Content uploaded by José Manuel Ramírez del Pozo Martín
All content in this area was uploaded by José Manuel Ramírez del Pozo Martín on Jun 07, 2022
Content may be subject to copyright.
Method based on
A methodological interface between Appraisal Theory and the
contextual models of Critical Discourse Analysis
José M. Ramírez (PhD)
Draft translation from Spanish to English 2022
(thesis extract, short version)
Source: “Análisis de los sistemas de valoración lingüística en la obra de Santiago
Ramón y Cajal”, by José Manuel Ramírez del Pozo Martín (doctoral thesis in Spanish).
Thesis Director: Laura Alba-Juez (PhD)
The author welcomes any comments, corrections, and observations on this draft.
Phone: (+34) 679 76 196
The priority challenge of this research was to try to reconstruct social values that
monitored the scientist's discourse.
The Method Based on Constituents (MBC) that we present was designed for the analysis
of evaluative resources both in the textual and in the pretextual phase, and taking into
account the peculiarity of the work of Santiago Ramón y Cajal.
In other words: firstly we try to analyze textual realizations, and secondly we try a
schematic and hypothetical reconstruction of decisive concepts in the thought of Ramón
y Cajal. These crucial concepts are primarily place/scenario and purpose, categories
(human, animate and inanimate participants), social groups, ideologies, and social values.
We have called it Method Based on Constituents or Methodology Based on Constituents
because it is based on constituents at four stages of their development.
First stage. Although the syntactic unit of analysis is the clause, we have analyzed
syntactic constituents of the clause as possible markers of the main valuation. In
a systematic way, we have analyzed the verbal nucleus, adjectives, and adverbs,
but also, in a non-systematic way, other syntactic constituents.
Second stage. We have established in contextual models a series of conceptual
constituents, the same in the six analyzed texts, according to the visual synthesis
template that we insert below. In a later stage of analysis, this allows us to compare
the six contextual models.
Third stage. The social values obtained are not idealizations proposed in an a
priori way, but are constituted by the textual evaluative realizations themselves.
They are presented as instructions for the evaluative realizations to which they
have given rise. Their conceptualizations have a hypothetical and potential
Fourth stage. Nor are possible ideologies raised in an a priori way, but are
constituted by the elements of the contextual model typical of the relevant social
group of the author in each discourse: especially social values, categories of the
participants, and their relationship, purpose, and place or scenario.
Note: other possible factors and mediators in communication, such as shared
knowledge, companies, or institutions, might be included in the analysis depending on
the text and its context.
Introduction to Part III: The Methodology or Method
Objectives of our research, hypothesis, corpus, theoretical framework, and the
individual character of doctoral research determine the method that we present. It is a
quantitative-qualitative method, divided into two areas: on the one hand and mainly, the
analysis of evaluations; on the other hand, and with a secondary and dependent character,
the contextual models. We have developed the quantitative method applied to the analysis
of the interpersonal semantics of the corpus on the basis of some very recent
methodological contributions. We have started from the evaluative functional
relationship proposed by Alba-Juez (2017), which has allowed us to define the what of
our analysis. The design of the method is an adaptation and extension of the seven-step
proposal by Fuoli (2018), whose purpose is to optimize reliability, replicability, and
transparency. In the first place, we quantify the frequencies of use of the evaluation
systems in six texts of different genres written by the same author, Ramón y Cajal. The
technique chosen is manual annotation, assisted by the UAM Corpus Tool software.
Secondly, for our reconstruction of contextual models proposed by critical discourse
analysis, we have developed an ad hoc qualitative interface that applies abductive
reasoning (Eco, 1983; Aguayo, 2011) and is supported by a series of parameters of the
theoretical frameworks used. The purpose is to reconstruct the schemes of contextual
models, whose character is hypothetical. We will give special relevance to trying to
rebuild social values. Thus, our method applied to discourse analysis is an attempt, whose
validity will depend on the explanatory capacity of these contextual models.
1. Research questions and hypotheses
The objective of this thesis is to explore the cultural foundations of the work of
Santiago Ramón y Cajal, as an exponent of the international scientific community and,
on a more general level, of the society of the time. We intuit with Bally ( 1965)
that evaluation is a linguistic mechanism that permeates all linguistic productions with
emotion, but we also consider, with John Dewey ( 2008a), that evaluation
participates in our knowledge of the world and in their communication, and with
Voloshinov (1992), that every text is dialogical, that is, that it establishes a dialogue in a
broad sense with previous and subsequent texts. Thus, we consider that evaluation is also
an expression of science and culture.
The first problem we face is the feasibility of applying Appraisal Theory and
Critical Discourse Analysis to texts of a scientist, in this case, Ramón y Cajal. There are,
to our knowledge, no precedents for a linguistic analysis of the work of Ramón y Cajal,
nor do we know, at the time of writing this thesis, that the analysis of research texts in
Spanish of the so-called exact, physical, and natural sciences, has been approached
systematically with one of these two theories, although Appraisal Theory has been
applied to popular science texts in English (Fuller, cited by Martin and White, 2008: xi).
We recall that, for Van Dijk, Critical Discourse Analysis, given its eclectic and
multidisciplinary nature, is compatible with other types of discourse analysis, and we also
defended pages before the relevance of applying this theoretical framework to the analysis
of a histology text (II, 2.4.2).
This research had as germ a master's thesis in which we already qualitatively
analyzed the evaluative realizations of the article by Ramón y Cajal «Estructura de los
centros nerviosos de las aves» (“Structure of the Nervous Centers of Birds”, 1888),
confirming its presence. When studying various texts of the scientist, we observe very
striking facts. For systemic-functional linguistics, the three contextual parameters that
model a text are field, mood, and tenor, which will therefore be our starting point. In a
first approximation, in that master's thesis we observed, for example, that although in his
scientific article Ramón y Cajal did not seem to express emotional reactions when
describing the nerve cells he observed through the microscope, he did, and very
noticeably, when describing in his memoirs the same nerve cells:
El jardín de la neurología brinda al investigador espectáculos cautivadores y emociones
artísticas incomparables. En él hallaron, al fin, mis instintos estéticos plena satisfacción.
¡Como el entomólogo a caza de mariposas de vistosos matices, mi atención perseguía, en el
vergel de la sustancia gris, células de formas delicadas y elegantes, las misteriosas mariposas
del alma, cuyo batir de alas quién sabe si esclarecerá algún día el secreto de la vida mental!
(Ramón y Cajal, 1981: 98-99)
Changes in field, mode, and tenor, according to Appraisal Theory, should allow
us to explain such relevant differences in the expression of aesthetic emotions, apparently
neutralized in the scientific article of 1888. In that first approximation in our thesis
master's degree, however, we also found at least one surprising evaluation that seemed to
elude these three parameters. «Diríase que cada elemento es un cantón fisiológico
absolutamente autónomo» (“It could be said that each element is an absolutely
autonomous physiological canton"), this is what Ramón y Cajal writes in his 1888 article,
introducing in a histological text a daring cognitive metaphor whose source is a political
concept, and expressing it through a heteroglossic resource, the impersonal voice and the
conditional, which allow him to distance himself from his own statement. It is, in other
words, a key metaphor in Ramón y Cajal's discursive strategy, but whose source domain
is deliberately and explicitly alien to the experiences of the field of a histology article.
At this point in our research, our purpose is to quantify the frequencies of use of
evaluative systems, to obtain data that allow us to compare such frequencies in scientific
texts and those of other genres by the same author, and to approach the contextual factors
and sociocognitive mechanisms that can explain the supposed variations of these
Thus, we can define the following research questions:
Are there differences in the frequency of use of Ramón y Cajal's evaluation
systems in different texts, typical of different contexts and discursive genres?
On what contextual factors depend the variations, if these variations exist?
Does sociocognition influence evaluation? It does so through contextual models?
The garden of neurology offers the researcher captivating shows and incomparable artistic emotions. My
aesthetic instincts found in it, at last, full satisfaction. Like an entomologist hunting for butterflies of
colorful hues, my attention was chasing, in the orchard of the gray substance, cells of delicate and elegant
shapes, the mysterious butterflies of the soul, whose flapping of wings who knows if one day will clarify
the secret of mind life!
Our guiding hypothesis is the following:
The linguistic evaluation systems used by Santiago Ramón y Cajal vary according
to discursive genres.
The second hypothesis or sub hypothesis is the following:
Some evaluative realizations can only be explained by means of the notion of
2. Corpus and research method
2.1. The corpus
Santiago Ramón y Cajal is the author of an enormous written production. Our
corpus is made up of six texts corresponding to six different discursive genres, which we
define succinctly according to field, mode and tenor: 1, a histological scientific article,
originally published in a research journal self-published by the scientist himself and
directed to international specialists in their field of knowledge; 2, an essay on science,
which although it was originally his entrance speech at the Academia de Ciencias Exactas,
Físicas y Naturales (Academy of Exact, Physical and Natural Sciences), in 1897, it was
soon published for the first time in a book, in 1898, and it is explicitly and primarily aimed
at young people students who start in any branch of scientific research; 3, a technical
photography manual, published in a book and aimed at fans of this artistic discipline; 4,
an autobiography, published as a book and addressed to any cultured reader interested in
the life of the scientist; 5, a volume of aphorisms on various subjects, published as a book
and addressed to any cultivated reader; and 6, a fictional story, self-published as a book
and aimed at any cultivated reader. These texts have been chosen taking into account, in
addition to their mutual differentiation according to discursive genres, their scientific and
cultural relevance, which we will explain very briefly below. Due to the limitations in
time and resources of doctoral research, we have selected fragments that together do not
exceed twelve thousand words. We take text 1 as a statistical reference, so, despite its
length, we will analyze it completely. These are the texts of our corpus:
Text 1. The article «Estructura de los centros nerviosos de las aves» (“Structure
of the Nervous Centers of birds”, 1888). The author rejects the reticular hypothesis with
data and advances for the first time the hypothesis that will later give rise to the neuronal
Text 2. First chapter of Reglas y consejos sobre investigación científica. Los
tónicos de la voluntad (Advice for a Youg Investigator. Tonic of the Will  2019: 23-
31). The author defends the scientific method and the role that the will plays in research.
Text 3. Chapter XVIII of Fotografía de los colores (Color Photography 
2007). The author describes the principles, techniques of industrial reproduction, virtues,
and defects of the photography of the time. Let us remember that the shortcomings of
photography at the time led him to communicate his histological observations through
Text 4. Sixteen paragraphs of Recuerdos de mi vida (Memories of my Life 
2017: 199-201). The author recalls his 1888 findings in the context of his autobiography.
Text 5. The first twenty aphorisms of «Pensamientos de tendencia pedagógica y
educativa» (“Pedagogical and Educational Trend Thoughts”), chapter VIII of Charlas de
café. Pensamientos, anécdotas y confidencias (Coffee Talks. Thoughts, Anecdotes and
Confidences  1967: 213-218). The author reflects on the will, knowledge,
education, and creativity in science.
Text 6. Part I of «El fabricante de honradez» (“The Honesty Maker”), from
volume Cuentos de vacaciones. Narraciones seudocientíficas (Vacation Stories.
Pseudoscientific narratives  1999: 63-67). The author narrates a peculiar fictional
pseudoscientific experiment. This experiment tests positivism and will, as well as the
relationship between science and society.
2.2. The research method
We are going to expose a mixed, quantitative-qualitative method. Through the
quantitative method, we will explore the evaluative realizations with the aim of obtaining
data that allow us to statistically compare the frequencies of use of the evaluation systems
in the six analyzed texts. Through the qualitative method, on the one hand, we will try to
interpret and explain the reason for the variations in the use of the frequencies of use of
the evaluative systems, and on the other hand, applying abductive reasoning to the
evaluations found in the previous analysis. We will try to make a reconstruction of the
mental schemes that constitute the possible contextual models and, in particular, the social
values. If Critical Discourse Analysis is correct, the contextual model would consist of a
complex series of schemes that can be specified in textual evaluative realizations: these
schemes must be composed, at least, of purposes, issues, actors, social representations,
scenarios/places, values, and discursive strategies, as well as ideologies. In terms of Van
Dijk, we will speak of values and attitudes as mental models, and by social
representations, we will understand how the speaker or writer configures in his mind the
participants mentioned in his text and their interlocutors, as well as the relationships that
are established among themselves and with himself. In this abductive reasoning, we will
not rely on mere creative intuition, but rather we will follow a series of parameters
obtained from Critical Discourse Analysis, such as the notions of social actor, ideological
group, and power, to which we are going to add parameters typical of systemic-functional
linguistics. These contextual models will be hypothetical; as new hypotheses, their
acceptance will depend on their ability to explain evaluations. It is an attempt to develop
an interface that links appraisal with the mental schemata that constitute contextual
This mixed method, with a double theoretical framework, involves several issues,
of which we highlight at least three. The first is the priority we give to valuation in our
analysis, whose data are textual realizations. The second is that our center of interest will
later move from the text to the psyche, the mind of the speaker or writer, who subjectively
assesses the context of communication; with this, we do not mean that we can enter the
author's mind, but rather we will try to reconstruct the mental schemes that the author
considered relevant when writing each of his texts and that, therefore, must have left clues
in them. The third, which follows from the previous one, is that in the contextual model
created by the speaker or writer we expect to find the interface between evaluation and
social representations or ideologies. We could formulate this last question with a
rhetorical question: what objective social circumstances of communication determine the
evaluations present, for example, in the story «The Honesty Maker»?
2.2.1. Evaluation systems analysis method
According to Rodríguez Lifante (2016: 29), the method designates the entire
research process, whose choice is determined by what, while the design establishes a
specific procedure for obtaining data, and the technique, finally, designates the
instruments with which data are collected. This categorization forces us to try to be
exhaustive in the development and exposition of the method.
Let us remember that the impetus for the development of Appraisal Theory was a
field study undertaken in Australia between 1990 and 1995, the literacy project
Disadvantaged Schools Program's Write it Right; that is, the challenges of this program
and the data obtained were the experimental basis for the development of the Appraisal
Theory (Martin and White, 2008: xi), so, in no way the theory was a prior construct that
was imposed on data. However, although Martin and White (2008) presented a topology
of evaluative resources in the English language, organized into three domains of
interpersonal semantics, they did not present a method of analysis. One of the objections
to their approach lies in the fact that the examples of their theory could be taken ad hoc
to illustrate their categories and that these categories are unproblematic cases that are
presented as self-evident (Fuoli, 2018: 3).
In recent years, other researchers have been developing qualitative and
quantitative research methods adapted to the requirements of research purposes and
corpus (Fuoli and Glinn, 2013; Fuoli and Hommerberg, 2015; O'Donnell, 2014; Taboada
et al, 2014; Alba-Juez, 2017). Thus, there has been a progressive methodological
refinement. The evaluative functional relationship presented by Alba-Juez (2017) has
allowed us to clearly delimit the what of our research: the phase, the linguistic levels, the
degree of (in)directness, position along the evaluation continuum (position between two
poles), parameters of evaluation and the mode. Our design is an adaptation and
specification of the seven steps proposed by Fuoli (2018), which this author has theorized
after several previous discourse analyses, with the aim of optimizing transparency,
reliability, and replicability. We also note that these methodological difficulties are shared
by other branches of linguistics and semantic analysis. The method designed by Gerard
J. Steen and his team (2010) for the detection of linguistic metaphors has served us, if not
as an example or guide, at least as an inspiration, precisely because of its transparency,
reliability, and replicability.
126.96.36.199. The what
We consider appraisal, valuation or evaluation, with Alba Juez (2017: 6), as an
evaluative functional relationship, where evaluation is treated as an equation of six
E = F(PH, Ll, Deg, ContPos, P, Mo)
In the preparation of our method, we start from this same definition and these
arguments or variables: phases of evaluation (PH), linguistic levels (Ll), degree of
(in)directness (Deg), position along the evaluation continuum (ContPos), parameters (P)
and the mode (Mo). Our quantitative analysis, as we will argue below, will target the
textual phase, the lexical, morphological, and semantic-pragmatic levels, all degrees of
(in)directness, the position in the evaluative continuum between two poles, the parameters
provided by the topology of Martin and White's interpersonal semantics and the written
Phase of evaluation (PH). Considering three valuation phases (Alba-Juez y
Thompson, 2014), pre-realization, textual and metaevaluative, in our quantitative analysis
of the evaluations we will focus on the second, the textual. The metaevaluative phase
would imply a task of introspection about our own reactions as interlocutors of the texts
of Santiago Ramón y Cajal, a reaction mediated by our condition as linguists outside the
communities of readers and histologists of the late nineteenth and early twentieth
centuries. Then, the results would shed more light on our reaction to the texts than on the
reaction of the intended recipients of the scientist. On the other hand, as we will see, in
our subsequent qualitative analysis we will try to enter the pre-realization phase, which
we are going to consider, following Van Dijk, as a sociocognitive phase, to the extent that
evaluative realizations can be the result of sociocognitive processes based on shared
social values, social representations, and human groups. In the analysis of this textual
phase, we will take into account the context in a broad sense, including sociocultural
elements that shed light on the production of evaluative realizations. An example is the
unusual cognitive metaphor implicit in the title of the story "The Honesty Maker", which
we could formulate as HONESTY IS AN INDUSTRIAL PRODUCT, which could
indicate a heteroglossic resource with an ironic function, as is later confirmed in the rest
of the story.
Linguistic level (Ll). Our quantitative analysis will especially study the lexical,
morphological, and semantic-pragmatic levels. The lexical level is central in the
interpersonal semantics of Martin and White and it is the most easily identifiable.
For cognitive or linguistic metaphor detection, we will follow the theoretical framework of Lakoff and
Johnson (2003) and a succinct ad hoc adaptation of Gerard J. Steen's (2010) linguistic metaphor detection
method. For its analysis, we will use as a tool the Spanish RAE dictionary (2001) and Spanish Alonso
Pedraz's dictionary (1982), which also provides meanings from the second half of the 19th century and the
beginning of the 20th, the time when these texts were written by Ramón and Cajal. In short, we will consider
as linguistic metaphors those expressions from which the dictionaries collect experiential meanings that are
more basic than that of the subject matter. We will only analyze those metaphors that, because they are
explicit, direct, or unusual, are very evident in the analyst's opinion and may point to or participate in an
However, Spanish is a language that is especially rich in morphological variations: thus,
the adverb dificilísimo (extremely difficult, Ramón y Cajal, 1888: 4) incorporates an
intensification of the lexeme difícil. Consequently, we must be attentive to the possible
emotional, affective, and pejorative uses of affixes. We anticipate that the analysis of the
semantic-pragmatic level, although it also rests on the analyst's subjectivity, depends
above all on his knowledge of the cultural context and pragmatics, which should allow
him to identify inferences such as presuppositions or conventional and conversational
implicatures, or a simple syllogism. However, prosodic resources such as quotation
marks, bold and underlining may be indicators or markers of a possible lexical or
semantic-pragmatic evaluative realization. A prosodic resource such as, for example, the
repetition of the adverb jamás (never) by Ramón y Cajal in his 1888 article, will be
quantified by the frequency of use of this lexical unit with an evaluative function. We
cannot fail to observe that the choice of jamás (never) instead of nunca (never) by the
scientist implies an intensification at the phonological level, due to the phonetic
forcefulness required by the pronunciation of the first word. Regarding the syntactic level,
resources such as thematization can also serve as indicators or markers of possible
semantic evaluative realizations.
Degree of (in)directness (Deg). In our analysis, we are going to consider both
direct and indirect realizations or, in terms of Martin and White, respectively, inscribed
and invoked evaluations. However, Fuoli (2018: 6) warns that this distinction is not
always clear. The analysis of the evaluative realizations invoked, which are contextually
dependent, relies in part on the subjectivity of the analyst, as Martin and White (2008:
62) already warn. To this problem, we must add the one called “Russian doll síndrome”
by Thompson (quoted by Alba-Juez, 2017: 4, and Fuoli, 2018: 6), a token in which an
inscribed evaluation can be interpreted to invoke other evaluations recursively and whose
categorization or type must be resolved contextually, taking into account the whole text.
Position along the evaluation continuum (ContPos). As it is explained by Alba-
Juez (2017: 20), the most accessible polarity parameter is positive-negative or good-bad;
however, “evaluative language can be found at any of the different points of a continuum
that includes different positions from one pole to the other”.
Parameters of evaluation (P). All the authors consulted agree on highlighting the
topology developed by Martin and White (2008) as the most complete and elaborated to
date. We can say that fifteen years after its first book publication, in 2005, the taxonomy
composed of the three domains ATTITUD, GRADUATION and ENGAGEMENT has become
the standard for evaluative realizations analysis. We have already described it in detail
when exposing the theoretical framework (I, 1.2.), but when approaching the design of
the method, we will try to establish decision criteria for the most problematic cases.
Mode of the evaluation (Mo). Our analysis will focus on the written mode. At
various points in this thesis, we have mentioned the important role that Santiago Ramón
y Cajal's drawings played in communicating his scientific findings. In our master's thesis,
we already did a linguistic analysis of the two drawings that illustrate the 1888 article,
applying the functional socio-semiotics of Kress and Leeuwen (2006) as a theoretical
framework. We have other drawings of the scientist (DeFelipe et al, 2007), made in his
adolescence and youth, but none of the other five texts that we analyze are illustrated by
the author, so we could not compare the interpersonal semantics present in his two
histological drawings according to discursive genres and concerning the chosen texts. The
Legado Cajal (Cajal Legacy) put us on the trail of the recording of two speeches by the
scientist that are preserved in the Biblioteca Nacional de España, in Madrid (National
Library of Spain, Ramón y Cajal, 1930). However, we have been able to verify that they
are readings of a selection of aphorisms from the volume Charlas de café (Coffee Talks).
On the other hand, we see that evaluation is not only characteristic of verbal language,
but that it is also a semiotic function. We wish to anticipate in this section a reflection on
another possible heteroglossic resource; because if we understand that dialogism is not
only linguistic, but also semiotic, we can observe that, for example, the mention of a
statue, a Byzantine Christ, in «El fabricante de honradez» ("The Honesty Maker", 1999:
64) would also be a heteroglossic resource: the statue is not present in the fictional
imaginary, in the narrative situation, but rather it is introduced as a quote to compare it
with the main character. It is a semiotic product whose mention could be considered an
expansive heteroglossic resource, which invites the formulation of dialogical alternatives
and, in this specific case, responds to the parodic intention of the text and is carried out
as irony. For the purposes of our analysis decisions, this case presents us with a problem:
whether we should categorize this mention as a heteroglossic resource, as an attitudinal
resource at the semantic-pragmatic level (Judgment? Appreciation?), or as both, that is,
heteroglossic and attitudinal at the same time. Martin and White (2008: 59-65) expressly
allude to semiotic products as the target of the evaluation, but in their analysis of indirect
realizations they do not deal with the incorporation of non-linguistic semiotic products
outside the immediate context of communication into the text; they do deal with
“comparable metaphors” of the type “like a market” or “like a mob of cattle”, which they
consider an attitudinal resource. In any case, Martin and White do not consider similar
cases as expansive heteroglossic resources of the ENGAGEMENT domain (2008: 111-
116). Thus, for consistency with our main theoretical framework, we will treat these
resources as merely attitudinal.
188.8.131.52. Design and technique
According to Fuoli (2018: 1), manual annotation facilitates a detailed and
comprehensive analysis, which would not be possible with merely automatic techniques,
because evaluation in discourse is contextually dependent, and, at the same time, the data
obtained in the manual annotation contribute to the construction of the theory and to refine
the model. We have already referred to the finding, by Martin and White, of the role that
subjectivity of the linguist plays in the analysis of texts. Fuoli (2018: 2) notes that to date
there is no standardized analysis protocol, and argues that this lack is an obstacle to
achieving analyzes that are both transparent and replicable.
Fuoli outlines several challenges in the analysis of interpersonal semantics and
divides them into challenges for identification and challenges for classification. Among
the challenges for identification (2018: 4-18), he lists the following:
1, evaluations can be made using a wide range of expressions of variable length
and complexity, expressions that are characteristic of any type of word, so it is not
possible to compile a definitive list;
2, appraisal is a highly context-dependent phenomenon;
3, according to what criteria we establish for the identification of the evaluation
units (unitizing), their textual boundaries will be different, which can lead to very different
results, especially in comparative analyses; in other words: what evaluation unit will we
choose? Two problems related to this are: a) the coordination of evaluative realizations
through the copulative conjunction, and b) the discontinuity of some evaluative
4, the distinction between inscribed and invoked evaluations and the problem,
already mentioned pages before, of Russian doll syndrome, that we will try to solve by
attending to the distinction between token and type suggested by Thompson (quoted by
Fuoli, 2018: 7, and Alba-Juez, 2017); another related problem is the possible existence of
multiple evaluative invocations, also observed by Thompson;
and 5, other problems related to the GRADUATION and ENGAGEMENT domains;
Fuoli mentions, for example, that although negation requires a heteroglossic resource, in
some cases its use is not "intersubjective", but “objective”, since sometimes it seems to
fulfill a descriptive function.
Among challenges for classification, Fuoli (2018: 8-13) lists the following:
1, the more one refines or descends in the classification of the semantic domains
and their subdomains, the more plausible several categorizations become; this challenge
can be resolved by allowing double and even multiple labeling, according to Macken-
Horarik and Isaac (cited by Fuoli, 2018: 9);
2, some expressions do not seem to fit entirely into any of the available categories,
either because the generic valuations are semantically underspecified, or because they are
highly context-dependent; according to Fuoli (2018: 9), an effective way to achieve a
more precise analysis is to adapt the framework to the specificities of discourse;
3, the very distinction between categories in the taxonomy is not always precisely
delimited; again, Fuoli (2018: 10) mentions Thompson for these cases and his proposal
to stick to the objective of the evaluation (“target at face value”), although he also
observes that the solution becomes more complicated as the analysis descends in semantic
4, some evaluative realizations do not refer to factual objectives, to facts, but to
possible or unrealistic scenarios, for which Fuoli distinguishes between actual and irrealis
scenarios, where evaluations play a different discursive role;
and 5, boundaries between dialogical Expansion and Contraction are not clear,
especially as it affects Entertain and Proclaim; Fuoli himself recalls that, for Martin and
White (2008: 103), the function of ENGAGEMENT expressions can vary systematically
under the influence of different co-textual conditions.
Fuoli proposes the development of an "Annotation Manual" that includes
identification and categorization based on three methodological principles: reliability,
replicability, and transparency. He distinguishes three types of reliability (2018: 13): a)
test-retest, which guarantees the precise reproduction of the results obtained by the same
analyst at time intervals; b) internal consistency, which ensures that the annotator treats
similar textual elements throughout the corpus in the same way; and c) interrater, which
guarantees that different annotators assign the same categories to textual units.
Unfortunately, the conditioning factors of doctoral research won't allow us to measure
this third type of reliability, the best, which can only be adjusted by comparing the
analyzes carried out by different analysts and the subsequent agreement on the coding
Replicability depends on reliability criteria and annotation guidelines being
explicitly formulated and made available to other analysts. Then, this is crucial to ensure
transparency (Fuoli, 2018: 13).
For the elaboration of this “Annotation Manual”, Fuoli proposes to follow seven
steps, under three procedural principles: 1) the explanation of all the elections; 2) that
annotation guidelines be tested and refined to maximum reliability, and 3) reliability is
assessed and its indicators are reported and discussed. Below we summarize the seven
steps proposed by Fuoli (2018: 14-21):
Step 1. Defining the scope of the project, by selecting the relevant categories,
which should be included in the draft “Annotation Manual” to apply to a random selection
of corpus fragments.
Step 2. Selecting and configuring an annotation tool, in our case the UAM Corpus
Tool. Fuoli (2018: 16-17) compares two corpus annotation tools, CAT (Content
Annotation Tool) and UAM Corpus Tool, tool that O’Donnell (2014) describes. The
chosen technique is UAM Corpus Tool, which we have been able to install on our
computer and that, in addition to allowing us to extract statistical data, has allowed us to
define the coding scheme and establish and display the categorical hierarchies; likewise,
it has allowed us to nest an analysis unit within another, a fundamental characteristic of
the annotation manual and method that we will propose, based on constituents.
Step 3. Writing the draft of the “Annotation Manual”; this draft must include the
main lines of the annotation scheme, categories, rules to be applied to the analysis, the
identification criteria of the analyzed units, as well as other contextual elements relevant
to the analysis, such as the intended recipient of data, or the main purpose of
Step 4. Evaluating reliability. In our case, we will test the reliability test-retest,
discarding both internal consistency, given the possible difficulties of application in
different discursive genres, and interrater. Fuoli indicates that both identification and
categorization should be assessed. Basically, this evaluation will consist of comparing
two annotations made on different occasions and, in our case, in the six texts. Fuoli and
Hommerberg, cited by Fuoli (2018: 19), describe the markers and coefficients that can be
used to assess reliability. We will explain our reliability test-retest in the “Annotation
Step 5. Refining the “Annotation Manual”. In response to the results obtained and
the difficulties encountered in applying the first draft to the analysis, it must be rectified.
Steps 5 and 4 constitute a kind of loop, a dynamic and iterative process. Whenever
necessary, categories should be rectified, ad-hoc rules created and examples provided.
Step 6. Corpus annotation or labeling. This sixth step should be undertaken once
reliability has been optimized. As individual researchers, we start with the drawback of
not being able to count on interrater reliability, the strongest of the three methods. To
avoid fatigue, and guarantee maximum stability and consistency, we will work in short
sessions, of up to two hours per day, and, as Fuoli (2018: 19-20) advises, we will separate
the tasks into two: first identification and then the classification.
Step 7. Analyzing results. Fuoli (2018: 20) mentions specific bibliography for the
statistical analysis of the data obtained. In our frequency analysis, we will use the chi-
square test or x2 to test the hypothesis of this doctoral thesis, that is, whether or not there
is a statistically significant variation in the use of linguistic evaluation systems in the six
texts. It is a test widely used in social sciences and linguistics. This test measures the
discrepancy between observed and expected frequencies (Spiegel, 1993: 268), and it has
been applied in various quantitative analyzes with Appraisal Theory and for various
purposes, among others by Wang (2013), Cheng (2014), and Alba-Juez (2017).
In short, Fuoli has proposed seven methodological steps that require a systematic,
contextually dependent investigation that requires more time than other more informal
approaches. In return, by scrupulously following all of Fuoli's steps, we hope to optimize
the quality of our analysis and its reliability.
Finally, we observe that the basic procedural elements of the Fuoli proposal
(2018) and of the methodology that we developed fit the model Matter Cycle of corpus
linguístics (Pustejovsky et al, 2017: 23), although we emphasize that, given the
contextually dependent nature of interpersonal semantics, the annotation has been
manual. This type of annotation has led us to a learning process that would not have been
possible with automatic labeling.
184.108.40.206. Annotation manual
220.127.116.11.1. Annotation scheme
Regarding the scope of the project, we start by considering that Ramón y Cajal
can deal with different issues in the six texts that make up our corpus and that he can
consequently evaluate people and their actions, phenomena, and semiotic products,
follow different discursive strategies, adapted to each purpose and each intended
recipient, with different evaluative resources. Therefore, for the purposes of our research,
the analysis of the three semantic domains of the current Appraisal Theory and their
respective subdomains is essential. It seems very complex to us the question of how far
to delve into the categorical hierarchy so that we can reliably quantify its frequency of
use according to the different discursive genres; this is a question that perhaps we can
only answer after doing the reliability test. The following table contains the categories
analyzed, each of them identified by an alphanumeric abbreviation:
Parameters: semantic domains and subdomains
Social esteem A21
Social Sanction A22
Likewise, the analysis of the frequency of use of the interpersonal semantics of
our corpus would also be incomplete if it did not collect information on the following
three variables, which are arguments of the evaluative functional relationship by Alba-
Juez (2017): a) linguistic level (Ll); b) degree of (in)directness (Deg); and c) position in
the evaluative continuum between two poles (ContPos). It is complementary information
to the main one, made up of the domains and subdomains of interpersonal semantics.
18.104.22.168.2. Program settings of UAM Corpus Tool (version 3.3v)
The installation of the program on our computer, guided by an automatic wizard,
has been simple. You need to assign a name to the project. Next, we have added the six
texts of the corpus. It was the first time we used this program, and this simple operation
has only been possible after several failed attempts. The program supports only the txt
Resumimos la configuración del programa con el objeto de facilitar una futura intervención, propia o
format, without attributes. Thus, the program imposes a series of typographical changes,
such as the loss of bold and italics. We have inserted footnotes at the end of each text. We
remark that in our corpus, except for a transcription error, we have faithfully reproduced
the original texts. We have kept the possible typos and orthographic uses of the time,
ignoring them in our analysis. We also remark that the punctuation of text 1 is more
unstable and less neat than that of other later texts by the author. We do not know to what
extent these differences are due to the author or to the intervention of typographers,
editors, or proofreaders, but we believe that the author is always ultimately responsible
for the texts published under his supervision.
After the dump of the texts and their incorporation into the corpus, it has been
necessary to identify the language, Spanish, in each of the files. The program, whose
menu is written in English, includes a help section. The annotation scheme is set in the
Layers. Following the menu options, we assign the Layer the name Parámetros and we
successively choose Manual Annotation, Design your Own, Segments Within A
Document, No (Do you need a special layer?: The usual case), and again No (Should the
program automatically segment the text for you?). In the next window, we have chosen
the option Edit Scheme, where we have been renaming parameters, creating subdivisions
through the options Add System and adding and renaming Features. After some initial
trial and error, this process is very intuitive, fast, and flexible and allows for immediate
error correction. The program allows you to use accents in Features, but not upper case.
To include all rungs of the taxonomy and display them on screen, we had to assign the
value 6 to the field Depth.
For the design of the evaluative variables, we have finally chosen to use the Fields
function. This function allows us to assign open values to the categories, not specified in
the program, but that we will codify in our annotation rules. The Help menu provides
information on the procedure for configuring these fields. We have added to all categories
a mandatory field ContPos (for position along the evaluation continuum), and, to the
categories of ATTITUDe, two other fields called Deg y Ll (for degree of (in)directness and
linguistic level, respectively). These fields are added using the code “Require_Field Name
The manual annotation process is started from the Files menu, clicking, for each
of the corpus texts, on the name that we have previously assigned to the layer, in our case
Parámetros. A window will open with the name Parámetros analysis for. We recommend
careful reading of the Help menu in this window (Help/Show annotation help).
In short, the default design of the program allows one analysis unit to be identified
within another or nested within it, but not to assign two or more categories to the same
analyzed unit (when the clause is heteroglossic, the ATTITUDE category will be indicated
as an open value in a Field). Units are identified by dragging and dropping the cursor. A
lower window allows: a) assigning the chosen category, up to the level of hierarchical
depth that we decide; b) adding comments, such as contextual information; and c) adding
open values for the evaluative variables, in Fields (to visualize them, you must click on
Although the program allows you to edit the text (for example, to correct
transcription errors), we have found that doing this after having identified units causes a
visual displacement in markers, so we recommend never using this resource. On the other
hand, it does allow you to reconfigure it to rename the categories or add others, without
such changes affecting the work already done.
22.214.171.124.3. Test analysis of a systematic sample of the corpus
It began with the first analysis of a systematic selection of sentences from the six
texts, without including titles or epigraphs: sentences 1-3 of Text 1 (T1), 4-6 of T2, 7-9
of T3, 10-12 of T4, 13-15 of T5 and 16-18 of T6. We first identify the units of analysis
and then we categorize them. The following week, we analyze another selection of
sentences, moving on to the next three sentences in each text, that is, sentences 4-6 of T1,
7-9 of T2, and so on. Another week later, we analyze another selection, moving forward
again three sentences in each text, that is, sentences 7-9 of T1, 10-12 of T2, and so on.
We let two weeks pass, and then, in a single session, we did the second analysis of all
These early analyses, in addition to serving the purpose of first assessing
reliability, have also trained the analyst's capabilities, both in terms of accuracy and speed.
The analytical practice itself improves the ability to identify the units and the different
semantic categories, resolving theoretical doubts and, at the same time, raising new ones.
The results of this first test were not satisfactory nor do we consider them relevant, since
we had not yet stabilized the annotation rules nor had we acquired a minimum ease with
the program or with processes. At the end of this same annotation manual, we deal with
the reliability test-retest carried out at the end of the complete analysis.
126.96.36.199.4. Annotation rules
The rules that we have established and that we describe below are presented in
order of application priority. Our purpose is to create an annotation manual that allows
certain exhaustiveness, establishes validation controls, and takes as a reference,
systematically, the attitudinal markers exposed in the work of Martin and White (2008).
A. Rules for identifying units of analysis
In their “Annotation Manual”, Fuoli and Hommerberg (2015: appendix) only
mark as units of analysis those that carry interpersonal semantics of the domains and
subdomains under study. From their analysis they rule out, for example, the evaluations
invoked and the Appreciation subdomain. We consider that this decision synchronizes
the identification of the units and their categorization, so we will not take it as a model.
For our part, following the seven steps of Fuoli (2018), we will try to differentiate both
decisions. To mark the analyzed units, we have applied a criterion of certain
exhaustiveness that turns syntactic and lexical units into the object of our attention, as
potential carriers of evaluations, without ruling out any semantic subdomain, degree of
indirect realization, or semantic-pragmatic level. Considering, with Martin and White
(2008: 36, 56, 58, 68-69, 97-98, 134), that appraisal can be expressed with very diverse
resources, we are going to follow the general criterion of treating as units of analysis
clauses, adjectives and adverbs, interjections (a point we already defended in II. 1. 2. 2.
6), those nouns that constitute unusual similes or metaphors, or that explicitly introduce
into the discourse referents outside their field,
and also the groups or phrases that, in the
position of adjuncts, fulfill an attributive function, a technique that Ramón y Cajal
habitually follows. Verbs that constitute unusual metaphors will not be indicated as an
identification unit, since they are the nucleus of the clause itself. We note that Ramón y
This decision is motivated by the frequent and decisive use that Ramón y Cajal makes of this resource,
especially in his story “The Honesty Maker”. The images of a Byzantine Christ, a prophet, or the statue of
a Greco-Latin god are foreign to the narrated events, introducing elements into the imaginary that serve as
contrast and comparison. In some cases, they are presented as metaphors, with an attitudinal evaluative
function, but they also tend to produce a reminder of the irony that permeates the entire text, reinforcing it,
which is why they could be considered heteroglossic resources. We will return to this question in the
interpretation of data, in the specific qualitative analysis of Text 6.
Cajal's syntax is very complex; for this reason, in the analyzed units we will mark a scope
that will correspond to its minimum extension; but in the specific case of adverbial and
adjective groups we will indicate only their nuclei, with some exceptions that we will
indicate in these rules. It is a methodological solution, since once an attitudinal meaning
is expressed, it tends to spread through the discourse, to color it, as Halliday observes,
quoted by Martin and White (2008: 43). The extent of this prosodic effect may be the
subject of qualitative analysis, in our subsequent interpretation of data. In short, this
criterion of certain exhaustiveness requires a very laborious process of identification and
subsequent analysis, but one that we hope will allow us to retrieve relevant statistical
information for our research, without prejudging results.
Rule A1. As a general criterion, following Fuoli and Hommerberg (2015:
appendix), we mark for each unit of analysis the smallest possible scope, which will
correspond to that of the clause, the syntactic group, or the word, depending on the unit.
First, we mark main clauses, then subordinate or nested clauses, and then, successively,
the other elements.
To identify clauses we will follow the basic criteria of generative grammar,
applicable to any language. According to Carnie (2013: 211), a clause is made up of a
subject and a predicate. This is consistent with systemic-functional linguistics. Let us
remember that, for Halliday (2000: xiii), a clause is a unit in which three different types
of meaning are combined: textual, interpersonal, and ideational. The subject and the
predicate would be the functions of the clause as an exchange or interpersonal
Carnie (2013: 211-213) explains that a clause can be:
1) main clause
Example (Text 1):
UAM Corpus Tool marks the unit by underlining it and it allows multiple underlines. In the examples of
the identification rules of this “Annotation Manual”, we have chosen to use square brackets to present the
extension of clauses and visually distinguish them from other analyzable units.
[Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una nueva era
de investigaciones cuyo término no se vislumbra]
and 2) embedded clause
Example (Text 1):
[Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una nueva era
de investigaciones [cuyo término no se vislumbra]]
In turn, among embedded clauses we can distinguish: 2a) complement clauses,
that in generative terms are sisters of the verb; 2b) specifiers, which serve as the subject
of the main clause; and 2c) adjunct clauses, that appear in this position.
Rule A2. In complex clauses with relations of coordination, disjunction, causation,
and consequence (Martin and Rose, 2007: 115-140), we will not treat the complex clause
as the unit of analysis, but rather its constituent clauses.
Example (Text 1):
[Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una nueva
era de investigaciones [cuyo término no se vislumbra]], [pues si bien el método analítico
descubierto por este autor permite resolver algunos problemas de estructura], [ha servido
también para poner sobre el tapete cuestiones nuevas y dificilísimas].
We note that Ramón y Cajal uses the gerund a lot for the addition to a clause of
consecutive or subsequent actions, which we will consider as new units of analysis.
Adjectives and adverbs
Rule A3. Following the categorization of the RAE online dictionary,
we mark all
adjectives and adverbs as the nucleus of their respective phrases, whether they fulfill an
evaluative function or if their function is merely ideational, referential, deictic, or logical.
We mark as a single complete unit the adverbial phrases (“si bien”, “sobre todo”) and
those adjective and adverbial groups that are presented as options (“24 o más”).
Example (Text 1):
Estructura de los centros nerviosos de las aves.
We do not translate the analysis examples.
Real Academia Española (Royal Spanish Academy).
Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una nueva
era de investigaciones cuyo término no se vislumbra, pues si bien el método analítico
descubierto por este autor permite resolver algunos problemas de estructura, ha servido
también para poner sobre el tapete cuestiones nuevas y dificilísimas.
Rule A4. We mark as potential carriers of evaluative realizations those nouns and
noun groups or phrases that constitute explicit or unusual similes or metaphors in the
context, as well as grammatical metaphors of the evaluative-noun adjective type.
Example (Text 6):
El fabricante de honradez
Example (Text 3):
[…] la inapropiación de los pigmentos industriales […].
Rule A5. We mark as potential carriers of evaluative realizations these nouns:
aumentativos, diminutivos, despectivos and superlativos.
Example (Texto 1):
Esta fibra nace con frecuencia de la raíz de una ramita
Note: This example is affected by our ad hoc rule C1.
Rule A6. We mark all interjections as potential carriers of evaluative realizations.
Note: interjections have not been found in our corpus.
Rule A7. The epigraphs and adjuncts constituted by a single noun phrase are
marked as a unit of analysis.
[Estructura de los centros nerviosos de las aves]
After applying all the identification rules, the initial fragment of Text 1 would
look like this:
In this aspect, the morphological resources of Spanish are very different from English (augmentatives,
diminutives, derogatory and superlative nouns).
Little branch in Spanish.
[Estructura de los centros nerviosos de las aves]
[Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una nueva
era de investigaciones [cuyo término no se vislumbra]], [pues si bien el método analítico
descubierto por este autor permite resolver algunos problemas de estructura], [ha servido
también para poner sobre el tapete cuestiones nuevas y dificilísimas].
Rule A8. In the course of the analysis, specific cases have been presented for
whose identification we have had to make ad hoc decisions. A8a: we consider direct
quotes as a single unit of analysis that constitutes the clauses to which they belong; we
will categorize them as heteroglossic resources and they will absorb the possible
attitudinal evaluation of the quote.
Rule A9. Finally, we have created the fields Id (identification of the grammatical
category of the unit), Deg (identification of the degree of (in)directness), Ll (identification
of linguistic level), and ContPos (position in the continuum), according to the following
variables (this coding will allow us to count the frequency of the evaluative function
performed by each type of unit in each text and the corpus as a whole):
Id: clause (C), embedded clause (CS), noun o noun phrase (SN), adjective (Adj)
and adverb (Adv). For the purposes of this typology, noun phrases that function as
epigraphs will be considered as main clauses (C), thus reserving the type noun or noun
phrase for the constituent elements of a clause.
Deg: direct or inscribed evaluative realization (Dir) and indirect or invoked
Ll: realization levels: lexical (Lex), morphological (Mor), semantic-pragmatic
(SeP), and, in the absence of the previous three, also the syntactic one (Sin).
ContPos: we will use codes 1, 0, -1, and # depending on whether it is a positive,
neutral, negative, or potentially ambivalent value.
Example (Text 1):
[Las investigaciones de Golgi sobre la estructura de los centros nerviosos han abierto (1) una
nueva (1) era de investigaciones [cuyo término no (-1) se vislumbra](1)] (1)
We do not encode +1, but simply 1, because UAM CorpusTool statistics generator does not recognize it
and, when creating searches, it fails. This peculiarity has invited us to review the entire analysis.
ContPos: 1. Since the position in the continuum of one evaluation and another are
made with different resources, we think that in the current state of research it is
problematic to quantify the intensity in scales and that assigning numbers can be
misleading. The signs, then, simply indicate whether the evaluation of the unit is positive,
neutral, negative, or ambivalent.
Another related phenomenon, and one that is especially frequent in the story (Text
6), is what we could call the instability of the position in the continuum when irony comes
into play, since the evaluation can be considered as negative or positive or as occupying
two different positions on the continuum. This instability, which we provisionally
indicate with the # code to mark irony, will later be coded as positive or negative
depending on the ultimate meaning agreed upon by interlocutors. We will return to this
issue in the interpretation of data, in the specific qualitative analysis.
B. Categorization Rules
The experience gained in the first attempts at analysis has led us to decide to
categorize the units into two distinct phases. In the first phase, we have categorized the
interpersonal semantics of clauses. In the second phase, we have categorized the
interpersonal semantics of the minor constituent units of each clause. This second phase
also allows us four things:
a) Carefully observe the attitudinal markers, the smallest appraisal details,
and lexical resources through which appraisal is expressed.
b) Study the interaction of the evaluative resources of the clause and its
constituent units, when one and the other belong to different domains or
c) Study and review the position in the continuum of the main clause, taking
into account the position in the continuum of the evaluative resources of
d) In short, correct errors and categorization oversights in the first phase. It
is a control function.
First phase: categorization of clauses
Rule B1. We first categorize clauses according to the evaluative taxonomy of
Martin and White (2008) in all texts; then we provisionally annotate variables Deg, Ll,
and Id. Any contextual observation, or about the decision process can be annotated in the
field Comment, especially troublesome cases.
Rule B2. As a general rule, when deciding which semantic domain is the most
relevant, we follow the following order of preference: ENGAGEMENT, ATTITUDe, and
GRADUATION. When the first domain is ENGAGEMENT, ATTITUDe will be encoded with
its alphanumeric identifier in the field Secondary subdomain (created only for the domain
Example (Text 3):
¿A qué se deben esa crudeza de tonos y pobreza de colorido peculiares de la tricromía
tipográfica? [C21, A32]
Rule B3. We consider as a marker for domain ENGAGEMENT, in addition to the
direct projection of sources and the use of negations, the presence of modal resources.
B3a) The expletive no,
whose value is of intensification of the evaluation of the
unit of which it is a constituent.
Example (Text 5):
[…] a menos que la esposa no aporte la compensación mental indispensable.
B3b) Comparative uses of negation.
Example (Text 1):
[…] aunque no tanto como las montadas en soluciones balsámicas clorofórmicas.
Rule B4. In general and for methodological purposes, we consider the most
relevant realization of the domain ATTITUDe the direct degree of realization.
Rule B5. In the absence of direct realization, we consider as markers of a possible
indirect degree of realization of ATTITUDe in the clause (Martin and White, 2008: 61-68)
the following, in order of priority: ENGAGEMENT resources, unusual metaphors, and
This is a special Spanish case: el no expletive.
lexicon, GRADUATION resources as the intensification of qualities and the quantification
of abstract entities and metaphors (Martin and White: 148-149, 152) (Ll: Lex), marked
syntax (Ll: Sin) and semantic-pragmatic selection (Ll: SeP). It is important to note that
Martin and White (2008: 148-149, 152) observe that intensification of processes and the
non-metaphorical quantification of number, mass, and extension can be cases of not
Rule B6. To identify a subdomain within ATTITUDe, we stick to the target of
evaluation, according to Thompson, quoted by Fuoli (2018: 10). We focus our attention
on the nucleus of the clause, the verb, to decide which subdomain is the most relevant,
because it is very common that in the same clause direct and indirect evaluations are
expressed, with diverse variables, or that its constituents include evaluations of other
semantic domains and subdomains.
B6a) In cases in which we cannot decide on the most relevant indirect realization
or when recursion of evaluation is multiplied, we will follow this order of priority: a)
Affect, b) Judgement y c) Appreciation.
Example (Text 1):
(a) [Las investigaciones de Golgi sobre la textura de los centros nerviosos han abierto una
nueva era de investigaciones cuyo término no se vislumbra], (b) [pues si bien el método
analítico descubierto por este autor permite resolver algunos problemas de estructura], (c)
[ha servido también para poner sobre el tapete cuestiones nuevas y dificilísimas].
(a), Judgement/ social esteem
(b), Appreciation/ valuation
(c), Appreciation/ valuation
In (a) the target of the evaluation is Golgi investigations, meanwhile in b) and c)
the target is the analytical method, that we consider to be a semiotic product, following
our ad hoc rule C2.
Rule B7. In general, intensifications of processes will not be considered attitudinal
markers, since, according to Martin and White (2008: 148), these resources are not
typically attitudinal. This rule has been established to respond primarily to the
categorization challenges posed by Text 3, or Fotografía de los colores (Ramón y Cajal,
1912). Here monoglossic clauses abound; this monoglossic clauses, as technical
instructions, do not seem to express attitude. However, there are also isolated cases in
Example (Text 1):
[Lavamos reiteradamente los cortes en alcohol].
We will return to this issue and these statements in the interpretation of data, in
the specific analyzes.
Rule B8. The presence of the heteroglossic resource of irony, which permeates the
entire Text 6, will be categorized in such a way in the corpus tool when it is an irony
carried out through the constituents and except for the title of the story itself. This rule
has been created ad hoc due to the challenges posed by the story, colored by an irony
placed at the service of parody, and will allow us to jointly identify and categorize other
resources of ENGAGEMENT and ATTITUDe, that would otherwise be obscured by the
Examples (Text 6):
(a) El fabricante de honradez
(b) Mas, antes de referir las hazañas del prestigioso personaje, debemos presentarlo a nuestros
(a) ENGAGEMENT/ Expansion/ proposition.
(b) ENGAGEMENT/ Expansion/ proposition.
Second phase: categorization of constituents and control
Rule B9. We categorize the constituent units of each clause correlatively according
to Martin and White’s taxonomy (2008) and rules B1-B8 and we assign, if necessary, the
variables Deg, Ll, ContPos, and Id. The non-evaluative constituent units will be
categorized as ideational/deictic or as heteroglossia markers, categories that we have
foreseen to indicate expressions that fulfill these functions.
Rule B10. We study the interrelationship of constituent evaluations with the main
evaluation of the clause and we verify and validate the latter.
Specific rules of some texts
Some texts have raised specific questions and have demanded new ad hoc rules,
which we have subsequently applied to the rest of the corpus.
Rule C1. This rule has been created ad hoc to address challenges posed by Text 1.
When diminutive nouns and other unusual lexicon respond to the need to establish a
categorical differentiation, we will not consider them attitudinal markers.
Esta fibra nace con frecuencia de la raíz de una ramita protoplasmática.
Observation: in this context, the potential metaphors of the terms fibra and raíz
are not unusual, so we have excluded them from our evaluative analysis. The term ramita,
which meets a categorical differentiation in Text 1, the scientific article, was analyzed as
a potentially evaluative unit according to our Rule A5, applicable to the entire corpus, but
later, in compliance with Rule C1, we have classified it as a function term merely
Rule C2. This complex and problematic rule, which should be understood as a
reviewable proposal, has been developed to meet the challenges posed, above all, by Text
2. How do categorize attitudinal statements that target issues related to epistemology? In
scientific texts or those that deal with science, we have been confronted with various
specific evaluation targets, for which we will follow the guidelines that we add in
a) Truths or expressions or formulations of truth, which we will consider semiotic
b) Methodologies, techniques, procedures, and instructions, in different degrees
of codification, which we will consider semiotic products (Appreciation).
c) Reflections, arguments, ideas and ideals, and models of research action, more
or less vague and schematic, but expressed, and that we will consider semiotic products
Example (Text 4):
Literally, fiber and root.
Tal fue la sencillísima idea inspiradora de mis reiterados ensayos del método argéntico en los
embriones de ave y de mamífero.
d) Specific research actions (Judgement).
Example (Text 4):
Escogiendo bien la fase evolutiva, o más claro, aplicando el método antes de la aparición de
la vaina medular de los axones.
e) Scientific and humanistic disciplines, which we will consider sets of verbal
f) Adaptation to the scientific method, which we will consider a scheme or set of
prescriptions for action (Judgement / social esteem/ normality).
We have not found or have not been able to adequately interpret in Martin and
White (2008) a criterion around the categorization of these evaluations, a circumstance
that may be due to the fact that the authors have developed their taxonomy by analyzing
other types of texts. As can be seen, in our analysis we will distinguish, on the one hand,
the categorization of attitudinal evaluations concerning the application of the scientific
method (Judgement/ social esteem/ normality), and, on the other hand, regarding specific
research methodologies (Appreciation/ valuation). We think that it is a debatable
Rule C3. The evaluation of the aesthetics of characters (Text 6), although literary
and therefore semiotic products (Text 6), will be considered as a Judgement about the
social esteem of such characters or, in some cases, as Affect, when the description arouses
strong feelings or emotions. We, therefore, consider characters, that are elements of the
fictional imaginary, like real people, and its aesthetic evaluation will be considered a
physiognomic criterion resource. As can be seen in the following example, the aesthetic
description of this character alludes to his intellectual abilities and it implicitly entails a
Poseía aventajada estatura, cabeza grande y melenuda, donde se alojaban pilas nerviosas de
gran capacidad […]
188.8.131.52. Reliability test
Since variables in our analysis are nominal, we will perform a chi-square or test
x2. It is the same test that we will use to test the main hypothesis of this doctoral thesis,
as we have anticipated when dealing with the seventh step of Fuoli, in III.184.108.40.206. Our
reliability test, unlike that of Fuoli and Hommerberg (2015: 332), will not be an interrater
one, but a reliability test-retest, which will simply measure whether our individual
analysis can be accepted as a hypothesis in the time interval between the complete
analysis of the corpus and a sample analyzed fifteen days later. For the test-retest sample,
we have randomly selected two contiguous paragraphs from the first text and then moved
forward one ordinal to select the paragraphs from the following texts, i.e.: 3rd y 4th of
Text 1; 4th and 5th of Text 2; 5th and 6th of Text 3; 6th and 7th of Text 4; 7th and 8th of
Text 5; and 8th and 9th of Text 6. This sample consists of 996 words, 7.99% of the 12,452
words that make up the corpus. For calculation, we have used the data analysis and
statistics program Xlstat, Addinsoft.
In the first attempt, we have descended to the maximum hierarchical depth in the
semantic domains. However, since some subdomains provided very few cases, we moved
up the semantic hierarchy again to meet chi-square requirements.
It is obvious that
problematic cases increase when descending in the semantic hierarchy and, conversely,
that when ascending in it, the correlation between the data of the analysis and those of the
This program integrates with Microsoft Excel.
For the ideal calculation of chi-square, the marginal sum of columns or rows cannot be null nor can there
be theoretical frequencies lower than 5.
The following table provides the theoretical frequencies that have been calculated
The result has been x2 (9, N = 579) = 1.141, p = 0,999. This result is not
significant at a statistical significance level of 0.05, usual in linguistics, so the correlation
hypothesis is accepted.
In conclusion, we think that the establishment of rules and differentiating, on the
one hand, the identification of analyzable units and, on the other hand, the classification
of the clauses and their constituents, has allowed us to acceptably systematize our results
concerning the first levels of semantic domains. We have applied a double process of
control and validation, where the classification of clauses is very dependent on that of
their constituents and focuses on markers. However, systematization becomes very
complicated when descending in the semantic hierarchy, since problematicity increases.
We insist, once again, that the true touchstone of this methodology would be a team
2.2.2. Critical Discourse Analysis Method
Like when preparing our methodology for the analysis of evaluative resources in
220.127.116.11., we can now also ask ourselves, following Rodríguez Lifante (2016: 29), about
the method, that designates the entire research process, the what, or object of
investigation, the design, which establishes a specific procedure for obtaining data, and
the technique, that designates the instruments with which the data are collected. We are
not starting from scratch, because our attempt to approach contextual models can be
considered an extension of the analysis of evaluative resources. We will try to ensure that
the methodology followed, based on constituents, is also consistent in the development
of the interface that we propose. Martin and White's Appraisal Theory (2008), with its
focus on the text, largely disregards the pre-realization phase in its current state. Let us
remember that Alba-Juez and Thompson (2014) distinguish three evaluative phases, pre-
realization phase, textual phase, and metaevaluative phase. To investigate the pre-
realization phase we need a theory that allows us to peek into the mind of the speaker or
writer and that allows us to explain their realizations, or at least try to do so. This is
precisely the gap that we expect contextual models to fill, mindsets of contextual models,
that are composed of sociocognitive elements.
Contextual models, therefore, are now the what. But it is a schematic what,
composed of numerous ingredients or mental models, in which, following Van Dijk
(2017: 122-123), we can distinguish in principle three basic categories: the place, the
participants, and the actions and communicative events. However, these are not closed
categories; its constituents, in other Van Dijk titles, will increase and nuance: social
actors, topics or issues addressed, purposes, strategies, social values, social
representations, social groups (professional, organizational, institutional groups),
epistemic communities, ideologies… A what, furthermore, that is not necessarily textual,
in the sense that they may not be indexed in the text, since, as Van Dijk (2017: 43)
explains, they may be implicit and presupposed, and they may be made explicit just under
In terms of design and technique, they are conditioned by the methodology of
analysis of evaluative resources: these already indicate the issues, people, and things
valued, which are the evaluation objectives. However, although our analysis of evaluative
resources consists basically of the categorization of linguistic data, obtained in empirical
observation, we must now qualitatively jump to the assumptions and implicit purposes
that stimulated the production of texts, social values, and ideologies that monitored the
verbal production. This reconstruction of contextual models, therefore, has a tentative
character and its proposals must be taken as mere conjectures or hypotheses, whose
acceptance will depend on their explanatory capacity of evaluative resources and, in a
more general plane, of discourse resources. In addition, given the schematic nature of the
elements that constitute the contextual models, we will present them in a similar
schematic way: as tables and as conceptual maps or visual synthesis. For the realization
of the tables, we will rely on the corresponding function of the word processor that we
use, and for the drawing of the conceptual maps, on the simple, flexible, and widely
available PowerPoint program. In this regard, we note that our first choice was the
CmapTools program, version 6.04, from the Institute for Human & Machine Cognition
(IHMC), University of Florida. However, we have encountered some difficulties in the
nesting function, for the graphical representation of human groups and subgroups: for
example, the fact that the group of reticularist histologists is in turn included within the
broader group of the international community of histologists, of which Ramón y Cajal
himself is also a member, a representation that alone requires two nesting levels.
We think that, within the contextual models, the very notion of social value is of
the utmost relevance for Appraisal Theory. For Martin and White (2008: 45), as we saw
when dealing with the theoretical framework, the evaluations of Judgement and
Appreciation (or, in general, evaluation of people and objects, respectively), can be
interpreted as institutionalized feelings. When judging or appraising, our evaluation is
based on values shared by the community. Van Dijk's sociocognitive approach, whose
work does not specifically deal with values, although they permeate it completely,
provides us with a more precise definition, as shared mental objects of social cognition,
which would form the basis of all evaluation processes, opinions, and ideologies (1999:
101). He further explains that while some values, such as truth and happiness, etc., seem
to be generally shared as criteria and goals, other values may be different from one culture
to another and vary in their order of importance (1999: 101). Thus, values have a
sociocultural status and they are pillars of the moral order of societies (1999: 102). Van
Dijk, consistent with this statement, rejects the notion of individual values (1999: 102).
Both definitions of social value, either as institutionalized feelings or as shared mental
objects of social cognition, can serve as a reference, but in our methodology, as we will
see, we will not start from the abstract to reach the concrete, but rather we will follow the
opposite path, traveling from the concrete, that is, the evaluative realizations, to the
abstract, in coherence with the methodology based on constituents that we have been
developing. Simply, in the most intuitive way and based on our own introspection, but
reusing the terms of Critical Discourse Analysis, we are going to consider a social value
as a mental scheme that monitors evaluations; this scheme can consist of a model (for
example, the basic characteristics of classical painting, which Ramón y Cajal defends
(1967: 253)), a model with which we compare things and semiotic products or a person
and his actions, and also instructions for evaluation, instructions to be carried out in texts.
What do we understand, for example, by Politeness? In pragmatic linguistics, outside the
theoretical framework of this doctoral research, several authors have studied Politeness
in-depth, among which Brown and Levinson (1987) stand out for the studies they have
promoted and the academic debate they have inspired. For Green, quoted by Alba-Juez
(2009: 89), the term Politeness encompasses various strategies to maintain or modify
interpersonal relationships. Dimitrova-Galacz (2002: 1-2) highlights precisely the lack of
a consensus on the definition of the term and the heterogeneity of the linguistic currents
that study the phenomenon, in addition to verifying the absence of a universal equivalence
between cultures. Our claim cannot be, therefore, to reach a definition of social value
Politeness. In our journey through the historical background of Appraisal Theory, we
already come across the old controversy between platonism and protagorism (II. 18.104.22.168.).
Returning to that debate, we will not try to define, for example, the social values of
Honesty or Politeness
in the platonist way, but, following the pragmatic use of
protagorism, which anticipates some aspects of current pragmatic linguistics, we will
simply observe that, in the scientific article by Ramón Cajal (1888), the Politeness of a
scientist towards another scientist (for example, Ramón y Cajal towards Camillo Golgi),
is carried out through the attribution of sources and an indirect degree of realization; in
other words, the social value Politeness in this particular context could be formulated as
follows: “The author of a scientific article must cite the works consulted and acknowledge
the authorship of the methodology used; its author will be valued or evaluated indirectly
and negative positions in the continuum will be avoided or modalized”. We cannot
assume that these social values are neither universal nor static, so they can be modified
from one text to another and according to other variables. Politeness, in the
epistemological essay by Ramón y Cajal (Text 2), seems indeed to be different: when the
quoted author is not a contemporary of Ramón y Cajal himself, we will find cases of very
negative attitudinal evaluations of the work or ideas of an author; but in the same text and
if the author is a contemporary one, the critical evaluations are softened. At the opposite
En adelante, los valores sociales, en cuanto esquemas mentales (no en cuanto su realización textual) los
escribiremos en redonda y con la inicial en mayúscula.
extreme, in the fictional text, the characters are valued directly and with ambivalent
positions in the continuum; some will be caricatured and vilified: from Politeness we have
passed, we could say, to an anti-Politeness, to hyperbole, to a parody that constitutes a
negative Politeness that threatens their face, in terms of Brown and Levinson (1987), and,
in Culpeper's terms, cited by Alba-Juez and Mackenzie (2015: 159), an example of
"Sarcasm or mock politeness".
Continuing with this reflection, we must allude to the complex issue of
epistemological social values. In the scientific article, we highlight the priority given to
empirical observations and direct inferences, not based on testimonies; as we have already
pointed out in other moments of this thesis, histologists reproduced the experiments to
confirm the observations of their colleagues, and Ramón y Cajal followed the same
procedure; however, the question of these supposed epistemological values leads us to a
terrain foreign to the theoretical framework of this thesis.
The possible attitudinal social
values involved in the scientific activity are going to be sheltered from a very broad
scheme, such as the scientific method itself, which we can understand in the context of
this thesis as a model of models and as a set of sets of instructions, that is, as a set of
social values used in science and normalized by research practice. It is true that, in the
observations of Ramón y Cajal (1888) through the microscope, we intuit a social value
associated with the aesthetic appreciation of nerve cells. This social value, this mental
scheme, could perhaps be called Truthfulness and, in the scientific texts of Ramón y Cajal,
it could be formulated as follows: “An attempt should be made to describe the concept of
the observations, without avoiding the expression of subjective difficulty of the observer;
in the absence of precise arithmetic and geometric form recognizable in the observed
things, it is acceptable to use similes and approximate quantities and to intensify
qualities”. This hypothetical social value is, we think, perfectly compatible with the
empirical theory of knowledge; according to Russell ( 1992: 511), “inadequacies
as we have seemed to find in empiricism have been found by strict adherence to a doctrine
by which empiricist philosophy has been inspired: that all human knowledge is uncertain,
inexact, and partial. To this doctrine, we have not found any limitation whatever”.
For this question and others related to the axiology of science, epistemological values, and the
mechanisms of reason, we refer among others to Rescher (1999), Echeverría (2002), and Mercier and
Although for reasons of textual organization, we now present the parameters of
the analysis carried out, before proceeding to the analysis itself, the truth is that the
definitive writing of this epigraph has been simultaneous, since we have been writing it
as, advancing in the attempt to reconstruct the contextual models of the six analyzed texts,
we were debugging it. The next "Annotation Manual for Contextual Models” is a
proposal, it configures a qualitative methodology, without a pretense of exhaustiveness,
it has a synthesis character, and is written as such.
22.214.171.124. Annotation Manual for Contextual Models: Parameters
Considering these parameters as an extension of our annotation manual for
evaluative resources, we organize them as rules, with the key MC, by “modelo
contextual” (Contextual Model), and identified with an ordinal. Some rules, due to their
hypothetical objective and their problematic nature, will be formulated as questions that
must be resolved.
Rule MC1. We identify the author of the text and the addressee, the scenario/place
of the communicative interaction (which does not necessarily coincide with the scenario
of social representation), and the support/format of the communicative event.
Example of Text 3, techno-scientific manual of photography. The author is Ramón
y Cajal; recipients are photography lovers and professionals; the place for communicative
interaction is a photographic laboratory; the support/format of the communicative event
is a book.
Rule MC2. We identify the purpose of the text.
Example of Text 1, the scientific research article. The purpose is to reject the
reticular hypothesis and propose a new avenue of research.
Rule MC3. We identify the people, characters, and their actions valued directly or
indirectly in the text, and indicate the degree of realization, their position in the
continuum, and, considering the interaction as a recognized core of Critical Discourse
Analysis itself, whether or not it is an author's potential interlocutor.
Example of Text 6, the fictional story. Doctor Mirahonda (see IV. 3. 6. 2. 2.).
Rule MC4. We identify semiotic objects and products, concrete or abstract, valued
directly or indirectly in the text. If they are grouped into categories in the text itself, we
identify them as such. We indicate the degree of realization and its position in the
Example of text 4, the autobiography. Histological maturation and staining or
dying methods (see IV. 3. 4. 2. 3.).
Rule MC5. We identify those people, actions, objects, or semiotic products that,
indexed in the text itself, may have served as an evaluation model for the entities detected
in the rules and, by way of conjecture, their position in the continuum according to the
evaluation that they propitiated. This rule allows us to include the source domains of
comparable metaphors and other possible metaphors and sources projected in the text and
not classified in the previous sections.
Example of Text 5, the series of aphorisms. A flower (see IV.126.96.36.199.).
Rules of reconstruction of social values
Rule MC6. In an attempt, we will consider social values as models or instructions
for evaluation. Although we will give them a name, we think that this is not as important
as the model that configures it or the scheme of evaluation instructions that constitute
each one. The attitudinal grammatical metaphors detected in the text can serve in the
Example of Text 3, techno-scientific manual of photography. Affordability.
Rule MC7. The epistemic values involved in scientific research will be grouped
under a complex social value that we will call Scientific Method (see IV. 3. 1. 2. 4.).
Rule MC8. Social values will be visually represented in blue and above (see IV.
3. 1. 2. 5)
Reconstruction rules of social representations
Rule MC9. We classify interlocutors, people, or characters and their actions in
groups, according to the categories indexed in the text and, failing that, according to our
knowledge of the cultural context, provided that this knowledge does not conflict with
the categories or classifications of the text itself.
Example of Text 2, epistemological essay on scientific research. Philosophers.
Rule MC10. We identify the place or scenario of the social representation of the
event model, which is not necessarily the place of the social representation of the
Example of Text 6, fictional story. Villabronca.
Rule MC11. We represent these schemes and groups in drawings: in the center
and in black on white, the evaluator (on the left) and his interlocutors (on the right); below
and in green, the physical and the least important, dominated or not preeminent; elements
outside the scope, or the imaginary, in warm colors, preferably below and to the left;
intellectual objects, in cool colors and on the sides.
Identification rules for ideologies
Rule MC12. Abstract ideas or systems of ideas indexed in the text that can
maintain a cause-effect relationship on a set of evaluative realizations and are associated
with a social group will be considered candidates for the status of ideology. They will be
represented in blue above the social values, hypothetically monitoring the discourse.
Rule MC13. We will indicate, for each text of the corpus, the social identity of the
author or his relevant social group in discourse, the most relevant social values involved,
and the social groups of the recipients. Ideology, due to its group character, must be
defined by the social identity or social group of the relevant author in each discourse and
must be consistent with the communicative purpose, social representations, and other
constituents of contextual models.
Rule MC14. We will study for each contextual model a relevant social value in
discourse and its possible function concerning the groups indexed in the social
representations, both in the social representation of communicative interaction and in the
social representation of the event model.
Rule MC15. We will reason about the possible inequality in terms of domination
or abuse of power of the group of the writer or the speaker over other social groups. We
will highlight processes of normalization, as an uncritical reproduction of a social
representation of inequality, and naturalization, as an explicit legitimation of such
representation through a supposed self-serving truth linked to the interest of the group.
In the methodological interface that we have just developed, we have started from
evaluative realizations, in the textual phase, to approach contextual models, in the pre-
realization phase. With the data obtained in the quantitative analysis of frequencies, we
will test the main hypothesis of this thesis: "The linguistic evaluation systems used by
Santiago Ramón y Cajal vary according to discursive genres". After exposing the
contextual models, we will make a comparison and test the secondary hypothesis or sub
hypothesis: "Some evaluative realizations can only be explained by means of the notion
of contextual model". Through the comparison between all contextual models, we will
try to reveal which components of these contextual models have a cause-effect
relationship with the evaluative realizations, and, where appropriate, we will evaluate if
they are necessary or sufficient factors to explain them. Finally, in light of the data
obtained, we will try to categorize the basic constituents of contextual models in the types
of context established by Fetzer, cited by Alba-Juez and Mackenzie (2015: 6-8):
linguistic, social, cultural, and cognitive, to which these same authors add a fifth context,
the attitudinal-emotional context.