reduce cognitive workload. Enactive cognition considers that mental processes are constituted by how an organism acts in the world and, consequently, by how the world acts on that organism (Varela et al., 1991; Hurley, 1998; Noe, 2004; Thompson, 2007; Manca, 2022; Grishechko et al., 2016). Finally, distributed cognition defends the idea that cognitive processes do not depend exclusively on
the action of an isolated individual but also on the social and physical environment in which the agent finds himself (Rupert, 2010; Hollan et al., 2000).
The four assumptions described highlight the role of the environment where the agent is located as part of the cognitive system. Viewing the context and circumstances in which the individual finds themselves is vital. Additionally, it is crucial to recognise individual emotional dimensions and feelings in a particular situation rather than simply focusing on the problem itself.
Thus, we add a new postulate to the previous ones: (v) cognition is affective and constantly interacts with the influence of effects or emotions (Colombetti, 2010; Colombetti & Thompson, 2008; White et al., 2009). The basis of this last postulate is that we can radically distinguish the computational conception of the cognitivist mind from the post-cognitivist one: the former neglects or does not estimate in depth the role of emotions in different cognitive processes. At the same time, post-cognitivism argues that mental processes should not be interpreted solely as symbolic information processes but rather as interaction processes that involve coupling between contextual and affective factors. In this view, the cognitive agent is considered the unit of analysis, encompassing both its affective dimension and its environment. According to Hipólito and van Es (2022), the continuous nature of time and the initial interactions of touch and bodily experience with the world indicate that cognition must be embodied.
Some theories from psychology and philosophy have examined human cognition and behaviour and the computational tools used to understand them. One such theory is post-cognitiv-ism, a branch of cognitive science that questions the theoretical frameworks of computationalism and representationalism when explaining complex adaptive behaviour (Hasselman, 2020). The author proposes a new framework (Radical Embodied Computation) that integrates ecological psychology, embodied embedded cognition, and principles of natural computation to understand how complex adaptive systems use semantic information to coordinate their behaviour. For instance, Casper and Artese (2022) argue that post-cognitivist approaches to cognition could align with computationalism if computations are non-representational and mechanistic. They evaluate this potential combination, highlighting overlooked issues and conclude that while integration between enactivism and computational-ism may be possible, it needs to be improved.
As per the complexity of science, enactive and embodied cognitive science approaches suggest that people are complex adaptive systems, inherently unpredictable and ambiguous. Machine learning systems are becoming more prevalent in all areas of social life, claiming to predict human behaviour. These systems are a form of artificial intelligence and are similar to the Cartesian and Newtonian worldviews, as they sort, categorise, and classify the world and forecast the future. Machine learning prediction of social behaviour damages those at the margins of society and is not only erroneous but also harmful (Birhane, 2021).
by Juan Antonio Dip and María Inés Silenzi
On the contrary, Villalobos and Dewhurst's (2017) work suggests that some theories of computation that do not rely on representation can be compatible with post-cognitivism. While post-cognitive approaches such as activism and autopoietic theory do not necessarily reject computationalism, the rejection of computationalism in post-cognitivism is based on the rejection of cognitive realism rather than the computation itself. Post-cognit-ivism offers a new perspective on cognition by challenging existing paradigms and providing new opportunities for research in various fields, such as psychology, cognitive science, ethics, and education. It does not entirely reject computationalism, but it does reject cognitive realism and opens up new avenues for exploration.
In this way, we question the explanatory capacity of cognit-ivist models, proposing the 'permeabilisation' of some post-cog-nitivist theses in research on the economics of education based on the proposed articulation between contextual and affective factors. Consequently, we suggest recognising the student (i) within a changing, dynamic and uncertain context such as the natural world and (ii) the role of their emotions within that (and not another) context, taking into account the nuances that they can be derived from the (iii) interpretation of a digitised written text.
These clarifications situate the student, anchor, and insert them in a specific context where emotions are not isolated but coupled in the environment where the agent is located. Thus, we postulate the post-cognitivist approach to cognition as an adequate and significant theoretical scaffolding for analysing our students' comments or educational practices.
As a 'general' bridge between these practices and the above-mentioned approach, we question the scope and limitations of SA (by lexicon-based method) in detecting students' sentiments and emotions. During the learning process, the student may find himself isolated from the teaching context or, conversely, directly interrelated (embedded) with that context. Thus, in the literature on the subject, concepts such as situated teaching, situated learning, and anchored instruction incorporate new theoretical contributions from the post-cognitivist approach to education.
The post-cognitivist approach involves the cognitive processes resulting from the reciprocal action between the student's sentiments and the educational environment (situation) in which they are inserted. Still, this section aims to provide a general proposal on how theoretical perspectives can aid in understanding educational processes, especially when the intention is to analyse the student's sentiments and emotions.
2.2. Sentiment analysis, emotions and cognitive theories
SA is a branch of computational linguistics dealing with human emotions and attitudes expressed in textual content. At its core, SA employs natural language processing (NPL) techniques to discern and quantify user opinions in written texts. In simpler terms, it is the art of identifying and understanding emotions and sentiments through the power of artificial intelligence. NLP is a
field of study that analyses human language and aims to transform natural language into a formal language, such as programming, so computers can process it and extract relevant information. Two main aspects of NLP are understanding human language and generating human language. However, comprehension is made more challenging by the ambiguities inherent in natural language. Applications such as speech recognition, document summarisation, question answering, speech synthesis, machine translation and others are based on NLP.
SA and emotion recognition are two crucial areas within natural language processing. Although they are sometimes used interchangeably, they differ in their approach. SA evaluates whether a text or opinion is positive, negative, or neutral, while emotion recognition identifies specific human emotions such as anger, joy, or sadness (Nandwani & Verma, 2021).
SA uses two main methods to uncover the emotional context of text: the machine learning (ML) approach and the lexicon-based approach. In the machine learning approach, pre-existing and labelled data sets are utilised. During the training phase, the system learns complex features inherent to textual data. This phase enables it to accurately identify sentiments in new test data sets by comparing them to its prior knowledge base. Alternatively, the lexicon-based approach focuses on the contextual semantic orientation of individual words within the text. By identifying and quantifying these semantic orientations, the method comprehensively measures the overall sentiment of the text, classifying it as positive, negative, or neutral (Wankhade et al., 2022).
Emotion detection, or emotional recognition, seeks to identify emotions such as joy, sadness, or anger. Many researchers are currently working on automating this process. According to Grishechko (2023), emotional expressions involve cognitive processes related to recognising, understanding, and communicating emotional states. However, detecting emotions from text is a challenge since other body activities transmit emotions and cannot be captured by text (some physical activities, such as heart rate, hand tremors, sweating, and tone of voice, cannot be captured by text) (Kratzwald et al., 2018). Every emotion is an epistemologically objective or intentional experience; it is always a cognition and, simultaneously, a subclass of experiences. In general terms, sentiment is the effect of emotion (Broad, 1954).
There are two main approaches to studying emotions in literature: categorical and dimensional. The first is based on fixed and discrete emotional units that classify emotions into specific categories, such as those proposed by Plutchik (1980), which include anger, acceptance, joy, anticipation, fear, disgust, sadness, surprise, or the Ekman (1992) division, which defines six emotional categories: surprise, happiness, anger, fear, disgust, and sadness. In contrast, the dimensional approach considers emotions as numerical scores in a two or three-dimensional space, taking into account dimensions such as valence (pleasant or unpleasant), arousal (excited or calm), and dominance (high or low) (Bradley & Lang, 1999).
Volume 8 Issue 2, 2024, pp. 52-65 doi: 10.22363/2521-442X-2024-8-2-52-65
A gap in the current literature exists due to a need for more empirical evaluation research on exclusive emotion detection (Hulliyah et al., 2017). Besides, the current literature on SA mentions a need for studies explicitly examining the cognitive dimension of emotions. Scholars have suggested more sophisticated and complementary models to traditional NLP activities to address this gap (Panikar et al., 2022).
Research on emotions in teaching-learning contexts has followed two approaches: In the first approach, emotion is a category of an affective domain that must be adequately developed. In contrast, the second approach focuses on integrating the student's emotions in the teaching-learning context and appropriately managing these emotions during the learning process (Dolianiti et al., 2019).
On the other hand, recent literature reviews have accurately summarised the use of SA tools in education. In this regard, Dolianiti et al. (2019) explore various techniques researchers have used to develop SA systems in educational data sets and examine how SA has been applied in academic decision-making and policy. Virtual learning environments that rely solely on text-based communication can worsen student distress, which is why SA is recommended as a solution to observing stress.
Grimalt-Alvaro and Usart (2023) analysed articles from 2006 to June 2021 to review the application of SA for learning assessment in online and hybrid learning contexts within higher education. The authors emphasised that students' comments can aid in identifying the challenges they face in the classroom. Negative, distressed, and questioning feelings can indicate when a student is experiencing these difficulties.
Thus, most existing papers have focused on determining the polarity of sentiments or classifying emotions but have yet to delve deeper into the analysis of cognition. Specifically, the analytical approach proposed in this article has yet to be applied, especially in Spanish. Our objective is to fill that gap in the literature, focusing, much more particularly, on the foundations of post-cognition to complement SA and emotions (the situated-embodied students). The following section explores how NLP and SA can use the opinions survey as a corpus to identify students' emotions and sentiments during the pandemic in Argentina. Emotions are crucial in shaping decisions and perceptions of the college experience.
3. MATERIAL AND METHODS
Considering the renewed theoretical analysis that cognitive sciences offer and explained above, we propose to address and evaluate a specific technique, such as SA. We will check if this tool allows us to attend to the student's affective dimension within the context where the student is located.
According to Henriquez Miranda and Guzman (2017), the lexicon-based approach is utilised in 42% of Spanish instances, machine learning in 35% of cases, and a hybrid model in 15% of cases. We also employ the lexicon-based approach to sentiment and emotion analysis. We choose to apply the categorical approach since models that use valence, arousal, and dominance
have not been widely employed in computational methods for analysing emotions in text. The categorical approach is the most predominant (Calvo & Kim, 2012). In addition, we implement the dictionary-based approach, which is a feeling lexicon approach in itself. This approach is valuable in addressing the challenges posed by acronyms or slang words, which often affect the accuracy of SA models.
The methodological procedure consists of four steps: (i) to collect information through an online survey as an instrument; (ii) preparation of texts to apply SA; correct spelling errors and eliminate connectors and redundant words, among others (this is our corpus); (iii) apply SA, following the guidelines of Jockers and Thalken (2020); (iv) evaluate the results, discuss the advantages and limits of SA, and analyse emotions.
We discuss a specific case to narrow down the focus of our task. We utilise a dataset from a survey conducted in December 2020 among 1,125 students at a public faculty in northern Argentina - the survey aimed to gather information about the students' virtual learning experiences during the pandemic. This public institution has approximately 5,000 students enrolled in three undergraduate and two graduate courses. The survey represents 22.5% of the students across all offered courses and programmes. It encompasses virtual learning experiences, teaching methods, home equipment, and teaching staff support. The data was collected through Google Forms, and the database was carefully reviewed to avoid duplicating comments or students.
As expressed in the previous paragraph, the survey sought to collect information about the academic experience in virtual-ity. However, one item in the survey asked the following: Which of the following statements do you consider proper? I have gone through stress, anxiety, or similar processes. Surprisingly, 68.3% of those 1,125 students stated that they had gone through stress, anxiety, or similar processes.
This question triggered the investigation of the students' writings on the item. Is there anything else you would like to add? The comments in this survey section are the corpus for analysing sentiments and emotions. The corpus is made up of 25,197 words that were extracted from over 600 comments collected during a survey that lasted approximately 20 days. The survey covered various educational topics, incorporating diverse language use and expressions of sentiment. The survey was conducted in December 2020 in Spanish. Respondents from diverse careers participated during the pandemic. To ensure linguistic consistency, the corpus underwent grammatical corrections and pre-processing to address potential informal or colloquial Spanish styles commonly found in online surveys. Thus, based on this corpus and the initial question that prompted this paper, we can identify components derived from the post-cog-nitivist approach explained in the literature section.
Figure 1 illustrates the interaction of components derived from the mentioned approach: the situated, embodied, and anchored student; the changing, dynamic, and uncertain context; and the analysis of digitised written text to determine sentiments and emotions.
by Juan Antonio Dip and Maria Ines Silenzi
Figure 1. Interaction of components derived from post-cognitivism and SA approaches
As Jockers and Thalken (2020) suggested in their study on text analysis with R, the open-source programming language R is used to analyse the comments. R is a programming environment and language designed, among other things, for statistical analysis. It has become a potent tool for data processing and manipulation.
Using the NRC Emotion Lexicon, we recognise sentiments and emotions (Mohammad & Turney, 2010, 2013). This lexicon includes words associated with Plutchik's (1980) emotions and two sentiments (positive and negative). The words have been accurately translated into Spanish, the language of interest for
our study. Despite cultural differences, most affective norms remain stable across languages. The current translations in 2022 have significantly improved (Mohammad & Turney 2024).
4. RESULTS
Table 1 details the respondents' profiles. The university campus is located in Posadas, Misiones, Argentina. The survey reveals that most participants are women, single and without children, who are pursuing undergraduate degrees (83.5%) and receive family support for their studies. In addition, the vast majority took the virtual modality in Posadas during the pandemic.
Volume 8 Issue 2, 2024, pp. 52-65 doi: 10.22363/2521-442X-2024-8-2-52-65
Table 1
Profile of respondents (n= 1125)
CHARACTERISTICS PERCENTAGES (%)
Women 66.6
Men 33.4
Singles without children living with their parents 47.2
Singles without children renting or student residence 42.2
Student of the Public Accounting career 66.2
Student with a BA in Business Administration 12.7
Student with a BA in Economics 4.6
University Secretariat 6.6
University Administrative Accounting Technician (TUAC) 9.9
Receive financial aid only from family (parents, grandparents) 50.2
Receive financial assistance from family (parents, grandparents) and work 25.5
Just work 24.3
Study in the common cycle - grade (1 st and 2nd year) 42.5
Study professional cycle - degree (3rd, 4th and 5th year) 43.7
Levelling and undergraduate cycle 13.8
He/She is from Posadas and studied virtually in Posadas 48.8
He/She is from the 'interior' or another province, but studied virtually in Posadas 15.6
He/She is from the 'interior' or another province, and studied at home 19.5
To reflect on the SA analysis and investigate emotions, we show an example of how the study of comments and their classification based on dictionaries is carried out. First, the sentiment polarity is analysed. The intensity of the polarity and its values range between [+ 1, - 1], where + 1 represents very positive and - 1 illustrates very negative. Firstly, we will analyse the five comments that resulted in the highest and lowest scores. The comment and the researched words or 'lemmas' are presented in Table 2.
Some comments receive a positive rating when they should be classified as negative. This bias could be because this tool makes capturing the denial of a positive feeling challenging, for example, the phrase they do not give students the opportunity...'. However, some comments, such as 4-5 in Table 2, are considered positive. In the case of polarisation with lower values, we find an entirely negative opinion, and those that follow it are very close to zero. Apparently, most have been classified as positive. It is worth noting that this bias towards positivity could
stem from the dictionaries used in the classification process, the challenge of negating a positive sentiment, or the presence of neutral words in the Spanish language (Figure 2).
What emotions are reflected in these comments? Figure 3 shows this classification of emotions in the students' comments. Many emotions are related to trust, sadness, anticipation, and fear.
Considering the context and situation in which students are placed, anticipation and sadness have been relevant during the pandemic. For example, anticipation is closely linked to the preparation and expectation of future events. Feelings of positive excitement are often associated with the anticipation of good experiences, while anxiety arises from unknown situations. Using the Plutchik wheel (Figure 4), we can deduce secondary emotions derived from the combination of those occurring in Figure 3. Thus, trust + fear (high percentages) show submission, trust + joy is related to love, and sadness + surprise involves disapproval or anticipation + anger results in aggressiveness.
by Juan Antonio Dip and Maria Inés Silenzi
Table 2
Polarity value and lemmas
POSITION VALUE-POLARITY WORDS-LEMMAS COMMENT
1 0.719 change gives opportunity student search for more help Change XXX's chair; they do not give students the opportunity; they seek to harm us more than to help us
2 0.702 like video conference attend I would like video conferences not to be held, since not all of us can attend
3 0.700 improve obtain useful data thus evaluate better learning process The questionnaire could be improved to obtain more useful data and thus be able to better evaluate the learning process
4 0.696 excellent public faculty Excellent public faculty
5 0.691 virtual class highlight tool ... I congratulate the XXX and XXX chair for the virtual classes. In addition, it is worth highlighting all the tools that
1 -0.138 difficult part teacher For the oral exams/midterms it was somewhat difficult with the mistrust on the part of the teachers
2 0.0500 exam method teacher Teachers' exam review methods are not adequate
3 0.0889 class difficult As for the teaching of XXX classes, excellent, not so with the midterms
4 0.0930 virtual exam biggest problem The virtual classroom for exams was the biggest problem
5 0.117 long short situation student lose survey have more weight much little subject The 1st semester was chaotic and extensive, the 2nd less chaotic but extremely short; therefore, in both situations
SENTIMENTS
Positive
Negative
0.0 0.1 0.2 0.3 0.4 0.5
%
Figure 2. Classification of sentiments
Volume 8 Issue 2, 2024, pp. 52-65
doi: 10.22363/2521-442X-2024-8-2-52-65
Figure 3. Emotions of comments (Plutchik, 1980)
Figure 4. The Plutchik wheel
by Juan Antonio Dip and María Inés Silenzi
Now, how does the recognition of emotions arise from written words? We use the comments in Table 2 to exemplify. Table 3 presents the identification of emotions according to the words in the NRC dictionary concerning the comments provided in Table 2. It is highlighted that, of the 151 words in the comments, only some are linked to emotions catalogued in the dictionary. In
Table 3
Distribution of emotions according to polarity value and lemmas
particular, the word problem is associated with emotions related to sadness and fear. Likewise, the word distrust is associated with three emotions: fear, anger, and disgust. On the other hand, the word 'opportunity' is linked to the emotion of anticipation and surprise. This analysis highlights the specific emotional connections present in such comments.
TERM SADNESS FEAR HAPPINESS ANGER DISGUST TRUST ANTICIPA- SURPRISE
ANALYSED TION
problem l l О О О О О О
excellent ОО2ОО2ОО
improve О О l О О l l О
distrust О l О l l О О О
faculty О О О О О l О О
opportunity О О О О О О l l
Based on Figure 1, it is vital to analyse the emotions and sentiments within the context of the students. This helps us understand that negative emotions such as sadness, anger, and submission result from issues related to the pandemic (lockdown) and the shift to virtual learning. These issues include poor internet connectivity in students' homes, inadequate servers for virtual classrooms at the public faculty, and the overall stress experienced during the pandemic. As we pointed out, it is crucial to recognise the student within a dynamic and uncertain context to understand the meaning of each sentiment and emotion. This includes considering the role of their sentiments within that context, especially given the nuances that can be derived from interpreting the written comments (post-cognitivism approach).
5. DISCUSSION
Although sentiments and emotions are detected, it is possible to formulate some criticisms of the lexicon approach, specifically the one used in this article. For example, denial expressed through words like no or never can completely reverse the meaning of a sentence, significantly altering the emotional context. This issue leads us to think that omitting this feature may limit the clarity of the analysis since negation plays a crucial role in expressing emotion and attitudes in the Spanish language. Currently, work is being done to resolve these issues with machine learning techniques (Jiménez Safra et al., 2021; Rendón-Cardona et al., 2022) for Spanish and English (Shaik et al., 2023; Alizadeh & Seilsepour, 2024).
In medicine, Giménez-Moreno and Ricart-Vayá (2022) suggest that specific terms with opposite emotions may appear, and their polarity depends on some combinations. For example, the
positive adjective efectivo is identified as negative when accompanied by the words solo or algo. This phenomenon could also be present in our analysis.
On the other hand, the familiar 'bag-of-words' assumption may need to be more concise since the affective meaning is not simply expressed by the lexicon used but is also an effect of the linguistic structure (Calvo & Kim, 2012). This problem becomes more evident in languages such as Spanish since this technique makes capturing the denial of a positive feeling difficult, as mentioned above. Furthermore, these lexicon-based techniques have often been criticised for needing more validation and accuracy. However, Mohammad and Turney's (2010) dictionary is used in this paper since we explain why it is a good dictionary.
The previous criticisms briefly, although not exhaustively, account for SA's limitations concerning the scope of computational resources needed to capture the complex interaction context/affect/language. We maintain that such criticisms can be expanded and corresponded from an epistemological perspective, accounting for the same issue: the limitations of capturing the underlying complexity.
Computational criticism highlights the explanatory limits of cognitivism by being unable to consider the different nuances. It reflects that in NLP (SA) applications, emotions are not simply unrelated labels but are rooted in emotional experience. Therefore, a range of greys that escape the form can be captured in which language is processed and understood. To argue our position, let us look at other criticisms that comprise them.
1. Simplification of cognition. The post-cognitivist approach criticises cognitive reductionism because it analyses the complex and varied aspects of the mind and human behaviour in terms of
Training, Language and Culture Volume 8 Issue 2, 2024, pp. 52-65
'The relationship between NLP (SA) and the post-cognitivist approach is that the post-cognitivist approach enhances NLP applications by recognising emotions in language and considering how those emotions are expressed within the context where they occur. This means that it unevenly influences the processing and interpretation of the content. Explicitly speaking, keeping post-cognitivism in mind helps enormously when reading the results, with the need to look for other ways to complement the sentiment and emotion analysis method based on lexicons' more complicated processes. Similar to computational critics, post-cognitivists place emphasis on the limited scope of contextual issues.
2. Excessive neutrality. The lexicon approach to SA based on NLP may be biased toward neutrality. It models and understands language neutrally, possibly without incorporating subjectivity and the complexity of emotions highlighted by the post-cognitivist approach. From a computational perspective, it is argued that the sentiments found in the texts (as we do in this paper) could be misinterpreted if the various nuances they are studied are not considered.
3. Simulation of superficial empathy. When considering sentiment and emotion analysis through NLP from a computational standpoint, it is essential to recognise that it can somewhat simulate human empathy. However, it is crucial to understand that NLP is based on predefined programming languages and lexicons and may only encompass some of the nuances highlighted by post-cognitivism. This includes non-verbal cues, body language, reactions, and specific contextual factors that influence the derivation of emotions.
4. Labelling of options. Surveys are tools used to capture and analyse emotions and opinions. However, they often cannot account for the nuances emphasised by post-cognitive approaches. Furthermore, the post-cognitivist postulate on emotions goes beyond categorising sentiments and perhaps some emotions in a binary way (positive/negative - bad/good). Instead, it promotes understanding emotions more thoroughly by integrating perception, cognition, attitudes, and gestures derived from the body. Remember that in this approach, the different emotional nuances that occur within a particular context are considered fundamental, and surveys can perceive this occasionally and under specific conditions.
5. Classification. The computational tool for analysing emotions in texts has two significant limitations. Firstly, the annotated corpora, considered the gold standard, are primarily small and homogeneous. Secondly, emotion identification is often simplified as a sentence-level classification problem, as shown in the results. When classifying sentiments and recognising emotions, the analysis of sentiments and emotions would not consider the complex interaction of context, affect, and language to truly understand sentiments and their causes (Cortal et al., 2022).
rudn.tlcjournal.org doi: 10.22363/2521-442X-2024-8-2-52-65
Despite these critiques and its limitations, SA can still be a valuable data analysis technique for researchers following the post-cognitivist approach. However, it should be used with other qualitative and more in-depth computational methods to fully explore the complexities of emotional experiences within their respective contexts. This suggestion, capturing the complexity of emotions, goes hand in hand with Clark's (1998) methodological proposal. The author describes 'post-cognitivism' as a new movement within the cognitive sciences. He believes this new approach can transform our understanding of old questions about biological cognition. It aims to comprehend the complex and temporally rich interaction between the body, brain, and the world. Post-cognitivism also seeks to introduce new methods, concepts, and tools that do not simply add to but replace the old explanatory tools of computational and symbolic analyses.
Additionally, Clark (1998) suggests that it challenges us to rethink and possibly abandon old familiar distinctions between perception, cognition, and action in favour of more novel ones. Post-cognitivist approaches provide new methodological tools for researching cognitive sciences. They advocate a methodological prescription that requires more focus on the body, time, and context than has been given so far. Within this reading, we would find claims such as those of Gomila (2008), Gomila and Calvo Garzón (2008) that advocate (re)attending to certain phenomena/problems/issues that could be reworked from this.
To address criticisms of SA, we suggest revisiting certain postcognitivist assumptions. By doing so, we can enhance the computational sophistication and design of natural language processing tools while better understanding the interconnection of language, context, and emotions. Such sophistications would be aimed at efficiently capturing (although the lexicon-based approach seeks to be independent of the domain) the nuances resulting from its relationship with the context (Dolianiti et al., 2019).
In this line, some post-cognitivist assumptions complement the analysis of sentiments and emotions, especially in its epi-stemological facet. Hovy (2015) noted that the challenge in SA extends beyond mere sentiment classification to encompass the development of more comprehensive explanations. Exploring techniques to achieve this represents an intriguing and enduring research endeavour.
6. CONCLUSION
The paper presents the following key findings. First, the SA is a handy tool to detect feelings and classify emotions as long as the context in which the students are inserted is kept in mind, as well as the postulates presented by the post- cognitivism (situated-embodied student). The sentiments and emotions associated with them are not the same in a pandemic context as in a context of absolute normality. The uncertainty of Covid exacerbated any excitement over familiar features like mandatory isolation. Second, regarding emotions, it was detected that students experienced anticipation and sadness during the pandemic, mainly with a high percentage of submission. Under the context
by Juan Antonio Dip and Maria Inés Silenzi
'Under the context analysis and the situated-embodied student theory, anticipation is related to the expectation of uncertain future events and negative emotions such as sadness, anger, and submission resulting from issues related to the pandemic (lockdown) and the shift to virtual learning'
analysis and the situated-embodied student theory, anticipation is related to the expectation of uncertain future events and negative emotions such as sadness, anger, and submission resulting from issues related to the pandemic (lockdown) and the shift to virtual learning. Third, we demonstrated how emotions are assigned to each word using SA based on lexicons to enhance the reader's understanding. Besides, diverse criticisms of SA were proposed using the post-cognitivist approach.
The relationship between NLP (SA) and the post-cognitivist approach is that the post-cognitivist approach enhances NLP applications by recognising emotions in language and considering how those emotions are expressed within the context where they occur. This means that it unevenly influences the processing and interpretation of the content. Explicitly speaking, keeping post-cognitivism in mind helps enormously when reading the results, with the need to look for other ways to complement the sentiment and emotion analysis method based on lexicons.
The SA methodology has limitations, such as the omission of new or domain-specific terms, which may lead to potential inaccuracies by not fully capturing the nuances of the context. Additionally, it may not account for the neutrality of certain words, which can introduce noise, and it may not classify sentiments or emotions. As previously mentioned, the sentiment of words and phrases can vary depending on the context and domain of the text. It is important to note that it can be challenging to distinguish sarcasm or irony, especially in Spanish among young people, where it is frequently used. This could lead to potential inaccuracies in this paper.
It is important to note that in university studies, especially in specific regions of countries (such as the one discussed in this paper), regionalisms and idioms can result in errors or inaccuracies in classifying sentiment and emotions. This is because these expressions represent a specific group of words with a meaning or sentiment different from the literal meaning of the individual words. For instance, Angacito/a, how quickly I solved the exam! This could have two meanings due to sarcasm: that he/she solved it quickly because it was difficult and he/she did not write anything or because he/she is intelligent and solved it very quickly. Angacito/a is a regionalism representing the diminutive of poor (poor boy/girl). In this context, it is an exclamation used to express self-compassion, sincerely or sarcastically.
Post-cognitivist theory and its derivations can aid in capturing the subtleties of the text that SA might miss. Thus, understanding the context in which students express their opinions, delving into various technological elements (such as cameras) and using qualitative tools (like interviews and focus groups)
can help analyse emotions and classify them accurately. However, some tools suggested here, such as developing new software, purchasing hardware for capturing images, student expressions, or gestures, and conducting randomised controlled experiments to support SA analysis, can be expensive.
Besides, post-cognitivism can help limit SA based on context-dependent lexicons. This is important because the meaning of words like good can vary based on the specific place and situation (the context). For example, good can be positive in a phrase like good class from the teacher, but it can be negative in a phrase like what a good beating the teacher gave me with the exam. Nevertheless, for researchers who are inexperienced or without knowledge of the post-cognitivism approach, it can be time-consuming, as they would have to spend time reading and may not be accustomed to the terminology used by the exponents and defenders of this approach.
Both approaches examine how emotions or emotional states influence our interpretation and perception of specific situations or events. The criticisms mentioned concern the accuracy and scope of their work. While both focus on studying human emotions, they must more effectively capture their expressions.
Our contribution goes in this direction: recognising that the human experience is a complex interaction, making it essential to consider the emotional nuances within the context in which they develop. Addressing this issue from the post-cognitivist approach is one of several ways to carry out this task.
Post-cognitivist approaches and other analysis recognition techniques are applied to different fields of economics and education. This includes evaluating undergraduate/postgraduate courses and students' perceptions of university education quality. Economic education research aligns with the latest advances in cognitive sciences, indicating a growing research area.
The present study has implications for implementing and conducting future research to improve existing lexicons or create new dictionaries. These must incorporate regionalisms or idiomatic expressions that take into account different contexts in which the analysis of student texts is carried out. In terms of practice, the paper shows how the post-cognitivist approach allows us to consider the student beyond their mind and the traditional mechanisation of learning (classical cognition). Thus, their texts and opinions can be interpreted through the concept that a student's behaviour is a cognitive activity that extends beyond the brain through the interactions the body maintains with the world, that is, its context.
To improve this paper in the future, it would be beneficial to conduct a more in-depth investigation into the role of post-cognitive approaches and their complements to other tools, such as machine learning, for analysing sentiments and emotions in languages like the one proposed in this paper. This is particularly important for Spanish, where such tools are scarce. By incorporating interdisciplinary research and qualitative tools, assembling lexicons for emotions and sentiment can be significantly improved, ultimately leading to more profound results in line with the methodology proposed in this article.
Training, Language and Culture Volume 8 Issue 2, 2024, pp. 52-65
References
Alizadeh, M., & Seilsepour, A. (2024). A novel self-supervised sentiment classification approach using semantic labeling based on contextual embeddings. Multimedia Tools & Applications. https://doi.org/ 10.1007/s11042-024- 19086-y Birhane, A. (2021). The impossibility of automating ambiguity. Artificial Life, 27(1), 44-61. https://doi.org/104162/artl_a_00336 Bradley, M. M., & Lang, R J. (1999). Affective norms for English words (ANEW): Instruction manual and affective ratings. University of Florida Press. https://pdodds.w3.uvm.edu/teaching/courses/ 2009-08UVM-300/docs/others/everything/bradley1999a.pdf Brand, A. G. (1987). The why of cognition: Emotion and the writing process. College Composition and Communication, 38 (4), 436-443. https://doi.org/10.2307/357637 Broad, C. D. (1954). Emotion and sentiment. Journal of Aesthetics and Art Criticism, 13(2), 203-214. https://doi.org/10.2307/425913 Burdman, F. G. (2015). El poscognitivismo en cuestión: Extensión, cor-porización y enactivismo. Principia, 19(3), 475-495. https:// doi.org/10.5007/1808-1711.2015v19n3p475 Calvo, R. A., & Kim, S. M. (2012). Emotions in text: Dimensional and categorical models. Computational Intelligence, 29(3), 527-543. https://doi.org/10.1111/j.1467-8640.2012.00456.x Casper, M. O., & Artese, G. F. (2022) Maintaining coherence in the situated cognition debate: What computationalism cannot offer to a future post-cognitivist science. Adaptive Behavior, 30(1), 3-17. https://doi.org/10.1177/105971232096705 Clark, A. (1998). Embodied, situated, and distributed cognition. In W. Bechtel & G. Graham (Eds.), A companion to cognitive science (pp. 506-517). Blackwell Publishing. Clark, A. (2008). Supersizing the mind: Embodiment, action, and cognitive extension. Oxford University Press. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1),
7-19. https://doi.org/10.1093/analys/5 8.1.7 Colombetti, G. (2010). Enaction, sense-making and emotion. In S. J. Inpenne & O. E. D. Raolo (Eds.), Enaction: Toward a new paradigm for cognitive science (pp. 145-164). MIT Press. Colombetti, G., & Thompson, E. (2008). The feeling body: Towards an inactive approach to emotion. In W. F. Overton, U. Muller, & J. L. Newman (Eds.), Developmental perspectives on embodiment and consciousness (pp. 45-68). Lawrence Erlbaum. Cortal, G., Finkel, A., Paroubek, P, & Ye, L. (2022). Natural language processing for cognitive analysis of emotions. Semantics, Memory, and Emotion. https://dx.doi.org/10.48550/arXiv.2210. 05296
Dawson, M. (2013). Mind, body, world: Foundations of cognitive science. Au Press.
Dokic, J. (2006). From linguistic contextualism to situated cognition: The case of ad hoc concepts. Philosophical Psychology, 19(3), 309-328. https://doi.org/10.1080/09515080600690102 Dolianiti, F., Iakovakis, D., Dias, S., Hadjileontiadou, S., Diniz, J., & Hadjileontiadis, L. (2019). Sentiment analysis techniques and applications in education: A survey. In M. Tsitouridou, J. Diniz, & T. Mikropoulos (Eds.), Technology and innovation in learning, teaching and education (pp. 412-427). Springer. https://doi.org/ 10.1007/978-3-030-20954-4_31 Ekman, P (1992). An argument for basic emotions. Cognition and Emotion, 6(3-4), 169-200. https://dx.doi.org/10.1080/0269993920 8411068
rudn.tlcjournal.org doi: 10.22363/2521-442X-2024-8-2-52-65
Gallagher, S. (2023). Embodied and enactive approaches to cognition. Cambridge University Press. https://dx.doi.org/10.1017/97810 09209793
Gardner, H. (1987). La nueva Ciencia de la Mente: Historia de la Revolución cognitiva. Editorial Paidós. Giménez-Moreno, R., & Ricart-Vayá, A. (2022). The expression of emotions in online medical consultations: A comprehensive Spanish-English analysis. Ibérica, 44, 239-262. https://doi.org/ 10.17398/2340-2784.44.239 Gomila, A. Y. (2008) Mending or abandoning cognitivism. In M. Vega, A. Glendberbg, & A. Grasser (Eds.), Symbols and embodiment: Debates on meaning and cognition (pp. 789-834). Oxford University Press.
Gomila, A. Y., & Calvo Garzón, F (2008). Handbook of cognitive science:
An embodied approach. Elsevier. Grimalt-Alvaro, C., & Usart, M. (2023). Sentiment analysis for formative assessment in higher education: A systematic literature review. Journal of Computing in Higher Education. https://doi.org/ 10.1007/s 12528-023-09370-5 Grishechko, E. G. (2023). Language and cognition behind simile construction: A Python-powered corpus research. Training, Language and Culture, 7(2), 80-92. https://dx.doi.org/10.22363/25 21-442X-2023-7-2-80-92 Grishechko, O. S., Akopova, A. S., & Grishechko, E. G. (2016). English language purism: History, development, criticism. Proceedings of Southern Federal University, Philology, 4, 185-192. EDN: VBUFXJ
Haugeland, J. (1988). Mind embodied and embedded. In J. Haugeland (Ed.), Having thought: Essays in the metaphysics of mind (pp. 207-237). Harvard University Press. Hasselman, F. (2020). Radical embodied computation: Reproduction of similarity by analogy as an order generating mechanism in complex systems. PsyArXiv Preprints, 1-30. https://dx.doi.org/ 10.312 34/osf.io/2htkr Henríquez Miranda, C., & Guzman, J. (2017) A review of sentiment analysis in Spanish. Tecciencia, 12(22), 35-48. https://doi.org/ 10.18180/tecciencia.2017.22.5 Heras Escribano, M. (2021). Pragmatism, enactivism, and ecological psychology: Towards a unified approach to post-cognitivism. Synthese, 198, 337-363. https://dx.doi.org/10.1007/s11229-01 9-02111-1
Hipólito I., & van Es, T. (2022). Enactive-dynamic social cognition and
active inference. Frontiers in Psychology, 13, 855074. Hollan, J., Hutchins, E., & Kirsh, D. (2000). Distributed cognition: Toward a new foundation for human-computer interaction research. ACM Transactions on Computer-Human Interaction, 7(2), 174-196. https://doi.org/10.1145/353485.353487 Hovy, E. H. (2015). What are sentiment, affect, and emotion? Applying the methodology of Michael Zock to sentiment analysis. In N. Gala, R. Rapp, & G. Bel-Enguix (Eds.), Language production, cognition, and the lexicon (pp. 13-24). Springer. https://doi.org/ 10.1007/978-3-319-08043-7_2 Hulliyah, K., Bakar, N. S., & Ismail, A. R. (2017, November 1-3). Emotion recognition and brain mapping for sentiment analysis: A review. In Proceedings of the 2nd International Conference on Informatics and Computing (ICIC) (pp. 1-5). IEEE. https://doi.org/ 10.1109/IAC.2017.8280568
Hurley, S.L. (1998). Consciousness in action. Harvard University Press. Jiménez-Zafra, S. M., Cruz-Díaz, N. P, Taboada, M., & Martín-Valdivia, M. T. (2021). Negation detection for sentiment analysis: A case study in Spanish. Natural Language Engineering, 27(2), 225-248. https://doi.org/ 10.1017/S1351324920000376 Jockers, M. L., & Thalken, R. (2020). Text analysis with R: For students
of literature (2nd ed.). Springer. Kratzwald, B., Ilic, S., Kraus, M., Feuerriegel, S., & Prendinger, H. (2018). Deep learning for affective computing: Text-based emotion recognition in decision support. Decision Support Systems, 115, 24-35. https://doi.org/10.1016/j.dss.2018.09.002 Manca, D. (2022). The situated mind and the space of reasons. On the match and mismatch between pragmatism and phenomenology European Journal of Pragmatism and American Philosophy, 14(2), 3000. https://doi.org/10.4000/ejpap.3000 Marsh, L. Y., & Onof, C. (2008). Introduction to the special issue: Perspectives on social cognition. Cognitive Systems Research, 9, 1-4. https://doi.org/10.1016/j.cogsys.2007.08.001 Mohammad, S., & Turney, P (2010, June 5). Emotions evoked by common words and phrases: Using mechanical Turk to create an emotion lexicon. In Proceedings of the NAACL-HLT 2010 Workshop on Computational Approaches to Analysis and Generation of Emotion in Text (pp. 26-34). The Association for Computational Linguistics. Mohammad, S. M., & Turney, P D. (2013). Crowdsourcing a word-emotion association lexicon. Computational Intelligence, 29(3), 436-465. https://doi.org/10.1111/j.1467-8640.2012.00460.x Mohammad, S. M., & Turney, P. D. (2024). NRC word-emotion association lexicon. Saif Mohammad. https://saifmohammad.com/ WebPages/NRC-Emotion-Lexicon.htm Nandwani, P, & Verma, R. (2021). A review on sentiment analysis and emotion detection from text. Social Network Analysis and Mining, 11, 81. https://doi.org/10.1007/s13278-021-00776-6 Newen, A., Bruin, L., & Gallagher, S. (2018). The Oxford handbook of
4E cognition. Oxford University Press. Noe, A. (2004). Action in perception. The MIT Press. Panikar, R., Bhavsar, R., & Pawar, B. V (2022, March 25-26). Sentiment analysis: A cognitive perspective. In Proceedings of the 8th International Conference on Advanced Computing and Communication Systems (ICACCS). IEEE. https://doi.org/104109/ICACC-S54159.2022.9785027 Pessoa, L. (2013). The cognitive-emotional brain: From interactions to integration. MIT Press. Plutchik, R. (1980). Emotions: A general psycho-evolutionary theory. In R. Plutchik & H. Kellerman (Eds.), Theories of emotion (pp. 3-33). Academic Press. https://dx.doi.org/10.1016/B978-0-12-558701-3.50007-7
by Juan Antonio Dip and María Inés Silenzi
Pluwak, A. M. (2016). Towards the application of speech act theory to opinion mining. Cognitive Studies, 16, 33-44. https://dx.doi.org/ 10.11649/cs.2016.004 Reggin, L. D., Franco, L. E. G., Horchak, O. V, Labrecque, D., Lana, N., Rio, L., & Vigliocco, G. (2023). Consensus paper: Situated and embodied language acquisition. Journal of Cognition, 6(1), 63. https://doi.org/10.5334/joc.308 Rendón-Cardona, P, Gil-Gonzalez, J., Páez-Valdez, J., & Rivera-Henao, M. (2022). Self-dupervised sentiment analysis in Spanish to understand the university narrative of the Colombian conflict. Applied Sciences, 12(11), 5472. https://dx.doi.org/10.3390/ap-p12115472
Rowlands, M. (2010). The new science of the mind: From Extended mind
to embodied phenomenology. MIT Press. Rupert, R. (2010). Cognitive systems and the extended mind. Oxford University Press.
Shaik, T., Tao, X., Dann, C., Xie, H., Li, Y., & Galligan, L. (2023). Sentiment analysis and opinion mining on educational data: A survey. Natural Language Processing Journal, 2, 100003. Shapiro, L. (2019). Embodied cognition. Routledge Press. Silenzi, M. (2012). Algunos aportes del enfoque incrustado de las Ciencias cognitivas a las ciencias de la educación. El rol del entorno en las prácticas educativas. Prometeica, 6, 48-65. https://doi.org/ 10.24316/prometeica.v0i6.31 Stanciu, D. (2023). Consciousness, 4E cognition and Aristotle: A few conceptual and historical aspects. Frontiers in Computational Neuroscience, 17, 1204602. https://dx.doi.org/10.3389/fncom.2 023.1204602
Thompson, E. (2007). Mind in life: Biology, phenomenology and the sciences of mind. Harvard University Press. Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind. MIT Press.
Varela, F., Thompson, E., & Rosch, E. (2017). The embodied mind: Cognitive science and human experience. MIT Press. Villalobos, M., & Dewhurst, J. (2017). Why post-cognitivism does not (necessarily) entail anti-computationalism. Adaptive Behavior, 25(3), 117-128. https://doi.org/10.1177/1059712317710496 Wankhade, M., Rao, A. C. S., & Kulkarni, C. (2022). A survey on sentiment analysis methods, applications, and challenges. Artificial Intelligence Review, 55, 5731-5780. https://dx.doi.org/10.1007/ s10462-022-10144-1 White, C., Ratcliff, R., Vasey, M., & Mckoon, G. (2009). Dysphoria and memory for emotional material: A diffusion model analysis. Cognition and Emotion, 23(1), 181-205. https://dx.doi.org/104 080/02699930801976770 Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin and Review, 9, 625-636.
Juan Antonio Dip
ORCID 0000-0003-3714-2478 и Juan.dip@fce.unam.edu.ar Universidad Nacional de Misiones, Argentina
María Inés Silenzi
ORCID 0000-0002-3003-6261 и misilenzi@uns.edu.ar Universidad Nacional del Sur, Argentina
