|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
SPANISH EVALUATIVE MORPHOLOGY:
PRAGMATIC, SOCIOLINGUISTIC, AND SEMANTIC ISSUES
VICTOR MOISES PRIETO
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Victor Moises Prieto
This document is dedicated to God; my wife, Monica; my son, Victor Emanuel; my
mother, Silvia; my father, Juvenal; my siblings: Raxil, Isaac, Daniel, Ruth, David, Javier,
Judith, Martin, Nali, Antoine, and Pipo; the linguistics and Spanish departments at the
University of Florida; Baptists in all the world; Hispanics in the USA; my home country,
Venezuela; and my second home countries: Colombia and the USA. .
I thank God for His help and company during all these years of fights and victories,
sadness and happiness, and defeats and triumphs. I feel forever grateful to my wife,
Monica; my son, Victor Emanuel; my parents; and my siblings.
I would like to express my special gratitude to my supervisory committee chair and
advisor (Dr. Diana Boxer) for all her support, understanding, teachings, and help during
these years of graduate school. At the same time, I thank my other supervisory committee
members; Dr. D. Gary Miller, Dr. David A. Pharies, and Dr. Shifra Armon. Each one of
them immensely contributed to my training, learning, and research during this time of
reading, preparation, investigation, and writing. Other professors who are not members of
my committee also contributed to this final product at some point: Dr. Caroline Wiltshire,
Dr. Joachim Camps, Dr. Theresa Antes, Dr. Eric Potsdam, Dr. M.J. Hardman, Dr. Ratree
Wayland, and Dr. Edith Khan. I also express my gratitude to Steve Flocks, my editor; and
to my classmates. In general, I feel appreciation for the Program of Linguistics at the
University of Florida for its office staff, Joan and Jolee; and for its entire faculty. Finally,
I would like to express words of gratitude to the Baptist Hispanic Church of Gainesville,
FL. since they constituted one of my main sources of data and information. I will never
forget these people, this department, this university, and this church.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
L IST O F TA B LE S ......... ............ .............................. .. .. ....... .............. viii
LIST OF FIGURES ......... ......................... ...... ........ ............ ix
A B STR A C T ................................................. ..................................... .. x
1 IN TR OD U CTION ............................................... .. ......................... ..
R research Q uestions........... .................................................................... ........ .. .. ...
L literature R review ................................................................ 3
Pragm atics ............................................................ 4
Pragm atic Theoretical Tenets .................................................................. ... 6
Sociolinguistic Perspectives ............................................ ........... ............... 8
Evaluative M orphology ......................................................... .............. 10
M orphology in general ....................................................... .... ........... 10
Evaluativeness in morphology .......... .................................... ..................11
M methodology ..................................... ................................ ........... 22
E thnography .......................................................................22
D ata and Participants ........................................................................... 23
D ata A n aly sis................................................ ................ 2 6
M orphopragm atics .............................. ...... ......... ............28
Conclusion ................................. .......................... ..... ..... ........ 30
2 EVALUATIVE MORPHOLOGY: FUNDAMENTALS ................ ......................32
E v alu ativ eness in G eneral................................................................ .....................32
T he M orphology of E valuatives ................................................................................34
Evaluativeness Cross-Linguistically........................................................................ 40
Non(morphologically) Evaluative Languages: Evaluative Syntax ...................40
Morphologically Evaluative Languages: Two Types.......................................43
Linguistic Postulates and Evaluative Morphology............................................47
Origins of Spanish EVALs ................................................................... 49
PIE and Latin antecedents ............................................................................50
Other non-DIM Latin suffixes................................ ........................ 52
D IM s form al gram m ar ....................................................... ..... .......... 52
D IM 's original sen ses..................................................................................54
Changes in meaning and productivity .......................................................55
A u g m entativ es ........................................................... ........ ........ ......... 5 8
-az o .............. ................. ............................................................... 5 8
-a l .........................................................6 0
o n ............................................................................................................ 6 1
-o te :..................................................................................6 2
Superlative ............. ..... ......... .... ...............63
C on clu sion .............. ................. ................................................................6 5
N o te s .............. ..... ............ ................. ......................................................6 6
3 SE M A N T IC IS SU E S ............................................................................................. 67
Sem antic vs. Pragmatic Polysemy ................................. ...................... ............. 67
Cognitive Semantic M odel: Radial Categories.............................. ............... 70
D im in u tiv e s............................................................................................................ 7 5
C ore Sense of D IM s ................................................ ............... 75
C h gaining L ink s ..............................................................86
A u gm entativ es ................................................................94
C o re S e n se ................................................................... ................................9 4
C h ain in g L in k s ................................................................9 5
Su p erlativ es ................................................................................................... 9 8
Conclusion ........................ ....................... .... 100
4 PRAGMATIC FUNCTIONS ..................................................102
D im inutives............................. .................. ...................... 104
Semantic/Neutral Uses of the Diminutive ....... ... .... .............. 106
Pragm atic U ses of D IM s ................................. ........ ............................110
The affection function ...... ..............................110
The derogation function ......................................... 122
The attenuation function or polite DIM ........................................... 126
A u g m en tativ e s ..................................................................................................... 13 3
Sem antics-D riven U ses ............................................... ......... 135
Pragmatics-Driven Uses .......... ................................136
The intensification function of AUGs ....................................................... 136
The attenuation function of A U G s .............................................................139
S u p e rlativ e s .......................................................................................................... 14 2
Conclusion ......................................... 145
5 SOCIOLINGUISTIC ANALYSIS ........................................ ... ........ 147
G group M parking ...............................................................149
T h e [C h ild ] D IM ........................................................................................... 14 9
The G endered D IM ....................................................................153
The Regional EVALs .............................................. ...............161
A Marginal Group-Marking Effect: Social Class............................................167
Societal Context/Speech Situation M parking ...........................................................168
The Inform al EV AL s.............................................. ................ ............... 169
DIMs and AUGs as informal context markers................ ....................169
The formal vs. the informal SUPERL...... ...........................................171
Familiar/intimate vs. non-familiar encounters .......................................172
Societal Power Structure and EVALs.................................................... 173
Conclusion ................................... ................................. ......... 179
6 CON CLU SION S ................................. ... .. .......... .. .............181
G en eral F in din g s............ .... ................................................................. ....... ........... .. 186
Implications ............................. ..........................192
Im plications for Theoretical Linguistics ................................. ............... 192
Implications for Applied Linguistics......................... ....................... 193
Lim stations and Further Research..................................... ........................ ......... 196
C o n clu sio n s.................................................... ................ 19 8
N o te s .............................................................................................. 19 9
A Q U E ST IO N N A IR E 1 ...................................................................... ...................200
B Q U E ST IO N N A IR E 2 ...................................................................... ...................20 1
C OTHER EVAL SPANISH SUFFIXES ............................. ..................... 202
D D A T A .................................................................................2 0 3
REFERENCES .................. ..... ......... .... .. ....... ........ 256
B IO G R A PH IC A L SK E T C H ........................................ ............................................270
LIST OF TABLES
1. Frequency of Spanish EV ALs ........ ...... ....................... ............... ... ............17
2 F requency of u se for D IM s ........................................................................... .... 133
3: Frequency of uses for AU G s ........... .......... ........ .. ................ ............... 136
LIST OF FIGURES
1. Frequency of Spanish EVALs ...................................... ........................... 18
2 T ree diagram of E V A L s......................................................................... .................. 38
3. New proposal for DIM's radial categories....................................... ..................... 86
4 R adial categories for A U G s ................................................................ .....................95
5. Semantic-pragmatic functions of Spanish DIMs................................................... 106
6. F requency of u ses for D IM s ............................ ................................. .....................133
7. Frequency of uses for AU G s ...................................................................... 136
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
SPANISH EVALUATIVE MORPHOLOGY:
PRAGMATIC, SOCIOLINGUISTIC, AND SEMANTIC ISSUES
Victor Moises Prieto
Chair: Diana Boxer
Spanish evaluative morphology (diminutive and augmentative suffixes,
prototypically; superlatives and pejoratives, marginally) has been the focus of many
studies in linguistics. However, important practical and theoretical aspects have not been
formally considered accurately or in entirety. The pragmatics and the sociolinguistics of
such Spanish suffixation is one such area in which more research is needed. My study
brings all of these morphological processes together under one major category:
evaluativeness. First I analyzed important pragmatic and sociolinguistic aspects of such
Spanish morphological phenomena. Second, I considered relevant semantic and
morphological issues. Theoretically, my study shows the need to redefine or clarify
pragmatic, sociolinguistic, and morphological concepts. Methodologically, my study was
ethnographic in its data collection and analysis. The data corpus consists of around 600
Spanish evaluatives (EVALs) found in spontaneous verbal interactions. My study also
shows the many different uses, meanings, and functions of these suffixes and connects
them to their basic functions, according to radial category models and pragmatic and
sociolinguistic categories. The multiplicity of functions of the morphemes analyzed here
(at times seemingly contradicting traditional Spanish grammar) is better understood using
an integrative analytical approach. By considering the potential morphological status of
such morphemes, their core semantic senses, their pragmatic functions, and their
sociolinguistic effects, I show the usefulness of this integrative approach to language
study. The following are the major conclusions observed:
* Conclusion 1: Pragmatically, diminutives are essentially attenuation, affection and
derogation markers; whereas augmentatives and superlatives are intensifiers, and at
times, augmentatives may serve as attenuators.
* Conclusion 2: Sociolinguistically, evaluatives may mark contexts (e.g.,
informality) and groups or segments of the society (e.g., children, women, and low
classes), which may reveal much about the power structure in modern societies of
the Spanish speaking world.
* Conclusion 3: Semantically, all the diverse meanings of these suffixes emerge
cognitively or conceptually from single core senses in each case via metaphorical
connections, inferences, or reanalysis.
* Conclusion 4: Morphologically, these suffixes are all part of one single category:
heads of evaluative phrases.
This is an analysis of the pragmatic and sociolinguistic effects of Spanish
evaluative suffixes (diminutives, augmentatives, and superlatives) in linguistic
interactions of everyday life. The study also indicates some important morphological and
semantic issues related to these pragmatic uses of these suffixes.
Spanish suffixation processes are many and have multiple functions. Explanations
for these have been the focus of many formal linguistic analyses in the areas of
morphology, semantics, phonology and syntax. Morphological analyses have shown how
complex words are formed and for which purposes. Semantic analyses have shown the
specific meaning and use of these affixes. Phonological studies have accounted for the
specific phonetic form of the affixes, their phonotactic constraints, and also the
allomorphy found in many of them (Miranda, 1999), among other aspects. In syntax,
specific lexical categories are affected by these processes, and also by the equivalency
between suffixed (synthetic) and non-suffixed (analytical) forms such as the prepositional
phrase and the adjective in Vera-Lujan's (1986) examples: "Pedro es de Valencia" and
"es ValenciANO" (from Valencia vs. ValenciAN). My study also addresses other crucial
issues in morphology and semantics.
Traditional accounts of Spanish diminutives, augmentatives, and superlatives leave
much unexplained. Phenomena such as navaj+ero (knife+agentive -er) with the meaning
of "someone who often uses a knife with criminal purposes" (Vera-Lujan, 1986: 32)
cannot be fully explained without further pragmatic analyses. For instance, we cannot
explain (based only on morphosyntactic and/or phonological and semantic accounts)
where the idea of criminality comes from, if we do not refer to contextual and social
We also could not explain the fact that some processes occur more frequently in
some social groups than in others (e.g., diminutives are used more often with women-
related words than with men-related lexical items). Such phenomena require a two-fold
analysis in our systematic account of language use. First, we must account for these
formal aspects of language (including these suffixation processes) from a morpho-
syntactic, semantic, or phonological perspective. Second, such phenomena need
systematic sociolinguistic and/or pragmatic explanations. This second analysis is the
main purpose of my study. In this way, my study applies formal discoveries in linguistic
research to our daily life contexts and situations.
My study focuses on a specific linguistic phenomenon called "evaluative
morphology" or more specifically "evaluative Spanish suffixation," observed in its
sociolinguistic and pragmatic dimensions. The Spanish evaluatives suffixes (or EVALs)
my study considers are diminutives (DIMs), augmentatives (AUGs), and superlatives
Suffixation is the most common resource for word formation in Spanish and the
other Romance languages (Mufioz, 1994). Given this fact, the data consist of a large and
rich corpus of EVALs. Moreover, while many pragmatic and sociolinguistic studies deal
with phonological and syntactic issues, very few deal with Spanish morphological
phenomena. My study should contribute to filling this linguistic research gap.
Given the lack of analyses of this type of Spanish morphology, the primary aim of
my study was to qualitatively describe the use of Spanish evaluative suffixes (EVALs)-
namely DIMs, AUGs, and SUPERLs-from a pragmatic and sociolinguistic perspective.
Schneider's (2003) exhaustive study on diminutives (especially English) indicates that
one reason for the problematic and puzzling state of DIMs (from a conceptual, scientific,
and academic perspective) is that these suffixes "have not, as a rule, been studied from a
pragmatic perspective" (p. 1). The objective is, therefore, to answer the following three
* Question 1: How do Spanish EVALs affect speech act performance?
* Question 2: What effects do Spanish EVALs have in linguistic interactions in
* Question 3: How can we account for the various meanings and uses of such
Question 1 deals with pragmatics. Question 2 deals with sociolinguistics.
Pragmatics and sociolinguistics are the two main areas my study focused on. Question 3
deals with semantics. Although not necessarily a semantic analysis, the semantic
denotations of the affixes I studied need to be clearly discussed, to account for the
different pragmatic effects of these EVALs.
Three areas were related to my study: pragmatics, sociolinguistics, and evaluative
morphology. Theories of morphology and semantics are mentioned as they relate to this
discussion. Some morphological issues are addressed in the section on evaluative
morphology. Semantics is compared to pragmatics in the first section of this literature
review. The order of presentation simply follows the relevance of such issues for the
goals and scope of the present research.
Pragmatics is the study of language use in context; or more accurately, "the
cognitive, social, and cultural study of language and communication" (Verschueren et al.,
1995: ix). However, it is necessary to further clarify the use of this term in the present
study. Since this is a relatively new field of linguistics, the conceptual framework is still
relatively vague. There are two major conceptual approaches to defining the field of
linguistic pragmatics: the holistic approach and the segmental approach. The segmental
approach concerns the different language components studied in linguistics
(phonetics/phonology, morphology, syntax, and semantics). It is called segmental
because it deals with one specific area of language competence. This approach considers
pragmatics one more language component. In other words, pragmatics is part of a native
speaker's competence (Hymes, 1972). This issue gave rise to criticism of Chomsky
(1965) who (in the view of a number of linguists) overlooked pragmatics as a language
component, or part of the linguistic competence of native speakers. In the conceptual
segmental approach, pragmatics deals with what semantics (and other grammar
components) do not fully account for. It involves the study of speech acts, conversation
norms, politeness, discourse structure, and (micro)sociolinguistic aspects.
Many linguists today agree with this view. Pragmatics is then considered part of
grammar (Moskowitz, 1998). Moskowitz attempts to define language by first considering
what language consists of (for example, rules of language structure or grammar).
Moskowitz indicates that grammar includes "rules of phonology ... of syntax ... of
semantics ... and rules of pragmatics, which describe how to participate in a
conversation, how to sequence sentences, and how to anticipate the information needed
by an interlocutor." (p. 530). Other proponents of such an approach are Jaworsky and
Coupland (1999), who take a discourse analytic view. They made an important
distinction between pragmatics and semantics: pragmatics deals more with the meaning
of utterances in specific contexts of use. All of these authors see pragmatics as a
component of language.
The holistic approach examines language use across all instances of language. In
this approach, every linguistic analysis accounts for language use. In other words, any
linguistic analysis that entirely and openly ignores language in use is superfluous and
disposable. Proponents of the holistic approach see pragmatics more as a perspective on
language than a component of language. Some proponents of the holistic approach are
Kasper and Rose (2001), and Verschueren et al. (1995). The main tenet is that every
language utterance is performed in a context, with a clear goal, and an ultimate (maybe
subconscious) intention. Kiefer (1998), in line with Verschueren, indicates that
pragmaticss can de defined as the functional perspective on language" (p. 272).
Both approaches are important: my study takes an eclectic approach. It examines
language use in specific contexts and across all instances (Ninio and Snow, 1996). A
segmental approach lets us answer questions such as:
* How do people process a communicative act in a concrete speech situation?
* What do people attempt to accomplish by communicating?
A holistic approach may help us examine the implications of these questions for all
components of language. This perspective view is epistemologically sound, since it
reminds us of the integrated nature of language. The segmental approach is
methodologically practical, since we can segment our object of study (language) for
academic and research purposes.
For the purpose of my study, then, a pragmatic function has to do with the ultimate
cognitive (and sometimes subconscious) intention of the speaker when uttering a
linguistic unit (phoneme, morpheme, lexical item, and so on); and more specifically the
intention of the speaker when uttering a bound morpheme. Finally, whether seen as a
component of language or a perspective on language, pragmatics must be part of any
linguistic analysis with integrative purposes because "a language user makes a systematic
analysis of the social context. This analysis is based on strategies involving schematic
knowledge structures (frames) about social, interactional and communicative behavior of
speakers" (van Dijk, 1981: 298).
It is important to note here that in the area of pragmatics, my study does not deal
with some formal pragmatic aspects such as presuppositions and implicatures. As a
pragmatic study, my study deals with ultimate intentions and effects in linguistic
interactions. More specifically, it deals with the (sometimes subtle) reasons for and
effects of using a diminutivized, augmentativized, or superlativized word instead of its
Pragmatic Theoretical Tenets
My notion of a speech act as a primary pragmatic unit comes from Austin (1962)
and Searle's (1969) Speech Act Theory. According to Sbisa (1995), one of the two main
ideas in Speech Act Theory is that any kind of utterance can be considered an act.
Examples of these speech acts are asking for information, requesting, asserting, and
complaining. Speech acts are normally performed via an actual linguistic utterance such
as a declarative sentence or an interrogative phrase, which is considered the locution of
the speech act. However, the intention or final goal of the speaker may be different from
that of the apparent syntactic form of the locution. For example, a question may be asked
without really asking for information but for another reason; greeting, for example ("How
are you?"). This intention of the speaker is called the illocutionary force in Speech Act
Theory. The effect on the addressee (the actual final result of the speech act) is the
Speech Act Theory demands a distinction between the semantic weight and the
pragmatic force of these suffixes. This is another fundamental principle of Speech Act
Theory. As Sbisa (1995) said, "a distinction has to be drawn between the meaning
expressed by an utterance [semantics] and the way in which the utterance is used
pragmaticss]" (p. 496). Jaworsky and Coupland (1999) showed the difference between
understanding a sentence based on the meanings of its words and the referential
meanings, and understanding the same sentence in relation to its intended meaning in a
Important differences exist between the basic semantics of Spanish EVALS and
their pragmatic functions. The basic semantic denotation of DIMs (Chapter 3) is littleness
(hence the grammatical name "diminutive"). The basic denotation of AUGs is bigness.
The basic denotation of SUPERLs is very (more common in modem Spanish) and most
(which can be considered a type of semantic shift). These would be, in technical semantic
terms, the intensions of these affixes. Now, many of the uses of such suffixes go beyond
these basic semantic effects. Again, in technical semantic terms, other extensions of these
Spanish suffixes could be associated with their semantic connotations, and these
ultimately may be linked to pragmatic norms (for a better understanding of the terms
intension, extension, denotation, and connotation, see O'Grady et al., 1997).
Although potentially connected (shown below), it is still necessary to keep the
semantic denotation and the pragmatic effects of EVALs separate, as two distinct
linguistic components. Even from a formal semantic perspective, it is recognized that
the fact that [some expressions] would be considered ... very strange ... in [a]
situation ... shows that over and above truth-functional [semantic] properties there
are other factors which decide our interpretation of linguistic utterances. One
suggestion for an analysis of these factors is to say that there is a set of
communicative norms which aim at making the exchange of information between
the participants in a speech situation as effective as possible. Basing oneself on
these norms one could say: one should not say p v q if one can say p or p & q, both
of which by virtue of their truth-conditions give more definite information than p v
q. One should utilize linguistic expressions as effectively as possible, making both
what one says and what one does not say relevant to how what is said is
understood. This is normally one of the implicit assumptions of linguistic
communication (Allwood et al., 2001: 37).
The norms (in bold letters above) the writers refer to in the previous quotation are
pragmatic in nature. That these are pragmatic norms is clear because Allwood et al.
directly referred to the Gricean Maxims (Grice, 1975).
My study adheres to Austin and Searle's notion of speech act as a major unit of
pragmatic analysis. Since context and communicative intentions beyond purely
grammatical rules (grammatical from a traditional view) are essential parts of our
everyday linguistic interaction, we cannot ignore pragmatics when studying EVALs.
Sociolinguistics is the study of language in society. One main objective of
sociolinguistics is to study the interplay of language, society, and culture in human
communication (Wolfson, 1989). Sociolinguistics, then, may observe linguistic variations
with various societal functions. A sociolinguistic analysis is needed to accurately
determine social motivations of Spanish suffixation processes. According to Lippi-Green
(1997), people "situate themselves in relationship to others, the way they group
themselves, the powers they claim for themselves and the powers they stipulate to others
... People use language to indicate social allegiance ... to create and maintain role ... in
such a manner that the linguistic varieties used by a community form a system that
corresponds to the structure of the society" (p. 31). My study aims at observing
sociolinguistic issues regarding evaluative suffix use.
Despite the apparent importance of such an endeavor, there is a paucity of formal
published work on the interface between Spanish morphology and sociolinguistics. The
scarcity of studies addressing the sociolinguistic implications of Spanish suffixation is
obvious when we examine important and extensive bibliographies on the topic both
recent and early (Bosque & Mayoral, 1979 and Pharies, 1994). The gap in dates of these
two bibliographies shows that this paucity in research has not improved significantly over
the years. At the Spanish morphology level, thus, we have few studies in the
sociolinguistic arena, unlike the numerous studies in the area of phonological variation
and sociolinguistic impact (Perissinotto, 1975; Caravedo, 1990; Alba, 1990; Calero-
Fernandez, 1993; Medina-Rivera, 1997). This is true in Spanish and also in English,
observable in studies on the interface between phonology and sociolinguistics, such as
Labov's (1966, 1972) pioneering work and others.
An extensive literature search yielded few formal linguistic accounts showing the
sociolinguistic implications of these suffixes. One isolated account requires mentioning
here. Mufioz (1994) assumes that some of these forms were cultas (learned, educated,
with culture/education) whereas others were not. Mufioz implies that some words with
certain EVAL suffixes characterized people with a low level of education. This would
give us a linguistic criterion to mark these people socially as belonging or wanting to
belong to the class culta (i.e., the high social class, the one with education, and normally
the one with power). Mufioz is one of the few studies connecting Spanish EVALs and
Other than this isolated mention, there is a significant lack of formal references to
the connection between some Spanish suffixation processes and its sociolinguistic
implications and effects on the interlocutors. Consequently, my study aims to account for
those still-unexplained processes and to fill this gap in the literature on sociolinguistic
research in the Spanish language. It is important to note that my study does not touch
upon major macro-sociolinguistic aspects such as language planning, language policies,
and the like. The focus here is primarily microsociolinguistic.
Before further defining the type of morphology my study focuses on, let us
consider some general issues in the realm of morphology.
Morphology in general
Theoretical approaches in the field of morphology do not deal clearly or directly
with the type of morphology discussed in my study. My study focuses on the pragmatics
and sociolinguistics of the morphological phenomena. As a conceptual tool, I treat these
linguistic units (Spanish EVALs) as separate morphemes, which is in line with both
traditional linguistic theories and more recent theories such as Halle and Marantz's
(1993) Distributed Morphology (or DM). This morphological approach considers both
stems and affixes as lexical entries or vocabulary items. Thus, my present study considers
DIMs, AUGs, and SUPERLs as separate morphemes, with relatively independent
semantic and phonological information. As a starting point, my study addresses semantic
affixes (Miller 2003, personal communication) and not other types of affixes (such as
phi-feature affixes observed by Chomsky, 1993). Henceforth, my study avoids the fuzzy
and unclear derivation-inflection dichotomy as a theoretical foundation because of the
arguments outlined below.
Evaluativeness in morphology
Evaluative morphology (more prototypically, diminutive and augmentative
morphology) has given rise to much research and interest. Many studies address this type
of morphology (Corbin, et al., 1999), probably because of its unusual characteristics in
relation to other types of morphology. Evaluative morphology has also been called
affective morphology (Beard and Szymanek, 1988; Volek, 1987). Mackenzie (2001)
called the DIMs, AUGs, PEJs (his list did not include SUPERLs) "affective" suffixes.
Bruyne (1989) called this type of morphology appreciative. Howard (1998) called it
expressive. Hualde et al. (2001) classified the Spanish diminutive and augmentative
suffixes as emotive or appreciative morphemes.
Chapter 2, which covers important fundamentals of EVAL morphology, elaborates
further on the definition of this type of morphology. A brief introduction is given here as
an overview. In simple terms, evaluative morphology refers to the synthetic marking of
features such as size and positive/negative emotional affect, which is the reason
"affective" is another term used to describe this type of morphological processes (Bauer,
1997.) Of course, Bauer recognizes that this would define mostly the core or prototypical
evaluative morphemes, but not the marginal ones, which also exist. Considering Sapir's
(1921) typology regarding morphological processes, evaluative morphology would be
classified within the affixation processes (Sapir, 1911, cited by Anderson, 1992:326).
In line with this simple definition, many authors such as Bauer suggest that the core
areas of evaluative morphology are diminutivization and augmentativization. However,
other areas (such as pej oration and endearment) are also described in this type of
morphology. Furthermore, languages exist (such as Spanish and Italian) in which these
two types of processes (DIM/AUG vs. pej oration/ endearment) are hardly kept apart.
Although Bauer includes concepts such as intensification, politeness, and modesty in the
realm of evaluative morphology, we can conclude that the prototypical elements of this
type of morphology are the diminutive (DIM) and the augmentative (AUG).
The specific types of affixes I studied have the following characteristics:
* They often attach to bases such as nouns (N), adjectives (A), adverb-verbs (Adv-
V), pronouns (Pro), and interjections; in order of importance/hierarchy as suggested
by Ettinger (1974), according to Bauer's (1997) citations.
* They are quantitative (augmentatives, diminutives, superlatives) according to Alvar
and Pottier (1987) since they express some type of degree or gradable quality;
without necessarily having any affective purposes.
* They may have emotive, appreciative, and expressive connotations, according to
Lang (1990). In fact, Volek (1987) describes emotive attitudes expressed by DIMs.
* They have similar functions and behavior as degree words (e.g., much, very)
accounted for as functional heads as in Abney (1987), Corver (1997) and Cinque
(1999). Cinque (1999), particularly, includes Moodevaluative markers in his list of
functional heads (pp. 71, 76). This is precisely the motivation for EVALS to be
considered functional heads.
Pragmatically, this type of morphology can be labeled as "expressive" and not
always "plain" morphology. Expressive morphology has to do with playful expressions,
and poetic, and/or ostentatious effects. Plain morphology, on the other hand, tends to be
purely semantic, and therefore does not express these characteristics (Bauer, 1997). More
accurately and importantly, however, is that the main difference is that expressive
morphology is conscious explicit knowledge, vs. the implicit knowledge of grammatical
markers (Miller 2005, personal communication).
Scalise's (1988) study on Italian, as cited by Stump (1992), presents distinctive
features of this type of morphology. Among these features are the following:
* Base semantic change (which should be taken cautiously, based on the arguments
* Possibility of consecutive application of more than one rule of the same type.
* Syntax-preserving features.
Stump describes evaluative morphology in the broader label of category-preserving rules.
However, Spanish EVALs do not really involve rules (discussed later). These suffixes are
probably more accurately analyzed here as syntactic heads.
Finally, EVAL morphology has been often considered to be morphology with
iconic tendencies. Since DIMs refer to littleness, and AUGs refer to bigness, many
authors think that the linguistic forms may express these gradable features iconically. In
line with tenets of Natural Morphology (Wurzel, 1994), some authors cited by Bauer
(1996) (such as Sapir, 1956 and Fischer-Jorgensen, 1978) stress this idea of the universal
tendency of EVAL morphology for iconicity. They argue that a preference exists for the
[i] vowel in DIMs and for [o] in AUGs, based simply on the commonality of some
phones in the EVALs that they observed. The front vowel, for example, may be
interpreted as expressing smallness, whereas the back vowel expresses bigness. However,
the iconic value of such phones is still unclear. Bauer (1996) shows that this is not the
case in his 50-language sample. He explicitly concluded that "there does not appear to be
any universal principle of sound symbolism operating in markers of the [DIM] and
[AUG]" (p. 201). The slight tendency among some languages may be caused by
coincidence; frequency of some sounds over others in a language; and most probably, this
tendency may be caused by an over-emphasis on some major language families, such as
Indo-European. Thus, he concluded that the tendency for iconicity might very well be
culture specific, not universal. Bubenik (1999) argued the same when he stated that "we
have to assume that various languages would rank differently on the scale of iconicity"
Spanish EVALs. Spanish morphology requires us to discuss Penny's (1993) view
of Spanish evaluative morphology. His diachronic account of Spanish grammar calls this
type of morphology "derivation". He molds his definition to include Spanish EVALs in
the category of derivational processes. Spanish derivation, in Penny's view, involves
adding suffixes to preexisting roots with two different purposes: forming new lexemes
(e.g., Lim6n+ada = Lemon+ade) and marking the speaker's attitude toward a certain
entity or reality (e.g., pejoratives). He called the first process "lexical derivation", and the
second "affective derivation". However, these suffixes should not be described as
Halle and Marantz's (1993) Distributed Morphology may account for fundamental
morpho-syntactic characteristics of Spanish EVALs. The Distributed Morphology
approach conceptually categorizes Spanish EVALs as distinct morphemes. Because of
the relative conceptual adherence to this morphological theory, my study does not make
the traditional distinction between derivational and inflectional affixes, even though
traditional Spanish accounts of these suffixes categorize them as derivational.
The derivation vs. inflection distinction is completely artificial and adhoc: no
consistent examples in real languages sustain such a distinction. Miller (1993) argues that
this dichotomy is not even a continuum, as Kiparsky (1982) and Bybee (1985) seem to
suggest. The language-particular realization of grammatical vs. semantic affixes is
categorical: one or the other. Miller (1993) clearly states that "many affixes cannot be
classified as either derivational or inflectional" (p. 13). Spanish EVALs give empirical
evidence for this statement.
Spanish DIMs, for example, do not change lexical category. If they attach to a
noun, the resulting word is also a noun. There may not be a semantic change at all. A
"car" may be exactly the same as a "car+DIM", just with smaller proportions. What
should we say about this DIM? Is it still a derivational suffix, even though it does not
affect lexical category or semantics? The solution adopted in my study and many others
is to simply dispose of this dichotomy in the treatment of such affixes; or, at least, to not
consider this dichotomy as fundamental for general morphological conclusions.
Ambadiang (1997) observed similar issues in the morphology of DIM formation: some
important features of the morphology of DIM formation are not properties of common
derivational processes (e.g., the relationship among the different allomorphs). Well-
known scholars in the field of morphology (Anderson, 1982; and Spencer, 1991) have
also recognized this problem. Spencer literally indicated that this type of morphology
"falls midway between inflection and derivation" (p. 197); one of many ad hoc
stipulations by which authors force the use of such dichotomy.
Penny's (1993) approach, like many other studies that use the derivation-inflection
dichotomy, faces categorization problems. He recognizes that in derivational affective
processes, there is no formation of new lexemes with different meanings from the base or
new semantic-syntactic categories, for which he uses the "cat+DIM" example. He argues
(accurately) that "cat" and "cat+DIM" could have essentially the same referent. Most
explanations of derivation (vs. inflection) characterize derivation as producing new
words, lexemes, or a large meaning change with regard to the base (Kurylowicz, 1964;
Bybee, 1985; Bauer, 1988; Bubenik, 1999). This presents a conceptual problem with
Penny's account. Some of his examples are also problematic. Such is the case with
"car+DIM", which can actually mean the same thing, contrary to what he stated. He does
recognize that there are cases that are difficult to account for, and this is precisely the
reason for not adhering to such a dichotomy as a major conceptual tool in my study.
Another conceptual problem is that Penny does not explain why affective processes
should be labeled derivational (considering that they may not produce or derive new
Penny classifies Spanish evaluative morphology within the affective derivational
processes, where he includes diminutivization, pej oration, and augmentivization.
Superlatives are not classified in this type of processes in Penny's account, but within
adjectival comparative processes. In classifying such morphological processes as
affective, we face another problem. One of the most common and important pragmatic
functions of DIMs and AUGs is speech act attenuation. In some cases, an assertion is
marked with a DIM to mitigate the degree of commitment of the speaker, which is hard
to connect to an affective purpose. Important pragmatic functions exist that are clearly
affective, such as marking proper nouns with the [dear] feature via diminutivization.
However, affection is just one of the several components or features of this
evaluativeness. The term evaluative covers all these different functions of such suffixes,
unlike the term affective.
Since they are a type of semantic suffixes, they are distinct from other affixes with
more grammatical functions; such as those "phi-feature" affixes discussed in Chomsky
(1993) and others in relation to Agreement (Agr). In syntactic theory, Agr is a collection
of (p-features (e.g., gender, number, person) common to the systems of subject and object
One of Lowie's (1998) conclusions most relevant to my study (after reviewing
important morphological theory and psycholinguistic models) is that "lexical
representations should contain or refer to properties defining syntactic, semantic/
pragmatic information. The syntactic properties can be seen as ... argument structures of
the lexical representation." (Lowie, 1998: 63). Since Chapter 3 focuses on EVALs'
semantics, and Chapter 4 deals with pragmatic issues, I briefly considered the important
morpho-syntactic aspects above as a preliminary base (for this pragmatic rather than
To finish this summary of morphological properties of Spanish EVALs, let us
briefly consider some frequency and productivity issues. Regarding frequency of Spanish
EVALs, approximately 75% (447 of the 573 words) of all the EVAL words in the data
analyzed are DIMs; around 20% of all the EVALs (around 100 words) have AUGs; and
only 5% (30 words) have SUPERLs. Thus, the data include four times as many words
with DIMs as words with AUGs, and AUGs are used around four times more than
SUPERLs (see table 1 and figure 1). Even though the table and figure below reflect only
the data analyzed, DIMs seem to be the most common EVAL elsewhere.
Table 1. Frequency of Spanish EVALs
DIMs AUGs SUPERLs Total
# oftokens 443 100 30 573
Percentage 77% 18% 5% 100%
Figure 1. Frequency of Spanish EVALs
Regarding productivity, we normally find in the data that SUPERLs attach to
adjectives only; AUGs attach to nouns and adjectives; and DIMs attach to nouns,
including abstract nouns that traditionally do not subcategorize for EVALs
("attitude+DIM" or "lack+DIM"); adjectives, adverbs ("now+DIM", "near+DIM"), verbs
("eating+DIM"), numerals ("one+DIM"), exclamatory expressions ("cheers+DIM!",
"certainly+DIM!"), other quantifiers ("many+DIM"), one prepositional use ("in-front-
of+DIM"). Even though it is not in the data, other uncommon uses of DIMs, such as one
with a pronominal (suyita or "yours+DIM") have been attested in modern Spanish.
Some may argue that a type of AUG attaches to verbs (llor6n = cry+AUG) and
makes up a noun or adjective. Some others may argue that the SUPERLs also attach to
adverbs (much+SUPERL). This is debatable in grammatical terms, but regardless, DIMs
are still much more productive than AUGs in the data and elsewhere, and AUGs are at
the same time more productive than SUPERLs.
EVALs illustrated: diminutives. Let us now exemplify or illustrate this concept
of evaluation with one of the most prototypical and studied EVALs; namely, the
diminutive. DIMs have received much attention in linguistic research, especially from a
semantic perspective (Hasselrot, 1957; Alexopoulos, 1994; Jurafsky, 1996; Scalise &
Grandi, 2001). In the Spanish language this has not been different (Alonso, 1937;
Gonzalez-Olle, 1962; Nafiez, 1973; Zuluaga, 1991; Howard, 1998). For Alonso (1937),
one of the first Spanish grammarians to formally account for DIM meaning and uses,
DIMs are more related to affection than to littleness. This shows the evaluative nature of
DIMs, since these suffixes are part of those affixes with which "nuestro pensamiento no
se detiene en las palabras ... sino que las atraviesa como la luz al aire y va a dar de un
modo peculiar en las cosas mismas o derechamente en el animo del pr6jimo...indudable
valor sistematico-estilistico" (Alonso, 1937: 43) (our thoughts do not stop in the words ...
but they go through them as light through air and end up, in a peculiar way, in the things
themselves or directly in the mood of the others...without a doubt, a systematic-stylistic
value). He perceives DIMs as having a fictive and/or ludic character, since they express a
somehow either imaginary or playful mood, which is in line with Dressler and Merlini-
Barbaresi's (1994) [-serious] feature. For Alonso the basic sense, or "sentido nuclear" as
he called it, of the DIMs is that of mood expression, which is consistent with Cinque
(1999). In general then, DIMs are semantic/pragmatic markers of a subjective evaluation/
appreciation of (often) littleness/affection towards the entities to which they refer.
Nafiez (1973) explains (based on his studies of traditional Spanish grammars) that
the diminutive indicates a conceptual quantitative distinction regarding the magnitude of
the entity referred to by the base. In this quantitative tendency, it often refers to the
general concept of smallness. However, Nafiez criticizes these traditional grammars for
their overemphasis on only this one function of DIMs. He recognizes other important
connotations in the DIMs not recognized by traditional grammarians such as Nebrija
(1492) and the anonymous grammar known as Lovaina (1955), among others. My present
study makes constant reference to these other meanings or functions. Dressier and
Merlini-Barbaresi (1994) present a more complete account regarding diminutives. They
describe diminutive formation as "evaluative (cf. the traditional term valutativi for Italian
diminutives, augmentatives, and pejoratives), that is, diminutives express an evaluation or
judgment 'as to value' (not 'as to fact'), according to the evaluator's intentions,
perspective and standards of evaluation" (p. 153). This is precisely one of the major
arguments for the evaluative label for the type of morphology analyzed in my study. In
the case of diminutives, the evaluation of the entity would be that of a (subjectively)
small, appreciated (or its opposite), and/or endeared one. According to Dressler and
Merlini-Barbaresi, the basic pragmatic connotation would be that of the [-serious]
feature, which is questionable according to the data and analysis presented here, as we
will see in the following chapters.
The following are uses of the DIM already documented. Jurafsky (1996) assigns
the [child] feature as the basic notion of these suffixes. My data and analysis seem to
favor the littlenesss] feature as the basic semantics of DIMs instead. Jurafsky also
presents other semantic senses that emerge from that basic notion mentioned above:
small/little, female, imitation, intensity/ exactness, approximation, and individuation or
partitive. We will discuss more about Jurafsky's arguments later. In relation to pragmatic
functions, the following have been cited: affection, contempt, playfulness, and child/
animal context marking. Dressler and Merlini-Barbaresi (1994), on the other hand, agree
with Jurafsky regarding the semantic base of the DIMs (i.e., [child]). In relation to what
he calls the "pragmatic base", he suggests the [-serious] feature at its core.
Before referring to other authors, let us consider some arguments against Jurafsky
and Dressier and Merlini-Barbaresi's proposal of [child] as the core sense of DIMs. It
seems more plausible to consider [+little], not [child], as the core semantics of DIM and
[dear], not [-serious], as its nuclear or core pragmatics. On the one hand, the label
diminutive approaches more logically the idea of littleness than the idea of childness.
More importantly, however, the [little] feature, with its primarily adjectival (and
consequently modifying) function seems to be more easily morphologically embedded
(as a bound morpheme in a word) than the feature of [child]. The latter is more a
referential item (more a nominal function, primarily), whereas the former is more of a
modifier. Thus, due to this important difference between the morpho-syntactic and
semantic behaviors of these two features, the [little] feature seems to represent the
nucleus of this bound morpheme. We will see, however, that morpho-syntactically
speaking, DIMs (and the other EVALs) present an important core: Mood/modality
function, in line with Cinque (1999).
On the other hand, the pragmatic core of [-serious] proposed by Dressler and
Merlini-Barbaresi misses many important pragmatic functions of such affixes, for
example, that of affection. Not only are there many uses of DIMs that are not [-serious] in
essence, but furthermore, many uses present challenging difficulties if we try to present
some type of semantic or pragmatic association with DIMs. Thus, neither directly (in
essence) nor indirectly (by semantic/pragmatic association) may many DIM uses be
bound by this [-serious] feature. Chapter 3 deals with this in more detail.
In the Greek language there have been important studies on DIMs, and particularly
on their pragmatics. Triantafyllidis (1941), for example, calls the DIMs also "caressives"
and assigns to this suffix the pragmatic function of request mitigation. Holton et al.
(1997) explains that one of the major functions of this type of affix is that of depreciation,
which is in line with the pejorative uses found. Other authors have related these suffixes
to speech acts (Sifianou, 1992). For Sifianou, they may serve as markers of informal
positive politeness and friendliness. Alexopoulos (1994) agrees with Sifianou in this
politeness marking function. Crocco-Galeas (2002) suggests the following pragmatic
functions: attenuation, understatement, meiosis of the speaker's commitment to the
illocutionary force of the speech act, and the mitigation of the interlocutors' obligations.
Giakoumaki (2000) also discussed what my study refers to as the "euphemistic" DIM.
In the realm of sociolinguistics, authors such as Giakoumaki (2000) and Daltas
(1987) refer to the gender-marking functions of DIMs. They observed that, in general,
females use more DIMs than males (they suggest that this is caused by more involvement
of women with children). Giakoumaki particularly observed that women use more
euphemistic DIMs than men in Greek.
My study takes primarily an Ethnography of Speaking approach (Hymes, 1962).
Hymes' SPEAKING model provides a structure for my study to perceive components of
the interactions analyzed: setting, participants, ends, act sequence, key, instrumentality,
norms, and genre. As can be observed, the model is very appropriate for my study since it
implies sociolinguistic and pragmatic perspectives. In many of the descriptions herein,
there is reference to one or more of these speech components to expand such
descriptions. My study is primarily a micro-sociolinguistic approach (focus on face-to-
face interactions and reference to speech communities), supplemented when appropriate
by macro-sociolinguistic issues (for example, Chapter 5 addresses gender-related issues,
which have been studied from a macro perspective). Since my study combines both
function elements (the pragmatic functions of EVALs) and form elements (the forms of
these suffixes), it resembles Schneider's (2003) formal-functional paradigm in his
analysis of English DIMs.
Data and Participants
The data corpus (Appendix D) that is the basis for the analysis in my study consists
of 580 (7 were not analyzed because after further analyses, they did not really constitute
examples of EVALs) tokens found in about 300 situations where evaluative suffixes were
used in spontaneous interactions in Spanish. The motivation for categorization of data
into situations or events is that ethnography (more details below), pays special attention
to speech events. Examples from this corpus were selected and discussed in detail for the
data analysis chapters. All the examples are given first in Spanish (as originally uttered)
in quotation marks with a literal translation into English in italics underneath. If the literal
translation seems unclear, another translation is given in parentheses, next to the literal
one, for clarification purposes. The use of the suffixes in the examples is marked by DIM,
AUG or SUPERL (in capitals) for easier reading.
Using the ethnographic approach (Hymes, 1962) for data collection, a large portion
of the data were taped conversations among native speakers of Spanish. These
conversations took place without any prescribed task in encounters such as church
fellowship meetings, services, phone conversations, television programs, and other less
structured interactions. A long subset of the data examples comes from one particular
speech community, a Spanish-speaking religious group in Gainesville, FL. where the
researcher is a member. By being a member of such a speech community under study, the
researcher could approach this not only as an outsider linguist and researcher but also
with an emic perspective, as defined by Pike (1958). In Pike's discussion of such a
methodological approach, it is clear that the emic perspective helps the researcher to
concentrate on the intrinsic socio-cultural distinctions that are relevant and essential
according to the members of the community, group or society under study. It helps the
researcher to discover which phenomena, properties, or features constitute the worldview
of a given society.
From this investigative perspective, it is critical that the researcher observe the
community as is. Therefore, in such a linguistic study, naturalistic data are the most
suitable. Boxer (2002) believes that "most good analyses of spoken discourse employ
data that captures spontaneous speech among interlocutors, since elicitation instruments
necessarily interfere with the naturalness of spontaneous discourse" (p. 10).
Radio and TV talk, one of the data sources for my study, is one of the many
techniques for data collection that Boxer (2002) surveys. Admittedly, she recommends
being cautious of this type of data, since it may not be representative of naturally
occurring conversations. However, most of the TV programs that contributed to the data
here were live spontaneous entertainment programs or talk shows in interaction with a
live audience. In this way, these are still samples of naturalistic data (admittedly, with
perhaps a slight difference in speech event). There are also several movies and soap
operas, which may be less naturalistic; but they still provide adequate samples of the use
of the suffixes under study.
The data collection in my study resembles that of Milroy's (1980). Milroy audio
taped spontaneous linguistic interactions by interviewing people. There were
interruptions (e.g., telephone calls) during such interviews, but the audio taping was not
interrupted at any time. People then were recorded in both types of interaction: being
interviewed and also interacting with people off-record (on-interview data and off-
interview data). These off-record pieces of interactions were crucial for her analysis. In
this way she captured spontaneous talk and avoided what Labov (1972) called the
Observer's Paradox. Milroy's argument was that these off-record pieces captured true
spontaneity in language use because of the interviewees being unaware of being
recorded. That was precisely the reasoning behind the taping that took place for the
collection of the data for my study. People were first unaware of being recorded for the
type of linguistic analysis I carried out; thus, their language use was truly spontaneous.
Unlike Milroy's method, the data in my study do not come from interview contexts but
purely spontaneous interactions.
The main speech community analyzed is a Spanish-speaking church of about 70
people where many nationalities converge in different types of social activities. Even
though a religious environment, the amount of social encounters is significant. Also, it
represents an encounter point for people of different nationalities. Most of the
participants were either Cubans or Puerto Ricans, with others from Honduras, Nicaragua,
Mexico, Argentina, Venezuela, Colombia, Bolivia, and other Latin-American countries.
As indicated above, some data also came from TV programs (soap operas, movies, talk
shows, entertainment programs, and so on) in a well-known Spanish-speaking TV station
based in Miami, Florida, USA: Univision. Some data also came from Internet sites or
chat groups, phone conversations, and everyday talk where the researcher was one of the
interactants. Direct participation of the researcher in such interactions was very useful,
especially when trying to understand the nature, purpose, and context of these linguistic
After collecting the data, I transcribed the suffixed words, paying special attention
to the contexts of those words to determine the meaning and intention of the suffixation.
Then, it was necessary to categorize the suffixed words into major sociolinguistic and
pragmatic categories. To perform such taxonomy, this categorization process considered
the different SPEAKING speech components described above and also common
sociolinguistic and pragmatic labels in the corresponding literature described below.
Subsequent to transcription and classification, triangulation of the data took place.
This is a common approach in studies of this type to tap into the participants' own
perspectives on interactions under analysis (Gumperz, 1982). Two triangulation
techniques were used: quasi-ethnographic interviews (as defined in Boxer, 2002) and
questionnaires (see appendixes). The questionnaires were given to Spanish L1 speakers.
Twenty-five L1 speakers' questionnaires were analyzed, and the discussion is presented
in the data analysis chapters (primarily Chapter 4 and Chapter 5). In the ethnographic
interviews that took place after the recording, the interviewees answered questions that
helped to either confirm or reject hypotheses in relation to particular instances. These
answers enriched and supplemented the analysis at the time.
These interviews (around 20 in total) with native speakers of Spanish also took
place. These informants' (in both the interviews and questionnaires) ages spanned some
forty years (from late teens to early sixties). Both female and male speakers participated
in the data triangulation process. Also, speakers from different countries (mostly Latin
American countries) participated. Some of the interviewees contributed to the data corpus
and others did not. These interviewees that did not contribute to the data corpus
commented on the grammaticality, potential meaning, and effects of the expressions they
were given. During these quasi-ethnographic interviews, the interviewees answered
triangulatingg" questions with simple non-technical terms which were useful to determine
the participants' opinion of the semantics, pragmatics, grammaticality, and
sociolinguistics of the suffixes used. These questions served to delve more deeply into
the interviewees' perception of such linguistic forms, since much of this is not at the
conscious level of the speakers. The following are examples of the types of questions in
* Can you tell me the difference between carro (car) and carrlTO (car+DIM)?
* Why didn't you say carro instead of carrlTO at this moment?
* What can you tell me about the person or activity you hear in this recording?
* Can you paraphrase what you just heard?
* What is this person trying to do or achieve here? What's her/his ultimate intention?
* Can you imitate a person with the following characteristics (e.g., a rich lady at a
* What do you think is going to happen after this point in the recording (the tape
* Is this acceptable or good Spanish for you?
* What degree/size do the following features/items have?(car vs. car+Dim; blue vs.
* Why do you think this word can have an ito/-ote/-isimo but this other word
Many of the questions used in the questionnaire and the interviews had a
psycholinguistic motivation based on Reeder's (1996) speech act comprehension
strategy. This technique was used to test children's pragmatic awareness (of speech acts)
in Reeder's study. The children undertook 12 task items in a randomized fashion. Before
receiving a stimulus sentence, they were given an explanation of the context of the
sentence they were about to hear ("Would you like to look at the books?"). This sentence
took place in a classroom format, and a teacher had uttered this sentence to indicate to the
students to start the reading section of the class. In the testing section, after this
clarification, the children were asked which alternative (of two) or meaning they thought
the teacher had in mind when she said "Would you like to look at the books?" The two
response alternatives were: 1) "I want you to look at the books", and 2) "Do you want to
look at the books?" According to the general procedure of the experiment, #1 was the
expected answer and #2 was just a distractor. It would show whether the children
participating in the test were in fact aware of the actual speech act (command) given by
the utterance. In my study, these questions were intended to determine what this (already
internalized) pragmatic awareness of L1 speakers informed us about the different
functions of Spanish evaluative suffixes.
The results of this data analysis process are presented in Chapter 4 (Pragmatic
Functions) and Chapter 5 (Sociolinguistic Effects). The first one focuses on pragmatic
categories such as speech acts attenuation or intensification, politeness marking, and
related issues. The second addresses sociolinguistic categories such as in-group identity,
social distance, and social context marking.
In relation to the type of linguistic phenomena analyzed in my study and its goals,
this is a morphopragmatic study, since it shows the interface between morphology and
pragmatics. Morphopragmatics has been defined simply as morphologizedpragmatics
(Dressler & Merlini-Barbaresi, 1994: 55), or in other words, pragmatic functions
performed via morphological processes. This is a relatively new sub-field of linguistic
research. The first studies that were formally categorized as morphopragmatic appeared
in the mid eighties (for detailed explanations, definitions and history of this field, see
Dressler & Merlini-Barbaresi, 1994).
Recall that my study analyzes Spanish morphological processes from a
sociolinguistic and pragmatic perspective. It also looks at pragmatic effects caused by
Spanish bound morphemes. My study has this dual nature: pragmatic and morphological.
The reason for this is that "grammar and pragmatics are complementary domains within
linguistics" (p. 4) as defined by Leech (1983). Because of this double-fold nature
(pragmatic and morphological), my study falls under the category of morphopragmatic
Extensive review of literature in Spanish in the topic yielded only one study that
openly described itself as morphopragmatic (Cantero, 2001). Cantero presents adequate
theoretical accounts of this field in the Spanish language. Yet, we still need many more
formal Spanish descriptive studies in this new and important field of linguistic research.
My study also aims at contributing to filling this gap in the general field of
morphopragmatics, and more specifically in the field of Spanish morphopragmatics. We
definitely need morpho-pragmatic analyses because they show how bound morphemes
may not only obey basic semantic (semantic affixes) or grammatical (phi-feature
markers) requirements of the language (e.g., Spanish) but also may have pragmatic
Mey (1989), who is not a morphologist but a well-known pragmatician, pleads for
the investigation of this connection between morphology and pragmatics. Interestingly,
and in line with my study, Mey (as cited by Kiefer, 1998) shows some morphological
processes that express power and solidarity, which is the case of Spanish EVALs as
shown throughout my study. In conclusion, methodologically, this is a morphopragmatic
descriptive account of Spanish naturalistic data using an ethnographic approach. In
essence, my study is morphopragmatic, qualitative/descriptive, naturalistic, and
My study, in sum, represents an ethnographic attempt to qualitatively describe
morphological phenomena with pragmatic and sociolinguistic implications. The specific
type of morphological phenomena analyzed is Spanish Evaluative Morphology, which
involves prototypically processes such as diminutivization and augmentativization, and
marginally processes such as (absolute) superlativization and pejoration. Pejoration will
be analyzed only as a subcategory of DIMs and AUGs, not independently. My study
leaves suffixes that have already been accounted for as pejoratives (e.g., ejo, ajo, ucho,
uelo) out of formal consideration. There seems to be not much more to say about these
suffixes, pragmatically speaking. In fact, the term "pejorative" is essentially a pragmatic
one. Thus, any account of such suffixes is in essence, a pragmatic account. My study is
limited to the other EVALs: diminutives (DIMs), augmentatives (AUGs), and
superlatives (SUPERLs), which appear to have had up to this point insufficient or
incomplete pragmatic and sociolinguistic formalization.These processes represent
relatively productive and frequent suffixation in modem informal Spanish.
The literature reviewed has shown two important impetuses that motivated this
topic. On the one hand, there is not much formal research on the pragmatic and
sociolinguistic implications of such suffixes. On the other hand, most studies that do
analyze these aspects of such phenomena fail to account for important issues in this
respect. My study, then, is an attempt to fill this gap in the research.
It is a primary goal that, by the end of my study, the reader will have a fuller
understanding of three important related issues (three research questions): 1) The ways in
which Spanish evaluative suffixes (EVALs) affect speech act performance; 2) the effects
Spanish evaluative suffixes can have in linguistic interactions in society; and 3) the
(semantic-pragmatic) connection of the various and many meanings and uses of such
affixes. These aspects constitute the main goals in this work.
EVALUATIVE MORPHOLOGY: FUNDAMENTALS
Evaluativeness in General
The linguistic items analyzed here (e.g., DIMs, AUGs) as any other type of
evaluation expressed in linguistic form, are evaluative in the sense that they convey
(consciously or subconsciously) a type of value of the referents or audience, according to
the speaker's judgment. Hunston and Thompson (2000) give an overview of evaluation,
its discourse functions and how to recognize it, and they observe a lack of consensus
among linguists in regards to its delimitations, categorization and definition. The authors
distinguish two major types of evaluation-driven marking in language: affective (good-
bad) opinion, primarily in reference to entities; and epistemic or probability opinion (e.g.,
"it is fairly certain ..."), in connection normally to propositions. They view these two
types of speaker/writer's opinion as subcategories of evaluation. Hunston and Thompson
further elaborate on subcategories and add two more types of evaluation: expectedness
(i.e., how much resemblance to the norm) and importance (i.e., subjective value in
relation to degree of relevance).
They suggest that parameters such as expectedness and importance can be related
to the basic good-bad parameter. Hunston and Thompson argue that identifying
evaluation "is a question of identifying signals of comparison, subjectivity and social
value ... evaluation consists of anything which is compared to or contrasts with the norm"
(p. 13). This definition encompasses linguistic structures; attitudinal, interpersonal and
discourse-organizational functions; pragmatic inferences as well as conventional, coded
meanings. Since these authors recognize that (linguistic) evaluation can be achieved via
lexical, syntactic or morphological marking, then it follows that we may find evaluative
lexicon, evaluative syntax and evaluative morphology in human languages.
Evaluativeness in Morphology: Evaluative Morphology: Based on the
discussion above and arguments in the previous introductory chapter, evaluative
morphology refers basically to the marking of subjective appreciation of the referents via
bound affixes. The specific types of subjective evaluation analyzed here are those of
diminution, augmentation, and intensification. These correspond respectively to
diminutive, augmentative, and superlative affixes. Chapter I already elaborated on the
definition of evaluative morphology, but this section presents some important additions to
that discussion to place evaluative morphology in a broader context.
The category of diminution, as a general concept, is a universal category since it
may be expressed in all languages (Schneider, 2003). The difference from language to
language is the particular linguistic devices used (e.g., suffixes, separate lexical items,
tones). We then can logically expect its opposite, augmentation, to be found also in many
languages, if not all. Superlativization is a concept that some authors do not even
consider within the same category of diminutivization or augmentivization. However, as
shown in subsequent chapters, in many respects, superlatives (in Spanish at least) behave
similarly to diminutives and augmentatives. Furthermore, as mentioned above, SUPERLs
present a subjective evaluation of the referent. In conclusion, all these evaluative
processes (i.e., diminution, augmentation, superlativization) express the speaker's
attitude/appreciation towards a certain abstract or concrete entity, state or event. They
refer to entities when attached to nouns, states when attached to adjectives, and events
when attached to verbs. They may even evaluate modifications of actions/states when
attached to adverbs.
The Morphology of Evaluatives
Let us start with some simple notions. Traditionally, EVALs have been considered
derivational affixes (Fernandez, 1986). In Modern Spanish, DIMs attach to many bases
such as Nouns (N), Adjective (A), Adverb-Verb (Adv-V), Pronoun (Pro), and Interjection
(in order of importance/hierarchy as suggested by Ettinger (1974), according to Bauer's,
1997 citations). One restriction we observe in modern Spanish DIMs regarding
productivity is the impossibility of attachment to abstract Ns, which is noticed in the
unacceptability of *felicidadita (happiness + DIM), *dependencita (dependence+ DIM),
*inteligencita (intelligence+DIM), *entendimientito (understanding+DIM), *pacita
(peace+DIM). We also have a certain allomorphy in some dialects of Spanish regarding
the distribution of -ito and -ico. In some Caribbean varieties of Spanish, -ico is an
allomorph of -ito (or -cito) when following root-final [t]. The phonological motivation
for ico may be captured in an Optimality Theory approach (Kager, 1999) by making
reference to a well studied constraint called the Obligatory Contour Principle or OCP.
This principle is commonly understood as a constrain prohibiting the adjacency of two
identical elements on a tier (Myers, 1994). For the dialects that use this allomorph, then,
the OCP constraint is high-ranked; whereas this is a low-rank constraint for the "non-ico"
dialects. Speakers of dialects with a high-ranked OCP would have the following forms:
[kart-a] (letter+FEMININE) [kart-ika] (letter+DIM) but not *[kart-ita], for example. It
is a type of dissimilation process; dissimilation of two successive [t]s at the phonological
consonantal tier (t V t t V c).
AUGs have very similar contexts as DIMs, but they are still less productive. For
example, there are no attested examples of AUGs attaching to gerunds (*comiendote=
eating+AUG), contrary to attested cases of comiendito (eating+DIM). Even more, some
AUGs such as -azo and -6n are further restricted, probably due to the etymons that gave
rise to these suffixes, unlike the "pure" AUG -ote. For example, the adjective grande
("big") normally accepts the -ote AUG. Not any other AUG is normally found with this
adjective (grande+ote; *grande+azo, and ?grand+ n).
Superlatives attach to adjectives, not to adverbs (in the case of attachment to Ns,
usually these Ns are, at least potentially, adjectives also). It has been suggested that it
attaches to adverbs. However, let us consider the following arguments against attachment
of -isimo to adverbs:
* -mente (-ly) adverbs do not accept SUPERL (*lentamentisima; lentisimamente),
* Nor do monomorphemic adverbs: (ex. *bienisimo vs buenisimo).
In a few words, this SUPERL attaches to whatever may be used as an adjective. It does
not attach to pure adverbs (bien and -mente adverbs). Let us now look at this morphology
more accurately and in detail.
Morphological Theoretical Tenets: Even though my study has no real foundation
in morphological theory (it is more pragmatic in nature), there are still important
morphological conceptual aspects to clarify. There have been two major theoretical
approaches to morphology and its relation to the lexicon: The word-based lexicon (words
are stored in the lexicon as whole units, not as separate affixes and bases) and the
morpheme-based lexicon (only roots are stored and then rules are applied). My study
adheres to Lowie's (1998) and Miller's (1993) assumption that there is actually a
connection between the two. "Most linguists as well as psychologists will now agree that
instead of a choice between listing and active-rule word formation both strategies are
likely to interact in a complete model of producing and processing morphologically
complex words" (Lowie 1998: 7). These approaches complement each other. Some
morphologically complex words may be stored and accessed in the lexicon as a whole
(especially those less frequent, less productive and more opaque), and others may be
analyzed (those more frequent, productive, and transparent). For example, Lowie
mentions DIMs as a very transparent affixation process (modern illo, an exemption),
which would therefore be formed through rules.
One clear account that connects morphological theory, evaluative morphology and
cognitive fields is provided by Pounder (2000). Pounder set up a list of word-formation
functions, which she divided into primary and secondary. Even though this distinction is
not clearly explained, it seems to be related to the degree of productivity and frequency of
such functions. Thus, diminutivization, which she labeled as DIM('X') and defined as
"'X' is made smaller, diminished" (p. 118), is a primary function. In the data analyzed in
my study, DIM is definitely a much more productive and frequent suffix than the other
two evaluatives analyzed here (AUG and SUPERL). These other two evaluatives,
accordingly, were labeled as secondary in Pounder's list. Even though Pounder does not
overtly cite a "SUPERL" category, she has an "INTENS(X)" function, which resembles
our SUPERL. This INTENS('X') function is defined as "'X' is associated with a high
degree of expressive-emotional intensity or as present in an extraordinary degree" (p.
121). AUG('X') is simply defined as "'X' is increased'" (p. 121) and opposite to DIM. In
all these instances, "X" refers to the base for the affix or function in consideration.
PEJ(X), one of the functions of the DIMs and AUGs analyzed in my study, is also part of
the secondary functions and is defined as "'X' is evaluated negatively" (p. 120). In fact,
Pounder states that most secondary functions belong to evaluative morphology, the main
focus of my study. However, this is more a psycholinguistic approach, which leaves
important morphological aspects unanswered. Furthermore, it seems to be simply a
taxonomic model in that it provides classificatory labels more than theoretical
EVALs as morpho-syntactic markers: As in all Phrase Structure morphological
accounts since Baker (1988), my study relies on the notion that word formation and
phrase formation involve the same operations. Borer (1998) explains that "the thrust of
the argumentation in these works is to show that WF [word formation] phenomena
adhere to syntactic constraints and interact with syntactic rules, and hence are best
characterized as syntactic" (p. 157). Miller (1993) clearly indicates that "the order of
affixes obeys the same principles that govern sentence formation; this can hardly be
coincidental" (p. 16). My study proposes a single head-complement relation for EVALs.
In Distributed Morphology, DIMs, AUGs, and SUPERLs may be said to function
as operators which occupy the head of a whole functional phrase, which shares features
with what Abney (1987), Corver (1997) and Rijkhoek (1998) called "Degree Phrase".
According to Abney and Corver, degree elements (e.g., "so", "more", SUPERL) are
heads that select APs (and other phrases) as their complements. In this syntactic model, it
is said that they project a Degree Phrase (DegP) and select an AP. In their analysis, based
on the X X' X" structuring (common in syntactic models), the constituency is the
Because of the similar syntactic behavior of Spanish EVALs to functional phrases
(when compared to degree operators such as "very" and "more"), based on our definition
of evaluativeness above, and considering Miller's (personal communication) suggestions
and Cinque's (1999) treatment of evaluatives, we propose the following general morpho-
syntactic structure for Spanish EVALs:
Figure 2. Tree diagram of EVALs
Since DIMs, AUGs, and SUPERLs all have an evaluative function, it is reasonable
to unify them as incorporations of an evaluative head. The head of the tree above is the
suffix itself (DIM, AUG, or SUPERL), and the complement of this head could be phrases
such as NPs, APs, AdvP, or VPs (in short, any phrase that can be evaluated via these
suffixes). We would then have the following specific configurations:
1 2 3 4
EVAL-P EVAL-P EVAL-P EVAL-P
EVAL' EVAL' E AL' EVAL'
EV N(P) EVAL A(P) EV Adv(P) EAL (P)
..o .... .ti
car, -ito -ote grand -isimo much -ito trotando
In this hypothesized representation, there is left adjunction, in line with the
syntactic model above. In Kayne's (1994) antisymmetry hypothesis adopted by Rijkhoek
(1998), movement to the right and right-adjunction are illegitimate. Thus, the X(P)
constituents move to adjoin to Evalo or Eval, deriving words such as carrito (car+DIM)
in configuration 1, grandote (big+AUG) in 2, muchisimo (much+SUPERL) in 3, and
trotandito (jogging+DIM) in 4.
One potential problem with the previous proposal involves its account of the
reduplicative or iterative property of Spanish EVALs. On the one hand, iterative cases
(e.g., poquitititito = little+DIM+DIM+DIM) can be treated as successive adjuncts,
assuming it is necessary in the logical form (LF) component of the grammatical system,
where derivations get interpretation. On the other hand, if they are not processed at LF,
then they may be simply phonological copies (PF copies) or part of language play. My
study argues that these iterative cases do convey some type of interpretative force;
namely that of intensification. Even though my study suggests that it be further looked
into, especially by studies of a more morphosyntactic nature than this one, it hypothesizes
that these reduplication-like (or more accurately, iteration) cases constitute a type of
successive adjunction to EVAL-heads. Regardless of how they are treated, it is important
to note that this type of right-edge iteration seems to be a unique property of these
functional phrases (FP). FPs do not normally have this property. In fact, it seems to
violate the haplological constraint or OCP as formalized before (Raffelsiefen, 1996).
Usually, this type of iteration is permitted only on the left-edge, and strictly forbidden on
the right edge. In fact, Miller (in personal interviews) mentions the following examples to
show how this constraint works: boyish vs *fishish. Even though these are not examples
of morphological reduplication, these show that repeating similar syllabic groups on the
right edge is not permitted; which seems to be the only reason for the illformedness of
"fishish". To account for the type of iteration found in evaluative items, we would need
to propose that iteration may be an inherent property of EVAL-Ps (also seen in English
phrases such as "very very very good"). This is one thing that makes these Spanish
suffixes special; they seem to violate a strong constraint.
Non(morphologically) Evaluative Languages: Evaluative Syntax
"Evaluative syntax" is symmetrical to "evaluative morphology". The latter
involves the marking of EVAL features at the morphological level. Thus, we can extend
the definition to the area of syntax. We can think of"EVAL syntax", then, as the marking
of EVAL features at the syntactic level; or in other words, not synthetically but
analytically. Essentially, the difference between analytical and synthetic is that the latter
involves affixal instantiations of the former. One fundamental similarity is that both
markers occupy head positions on EVAL-P trees. We discuss specific examples below.
In relation to languages without EVAL morphology, it should be first noted that "it
is difficult to be sure from grammars that a given language does NOT have a particular
phenomenon", as Bauer (1997) clearly states, in relation to universal tendencies in
evaluative morphology precisely. However, based on traditional typologies and some
discoveries in the review of the relevant literature, this chapter presents some potential
examples. We could logically assume that languages traditionally classified as analytical
will have no EVAL morphology. Such category involves English, marginally, and more
clearly Chinese. The latter has been reported as marking all categories at the syntactic
level, with isolated (and often monosyllabic) words (Crystal, 1997). Even purely
grammatical functions such as tense would be marked lexically (with different lexical
items). Thus, it is easy to predict that the same is true for marking [+EVAL] features
English, on the other hand, has synthetic (work+ED) and analytical (WILL work)
features; which makes it mixed. However, there is a consensus among linguists that
English tends to be more analytical than synthetic. Actually, in historical perspective,
English has been changing from synthetic to analytic structures. In relation to EVAL
marking, we also see the two tendencies: Synthetic (John-Y, dogg-IE) and analytical
(LITTLE/DEAR John, LITTLE/DEAR dog) (Schneider, 2003). However, there are more
cases of analytical EVAL marking than synthetic EVALs. For example, there is no real
morphological equivalent of Spanish -isimo (one marginal EVAL suffix), which means
something like "very". English more often uses the separate lexical item "very" for this
intensive function, even though in modem English a few synthetic instantiations of this
function can be observed in cases like "the bestest" vs. "the very best". Also, English
DIMs such as "-let" (piglet), "-ling" (duckling), and "y-ie" (Johnny, doggie) are not very
productive (Schneider, 2003).
The bases for these few English diminutives are very restricted (often infantile
terms), unlike the analytical equivalents (little/small), which combine with many lexical
categories. That is why we have "lousy little..." in English instead of pejorative DIM
affixes such as the Spanish and Fula ones. This may be also observed in the fact that
analytical expressions such as "itsy bitsy, teeny weeny" (analytic diminutivization, even
if we consider final -y as a remote DIM) are common DIMs for certain English-speaking
groups (Schneider, 2003). Some other languages without EVAL morphology are
Samoan, and Quiche Mayan, of which Crystal (1997) says: "many of the features of
Anglo-American motherese ...-such as diminutives- were found to be absent." (p.237)
Thus, it is obvious that some languages have evaluative morphology while others
express related semantic nuances lexically; henceforth analytically. On the analytic side,
we have languages such as Chinese, English, Samoan, and Quiche Mayan. On the
synthetic/morphological side we have many Indo-European languages (e.g., German,
Latin, Dutch, Greek, Scottish), including more distinctively Romance languages (e.g.,
Spanish, Italian), and other non-IE languages such as Fula, Swahili, and Japanese.1
The marking of evaluativeness at one level or the other may have to do with
historical processes or with (typological) representational preferences. These two types of
motivations are in line with Jurafsky (1996), who believes that by "considering the
dependence of synchronic meaning on both historical and human cognitive context it is
possible to tease apart the seemingly paradoxical and unmotivated components of a
particular semantic category: [for example] the diminutive." (p. 562). Anderson (1992)
states that it is not uncommon across languages for the "internal structure of words [to]
derive from earlier syntactic constructions" (p. 348). He cites Giv6n's (1971) famous
aphorism "today's morphology is yesterday's syntax" (p. 348). Anderson (1990) also
supports this idea, since he thinks that "words have the form they do ... especially ...
because of the history of the language with respect to both grammar and to sound." (p.
278). However, it is important to keep in mind that this may not be the end of the whole
story. Recall that it is also possible for morphological anomalies to be altered in language
change to conform to the syntax of the language. Even though there may still be
unexplained phenomena across languages from a historical perspective, it definitely is an
explanation for the difference between evaluative morphology and evaluative syntax.
Diachronically, it is possible that those languages that do not have evaluative
morphology but syntactic or analytic EVALs, have simply not gone through linguistic
changes of the analytical-to-synthetic type. Those that do have EVALs at the
morphological level did go through (or are going through) the changes already. However,
even within the morphological EVAL languages, it is totally possible that those affixes
emerged not from lexical items but from previous bound morphemes. This is very
probably the case with Spanish EVALs. In this sense, reanalysis may have taken place
and previous non-EVAL affixes were re-interpreted or understood as EVALs. Below, we
discuss a language with evaluative morphology (or EVAL morphology language), Fula.
Anderson (1992) explains how noun class suffixes in Modern Fula can be traced back to
earlier pronominal elements. In the section below on Fula, we see that EVALs are also
class suffixes; thus, this could be a reason to hypothesize that EVAL class suffixes also
may have had a non-affixal or non-grammatical status. This can be a good example of a
historical motivation for EVAL morphology.
In conclusion, the reason for marking some categories at different levels across
languages may have an explanation in the history of the language, in basic human
cognition and cultural or pragmatic forces (last two elaborated later since they may be
Morphologically Evaluative Languages: Two Types
Let us start this section with a reference to Fula EVAL morphology, which should
help us observe the relationship of evaluative morphology to the Agr evaluatives such as
the Fula pejorative DIM. According to McIntosh (1984), Fula (West Atlantic branch of
the Niger-Congo language family) has a rich nominal declension system with 25 N
classes, 19 singular and 6 plural. That is why Anderson (1976) calls it a "class language"
and explains that the complete declension often contains 7 forms: citation form (class 1)
and its Pl. (cl. 2), DIM (3) and its Pl. (6), DIM Pej. (5), AUG. (7), and its Pl. (8). Stem
initial consonants (Cs) vary (often within 3 possibilities or grades such as stops, [-stop],
and prenasal stops) or agree/inflect with noun classes, which is known in Fula grammar
as "C gradation" or "alternation". Each noun class has a corresponding form or grade for
the initial C of the stem. For example, when "man" (dim-/rim-/ndim-) attaches to PL. -be,
the initial C grade is continuant ([r]), producing the form rimbe (men); when attached to
DIM -el, the grade is stop ([d]), producing the form dimel (little man); and when attached
to the DIM Pl. -on, the grade is prenasal stop ([nd]), deriving ndimon (little men).
All this morpho-phonological explanation in the previous paragraph should make it
clear that nominal stems are constituents of the same XP as nominal-class markers (i.e.,
noun roots must be accompanied by a nominal-class suffix). In our morphological model,
we could say that these nominal-class suffixes are functional heads whose complement is
an NP (whose head is the noun root). The class suffix limits the reference of the stem. We
need to keep in mind here that Fula EVAL suffixes such as DIMs and AUGs are just part
of these class-suffixes with clear and heavy class-marking features and functions. The
Fula language has 6 evaluative suffixes: ngel (DIM. Sing.), koyU (DIM. Pl.), kal (DIM.
Qu.), nga (AUG. Sg.), ko (AUG. Pl.), and the pejorative DIM ngum (McIntosh, 1984).
Mukoshy (1991) makes a distinction in the 25 N classes referred above: Basic
classes and subsidiary/secondary classes. Among the latter, Mukoshy included the
evaluative suffixes we are dealing with in this paper. In another paper, Mukoshy (1991)
clarified this distinction. The primary class refers to Ns that are fixed within a particular
class and will not change to another (for example, "horse" will always belong to the
nonhuman class). On the other hand, secondary classes (like DIMs) may take any N.
All these suffixes mark features or functions very essential to Fula speakers; as
important as gender reference in Spanish, or number reference in English. In Spanish, we
always find reference to the gender (often Masculine-Feminine) of entities in discussion.
In English, we cannot speak without referring to the number (Sing-Pl) of the entity we
talk about (more on this below in reference to Hardman's, 1978 linguistic postulates). In
the same fashion, Fula people cannot speak without reference to aspects such as class,
quantity, and animateness. According to Mukoshy (1991), in Fula, "a thing is either
ordinary/normal, large, small, or tiny." (p. 25). That is why he states that, unlike other
cultures, Fula people do not see things in terms of masculinity or femininity (like
Spanish-speaking people), but rather in terms of distinctions such as countable/
uncountable, animate/inanimate, and human-nonhuman.
On the other hand, these nominal suffixes, including the evaluative ones, mark
class membership. This is precisely the reason for classifying the Fula evaluative suffixes
as grammar-compliant (vs. pure semantic affixes such as Spanish EVALs), since they are
basically class/number and concord/agreement morphemes, unlike the same type of
morphemes in other languages. They are basically grammatical markers (i.e., they code
redundant information or information already present in the derivation). Anderson (1992)
pointed out that this type of affixes include some sort of [+Agr] features within the phrase
(feature absent in Spanish EVALs). These features can be imposed on class-markers
(such as Fula's evaluative suffixes in our case) due to the position they have in the larger
phrase and in relation to other phrase constituents. Anderson argues that in this Fula class
paradigm, DIM/AUGs act, morpho-syntactically, exactly as the PL(ural) suffixes. Thus,
both PL and DIM, for example, should get the same label.
The other type of morphological marking of evaluativeness corresponds to most
cases of morphological EVALs: semantic suffixes. These, again, are optional in nature
and not required by grammar-compliant rules. Spanish EVALs fall within this category
as do the rest of the Romance languages EVALs and those of many other languages.
This, at first sight, may seem to correspond to the inflectional-derivational distinction.
However, as argued before, this dichotomy is not principled and finds many counter-
examples. We simply refer to these two types of morphological marking as grammar vs.
semantic suffixes (i.e., as a terminological tool).
However, in our morpho-syntactic model, this distinction does not seem relevant,
since both markings may be accounted for by similar principles. We just need to keep in
mind that the difference between the two relies on LF processing. The "grammatical"
type (e.g., Fula) involves redundant features adjoined at spellout or assignment of
phonetic form-PF (not at LF) whereas the "semantic" type (e.g., Spanish) involves non-
redundant features required for LF interpretation. This distinction is only important when
further analyzing the specific grammars of languages with EVALs.
In conclusion, languages mark the category of evaluativeness (or use EVALs) in
one of two ways, synthetically affixall instantiation) or analytically (non affixal
instantiation), which may be in direct connection with the marking of language-specific
postulates. This takes us to the assumption that evaluativeness may be in fact a linguistic
Linguistic Postulates and Evaluative Morphology
Hardman (1978) defines linguistic postulates as "those recurrent categorizations in
the language... most directly and most tightly tied to the perceptions of the speakers" (p.
122). These are so imposed that the interactants view them as natural parts of their
universe or reality. The importance of this in linguistic research is simply that the more
powerful a postulate is, the more involved it is in a language's grammatical system. Very
succinctly, Hardman explains that these postulates are language specific, and that they
may be realized at different levels; morphologically or syntactically, for example, which
was shown in the previous section. Hardman mentions some examples of these postulates
in two different language families. On the one hand, we have postulates such as sex and
number in Indo-European languages. On the other hand, we have data source (something
similar to what O'Grady et al (1997) called "assertion" in Hidatsa) and humanity as
linguistic postulates in Jaqi languages.
We can assume, then, that these postulates originate as human conceptual
categories which may develop special significance in a particular culture and
consequently get coded in the language of that culture. We can also assume that this may
be the case for EVAL marking and for the evaluativeness postulate. If we extrapolate,
then, it would not be strange at all that so many languages exhibit this EVAL feature,
even though it seemingly ought to be culture specific. The important point is that
cognitive/conceptual categories (e.g., littleness) may have culturally very different
realizations. It would be similar to other linguistic postulates, such as number. Hardman
mentioned number in Indo-European (IE) languages, in which this category is a postulate.
Other non-IE languages may also mark number (e.g., Hebrew, Fula). Miller (1993) points
out that all languages have numerals, supporting the idea that number is conceptual (i.e.,
enumerating entities is a cognitive process).
Postulates can pervade languages/cultures and even language families, as the
arguments above suggest. If this is true with number and sex, there are no reasons to
believe that it may not happen with other postulates such as the type of evaluation
discussed in my study. Because of the observations mentioned so far and my findings in
the literature, no doubt DIMs (or EVALs) are "among the grammatical primitives which
seem to occur universally or near-universally." (Jurafsky 1996: 534).
On the other hand, we may need to resort to historical and anthropological
linguistics to give another possible answer to the question of the pervasiveness of
evaluativeness. As Hardman suggests, postulates such as sex and number are inherent in
IE languages. What this implies is that these features are almost unavoidable when
speaking an IE language. However, for non-IE speakers, those features are not of general
importance; they are features with which they do not need to be concerned. One possible
hypothesis for the existence of different linguistic features in a language is that these may
come from language ancestors or Proto-languages. In the case of IE languages, historical
linguists propose a Proto Indo-European or PIE stage (Jeffers and Lehiste, 1979). Thus,
we might hypothesize that one reason sex and number marking occurs in so many IE
languages is that these were also PIE's linguistic postulates (Miller, in personal
communication, clarifies that sex only became a postulate after the Anatolian branch split
off); they just spread to many of its daughters. Even though it might need more historical
reconstructions (which Jurafsky claims to have done for DIMs, as explained in the next
chapter), this is a very plausible explanation for understanding the existence of similar
postulates across languages. We have here another possible reason why the evaluation
postulate is found in many languages. Due to the importance of historical developments
in language, the next section presents a relatively brief overview of EVALs' history.
Origins of Spanish EVALs
It is important to look at EVALs from a diachronic perspective, even if the main
focus of analysis is pragmatic-sociolinguistic. A well-known sociologist of language,
Joshua Fishman (1972b), argued that sociolinguistics might well benefit from historical
perspectives. Fishman believes that "time [or historical] perspective deepens our
understanding of and appreciation for any particular sociolinguistic topic." (p. 146).
Jurafsky (1996): "In recent years, however, many scholars have begun to treat the
synchronic state of the semantics of a language as profoundly bound up with its
diachronic nature" (p. 533). Let us consider each one of Spanish EVALs, in order of
frequency. Even though the following is not a historical linguistic treaty, it should help
the reader to know somewhat more about the suffixes under scrutiny here.
The 24 Spanish suffixes below have been associated more or less with DIM
functions in the literature review (mainly based on lists by Gonzalez Olle, 1962; Gooch,
1967; Fernandez Ramirez, 1986; Penny, 1993; and Pharies, 2002) across dialects and
across time. In Appendix C, the reader can observed a table with these suffixes and some
information about their origins and meanings. Let us list them all here as a starting point,
even though some of them will be hardly discussed since they have either lost DIM
functions or are hardly productive: -acho, -ajo, -ancho, -allo, -asco, -culo, -eco, -ejo, -elo,
-ico, -illo, -in, -ingo, -ino, -iTo, -ito, -oco, -orro, -ulo, -ucho, -uco, -uelo, -ujo, and
When looking into the history of Spanish, we need to look first into the (mainly
Vulgar) Latin language, its main ancestor (the IE language that appeared in Italy probably
around centuries VI or VII, given that Wallace (1989) lists Latin inscriptions form 620-
600 BC). Admittedly, there are influences on Spanish from other languages such as
Basque, Greek, Arabic, Proto-Germanic, and others, as Entwistle (1948) clearly shows.
We can also look for more remote origins if we look at the ancestors of Latin, namely the
Proto-Indo European (IE) language, whose daughters include Italic/Latin (Crystal, 1997).
PIE and Latin antecedents
There were Indo-European suffixes withfunci6n minorativa (Gonzalez-Olle 1962:
177). Of these, according to Gonzalez-Olle, -lo- evolved the most in Latin (> -olo- > -
ulo-). But -ko- is also a reported IE suffix (even though some argue that it had DIM
functions, it was mostly an emphatic affix), which probably gave rise to the Latin DIM
-culus when combined with lo-. Thus, we may summarize saying that there were two
main DIMs in Latin, -ulus, DIM for nouns of 1st and 2nd declension, and culus, DIM for
the rest of the declensions, reflections of the IE equivalents -lo- and -ko-, respectively.
The other Latin DIM -ellus (or its allomorphs -illus/-ollus/-ullus according to root final
vowels) is believed to come from a morphonological variation of -ulus. In relation to the
use of this suffix, Gonzalez-Olle explains (which has been known already) that in root-
final liquids and nasals, assimilation takes place (of -root- Cs to the suffix ), after the
loss of its initial V, producing the variant -ellus, according to root final vowels (p. 178).
This would give us cases like *librelo > *librlo > *liberlo >libellus (Steriade, 1988).
However, whatever the specific origins of the LAT DIMs were, we may assert, as Pharies
(2002) did, that there were really only two LAT DIMs:-(c)ulus and (c)ellus (p. 366)
(four if culus is separated from ulus and cellus from ellus), at least in the form they
were really used in LAT. Let us see the development of these and other related suffixes
that gave rise to Spanish DIMs. References to specific phonological changes from Latin
to Spanish are based on historical accounts by Resnick (1981) and Penny (1993).
-ellus (Acc. -ellum > -ellos; the first is a short vowel, and Il is a geminate or long
). It has produced Spanish forms with an initial [e], its diphthong [ie], or the reduced
diphthong [i], whence the Spanish DIM -illo and its variants. In relation to the new final
vowel, there was a process of vocalic changes often implying a certain degree of vowel
(V) lowering or opening (u- o), according to description of common Latin-to-Spanish
phonological changes. Until the XIV century, the most frequent form was -iello, and for
approximately the next two centuries, -illo was the most common variant.
-(i)(c)alis (sometimes it had a short or long -i): Common Latin (LAT) stress
pattern rules causes the first -u of the suffix to be in unstressed position. When short
vowels (e.g., -i) were in this position, they were normally lost from LAT to Spanish
(SPN). It (plus rules mentioned above) would produce -iclo. Another common
phonological rule (i.e., -cl- -j-) gives us a new form -ijo with no clear relation to
vowel length. Vocalic changes such as -i e gave rise to -ejo, and in rare cases, when
the -i- was long, it might have produced -ijo. However, this last relationship described is
not totally clear. Pharies (2002) proposes more a semantic tie, rather than a phonological
one. The allomorph -ulus gave rise also to -ulo, which seems more a learned suffix,
unlike those discussed above.
-olus. (Late LAT) For this suffix, similar processes of vowel opening (u- o) and
vowel dipthongization (stressed short o-ue) take place. This gives us modern Spanish
DIMs such as uelo or uela. The former comes from the accusative masculine singular
form (-olum), which also implies a final-m deletion process, and the latter comes from its
feminine counterpart (-ola).
Other non-DIM Latin suffixes
-inus: It would clearly produce our modern -in(a/o) suffixes, according to the rules
mentioned above. However, we should keep in mind that this is not a LAT DIM suffix.
-ittus: (hypocoristic anthroponomy, which will be discussed under "Meaning
Changes"). This LAT suffix produces the widespread and productive modern -ito, even
though it should have produced -eto, which has also been observed as another DIM. This
may be related to the DIM series in -et (ete- eta- eto). Admittedly, this was not really a
LAT DIM since "Le latin ne possedaitpas de suffixe diminutive en tt-" (Hasselrot 1957:
9) (Latin did not have DIMs in -tt-). There is more on its original use and meaning
-*iccus: (o.u.o.)(Vulgar LAT but not LAT, Pharies, p. 306) This gave rise to -ico.
These are the most common historical accounts of the etymological origins of the
Spanish DIMs discussed in this paper. Let us now discuss some modem and historical
morpho-semantic restrictions on these suffixes.
DIM's formal grammar
Regarding Latin, there were clear grammar restrictions for the suffixes discussed
thus far. Some LAT suffixes are -ulus and -culus, which gave rise to several Spanish
suffixes such as -culo/-ejo. They attached to nouns and adjectives to form their DIMs.
The suffixes -ulus and -culus were in an allomorphic relation, probably morphologically
motivated. The morphological motivation was in relation to noun classes. -ulus was used
with nouns of the Ist and 2nd declension. Some examples mentioned by Pharies are:
porta ae >portula (1st declension) and servus i > servulus (2nd declension). On the
other hand, -culus was used for the rest of the declensions. For example,flos oris >
flosculus (3rd declension), manus us > manuscula (4th declension), and dies diei >
diecula (5th declension). In Latin, according to Varro, a well-known Latin grammarian,
suffixes such as -ulum (DIM in modern grammar) were used to mark one of the
differences among nouns (Kent, 1938). Varro, according to Kent's (1938) translation,
asserted that Latin nouns "are varied in form to show differences in those things of which
they are the names or to denote those things outside, of which they are not the names" (p.
381). Diminutives are used to mark differences with reference to the whole thing, not a
part of it. Plurality and smallness were the two main categories mentioned by Varro in
reference to this aspect of nouns. The examples of smallness are: homunculus, which
meant "manikin" (from homo, which meant "human being" plus the diminutive); and
capitulum, which meant "little head" (from caput, which meant "head" plus the
diminutive). Varro also comments on the possibility for recursiveness of DIMs in cases
where double and even triple DIMs can be found (Ex. Cista "box" cistula "little box"
- cistella "a smaller box" cistellula "very little box"). Such cases were used on a
sliding scale of greater diminution (i.e., 3 l's is smaller than 2 l's and 2 smaller than 1).
The other Latin DIM -ellus (etymon of Spanish DIM -illo), used to be an
allomorph (phonologically motivated) of-ulus for roots ending in a liquid or nasal
consonant. This happened especially when the first -u was lost (caused by the common
unstressed vowels deletion mentioned above) and then the final consonant of the root
(especially nasals and liquids) assimilated to the -1- of-ulus. Pharies mentions the
example of liberulus liberlus libellus (diminutivization of "book"). However, -ellus
started replacing -ulus regardless of phonetic environment (probably around the first two
centuries AD). The non allomorphic and productive -ellus used to have different
connotations depending on the lexical category of the base; with adjective bases it had an
attenuative function, and with nominal bases it was a diminutivizer and/or differentiator.
Pharies cites Gonzalez -Olle and Casado Velarde (1992) in relation to the
distribution of the three main Old Spanish DIMs: -uelo (pedazuelo "little piece"), -ejo
(portalejo "small portal"), and -i(e)llo (cosilla "little thing"). The first was used mostly
with root-final non-liquid sonorants; -ejo was used for root-final liquids, and -iello,
elsewhere. However, for the XV century, -i(e)llo started being used in phonetic contexts
of its allomorphs (which gave rise to the significant productivity of -illo during those
times.) Pharies cites Gonzalez-Olle's about two variants of -iello in Old Spanish: -iello
and -ciello. The variant with -c- (derived itself from -culus, and this from the IE suffixes
*ko+lo) is used for two-syllable bases ending in -e, iambic bases ending in -n/-r, and
monosyllables ending in consonants, which is similar to the distribution of -ito/-cito and
illo/-cillo in Modem Spanish (with few alterations and much dialectal variation;
DIM's original senses
Varro (1938), Hanssen (1952), Ettinger (1974) and Fruyt (1989) all show that Latin
made extensive use of DIMs, and these suffixes were normally associated with the idea
of littleness. The pure meaning of DIMs was, according to Malkiel (1989) "genuine
miniaturizing... [it] underlies the relationship of casa 'house' to casita 'small house', or
\eitia '(married) lady' to \ei)
(1977), in reference to the Hispanic Romance times (probably around the 9th century),
explained that the people then used to use DIMs (such as articiilus) in a concrete sense of
littleness of the base (p. 10). In the very first grammar written for any Romance
language, around Renaissance and Middle Spanish times, Nebrija (1492) considered the
DIM as one of the nine forms or differences for derived nouns, together with
patronymics, possessives, augmentatives, comparatives, denominatives, verbal,
participials, and adverbials. He defined derived DIMs as diminution of the original base:
ombre (man) ombrezillo (man+DIM) =pequeho ombre (little man); thus the basic
sense for early grammarians (Varro and Nebrija) is "little/small" (p. 93-94). Pattison
(1975) separated diminutive and augmentative suffixes from the rest of early Spanish
suffixes as "affective" or "appreciative". The rest he called "categorials" (p. 5), which
have basically a function of structural and logic order.
Changes in meaning and productivity
In general, there is a tendency of the type DIM > PEJ, not only in
Spanish, but it seems to be a universal tendency, at all times and in
all languages, as suggested by Pharies (in personal communication).
Pharies further clarifies that not only DIM>PEJ but also AUG>PEJ is
possible. Anything that is supposed to be big but is actually small
causes the DIM>PEJ process. On the other hand, anything that is
supposed to be small but is actually big causes the AUG>PEJ process.
This section emphasizes the former process (DIM>PEJ). In the case of
Spanish, much of the basic Latin sense of littleness in the diminutive
is lost in many of the Spanish suffixes; and in many of those cases, it
is the pejorative idea that prevails. Pharies (2002) thinks that this
is the normal trajectory or course of Latin DIMs that end up becoming Spanish
pejorative suffixes (p. 423), even though this is one of the attested functions in Latin, as
shown in the following example mentioned by Miller (2003) in personal interviews:
Graeculus = "lousy little Greek!" This is the most essential change in the semantics and
pragmatics of the diminutives. This can be observed in the following DIMs that now have
both diminutive and pejorative functions in Spanish: -ancho, -ejo, -ete, -ucho, and -
uncho. All these had a smallness sense in their Latin origins (e.g., -culus), but now this
littleness sense seems to be competing with pejorative connotations. Other suffixes exist
whose diminutive sense has been completely lost: -elo, and -ulo (from Latin -ellus and -
There are yet other less frequent phenomena in relation to the semantic
development of these Spanish DIMs: One is reanalysis, and the other is semantic
association. Some reanalysis is observed in -usculum, which becomes a suffix via a
wrong morphological analysis, which reanalyzes corpus-culum as corp-usculum (Pharies
2002: 507). Semantic extension or association is a plausible explanation for the origin of
the very productive DIM -ito, according to Pharies (2002), Gonzalez-Olle (1962), and
others. Originally, it is believed that -ittus was used with anthroponyms (especially as
nicknames for people). It is logical, semantically speaking, to see the connection between
these types of names and the diminutives since both share hypocoristic connotations. It is
not uncommon to see these types of names given to children, and this could have been
the explanation of its association with the sense of smallness, which probably was
extended later to inanimate entities. Also, Alonso (1937) considers more important for
the DIM the idea of hypocoristic and expressiveness than smallness. This, again, could
have been an even stronger reason to extend -ittus to its DIM behavior.
In relation to productivity, it is noticeable that many suffixes treated in this paper
are not productive. They are either productive only dialectally (like -oco in Chile), or
with very limited productivity (such as -ujo or -ucho), or with no productivity at all
(-uco, -ulo, -ueco, among others). There seem to be around four that are productive in
modern Spanish, namely -ejo, -ico, -illo, and -ito. Of these, -ejo is not really a
diminutive; it serves more pejorative and attenuating functions, as Pharies' example with
azulejo (bluish) shows. Thus, my study focuses on the productivity of the other three.
-ico. -ico was productive between the XV and the XVII centuries. Today it is
mostly in allomorphic variation with -ito in Caribbean Spanish (which has already been
discussed under "Formal Grammar"). Very few -ico diminutives are found in Spanish
literature of the 2nd half of the XVII century. In one example Pharies cites, there were
only five -ico DIMs found in Don Ramon de la Cruz's comedies. In the same literary
works, 206 times illo was found, and 1008 times for ito. Thus, obviously in this time,
-ito was the most productive DIM, followed by -illo, and finally -ico.
-illo. This suffix becomes generalized between the XIV and XV centuries. Before,
it was -iello, which was the Spanish DIMpar excellence. The suffix -illo is related to -
/llus, which is not frequent in Latin until the Post-Classical period (centuries I, II AD). At
that moment, -ellus started replacing -ulus, which was the most productive Latin DIM
until that time. Around the XV century, -illo competed hand-in-hand with -ito. However,
after this period, -ito wins the productivity race. Alvar and Pottier (1987) present -ito
even attaching to verbs (dormitar= "sleep+DIM").
-ito. The suffix -ito becomes more common during the XV century, and its
frequency and productivity continues to grow until today, when it represents the Spanish
DIMpar excellence. Until the XV century, -illo was apparently the most productive
Spanish DIM. However, Gonzalez Olle suggests a very interesting explanation regarding
this apparently long period of time between ittus and the appearance of -ito during the
12th and 13th centuries and then the long time before the great productivity of such affix
in the 15th century. Gonzalez Olle suggests that it may be the case that -ito was used
during these apparent times of absence but in uneducated or very informal environments,
which made it inappropriate for formal writings of the time. This is the reason this suffix
is not found in the documents that researchers normally have access to.
The productivity of -ito in the 15th century may be caused by two factors. One is the
linguistic pressure to use new expressive resources in the literature of the time, which
was in need of such a revival. The other factor, more at a societal dimension, was that
popular issues gained some prestige during this time caused by the social mobility (to
higher social status and positions) of members of the low social classes. Thus, the
language of such low social classes (among whose features were the uses of -ito),
similarly to other cultural manifestations, started acquiring more importance and respect,
to the point of including it in the literary works of the time. As can be seen, this is
obviously a sociolinguistic phenomenon, reflected here by the use of DIMs. This loss of
productivity of -illo is clear in the lexicalization of words ending in -illo, as Pharies
suggests. In Spanish, we now assume that words such as comilla (comma+ DIM =
quotation marks) bolsillo (bag+DIM = pocket) and others are monomorphemic, because
-ito attaches to them with a clear diminutive function. Admittedly, -illo still keeps some
productivity but at the dialectal level, as other DIMs.
This is often an adjectivizer and nominalizer that attaches to nouns. Pharies (2002)
recognizes two semantic senses in modem Spanish: a) Augmentative (it makes the base
N bigger than what is normal or convenient, or it intensifies the adjective base); and b)
Names the objects that can be used for hitting or the hit that can be given with such
objects. The suffix -azo comes from Latin -aceus, where it had the original function of
deriving adjectives of belonging from nouns. On the one hand, Pharies explains that the
augmentative sense appears first in the spoken Latin of the Western region, and the
"hitting" sense in the Spanish of the XV century, which represents a secondary evolution
of the augmentative sense. On the other hand, he recognizes that it is very difficult to
connect or to show the evolution from the original sense in Latin to the modern senses (of
augmentative nature, mostly). He agrees with Malkiel (1959) in the sense that this
evolution may represent a post-classical-Latin innovation. Malkiel, as cited by Pharies
(2002), believes that this evolution may be caused by a series of nouns which had this
suffix and referred to bulks or piles of something or big things. Another possibility for
this connection or semantic binding, using Jurafsky's (1996) terms, is that markers of
belonging or pertaining (such as Latin -aceus, English -ist, or Spanish -al or -ista)
normally imply not only a sense of belonging to the base but also having the qualities
expressed by the base in a characteristic way. We need to remember here that this is
probably the same type of evolution observed in Spanish -al. The sense of belonging is
such that the entity is characterized primarily by such a quality (A Latin Americanist, for
example, is a professional that focuses only or primarily in Latin American issues); in
this way, we can say that this quality is augmented.
My study does not elaborate further on the modern sense of -azo of nomen actions
(the naming of an action -in this case of "hitting") since it is not AUG. Some argue that
these two senses are so different that it may be necessary to see this as another
homophony in the language, which synchronically may be the case, but diachronically is
debatable. This second sense is not essentially or transparently an augmentative;
therefore, a discussion on the evolution of such sense is irrelevant in my study.
According to Pharies (2002), this suffix comes from Latin -dlis. Both in Latin and
in most uses in Modern Spanish, this suffix is an adjectivizer attached to nouns. In
Spanish, this has three main connotations:
* Expression of belonging or a similar relationship.
* Place naming, especially where items like plants are abundant.
In the second sense, it normally has the form of a nominalized adjective. Pharies
cites examples in both senses. Sense 'a' can be found in words such as: annual (annual)
(belonging to aho or "year"), internal (winter+the suffix al) meaning "relative to winter".
For the second sense, Pharies cites naranjal (orange+the suffix al) meaning a place with
many orange trees, and maizal (corn+the suffix al) meaning a piece of land where they
Even though Pharies does not mention the augmentative function of -al, in some
dialects of Modern American Spanish, this suffix is used with such functions, meaning
"much" or "a big amount of". This is in line with the second sense of the affix but also
with the third sense. We can clearly see how these senses mentioned above gave origin to
the idea of "much" or "big". Both give the idea of a significant number of something.
Now we can see examples such as pantanal (mud+a/) and dineral (money+a/). The latter
example was cited by Pharies as a collectiveness marker. In this sense, there may be a
clear contrast with the DIM; dinerito (a little money) vs. dineral (a lot of money). From
here, we can see how in some dialects of of modem Spanish, this suffix may be used as
Pharies (2002) observes two distinct origins and functions of this suffix: The
augmentative and the nomina actions (naming of an action). The AUG -6n comes from
Latin -6, onis, which in Latin attaches to nouns and verbs to designate people that are
particularly characterized by some action, feature or habit, undesired generally. In
Spanish, we still see this type of use in words such as llor6n, chill6n (cry+6n = person
that cries too much) andfrent6n (forehead+6n= person with a big forehead). Thus, this
type of function is still an AUG function. The nomina actions -6n comes from its Latin
counterpart -(i)o, onis. Even though different, these two origins and functions end up
influencing each other, as explained by Pharies. Finally, Pharies and Gonzalez Olle
(1962) recognize the diminutive function of this AUG, shown in the present analysis.
Pharies (2002) also recognizes the pejorative function of this AUG. Pharies suggests that
certain actions are considered negative if they are intensified. For example, it is bad to
sleep too much, whereas it is not necessarily bad to read or study too much. If the suffix
is added to verbs like to sleep, then that creates a pejorative connotation, which is not the
case with verbs such as to breathe or to study. This is in line with the findings in my
Regarding the second sense of 6n (nomina actionss, Pharies explains that
different authors believe that this second sense in many cases (e.g., rascaz6n
"scratch+6n", hartaz6n "swallow/eat + 6n") is also interpreted as an augmentative, and
that in some cases it is almost impossible to separate these two uses (baj6n "pull
down+6n", visit6n "to visit+ 6n"). In light of this discussion, many such uses of -n in
the data under scrutiny here have been labeled as augmentatives, even though we have to
recognize the different origins and uses.
According to Gooch (1967), -ote became very productive in the 19th and 20th
centuries. Pharies (2002) argues for the hypothesis of the Catalan origin of such suffix
instead of the French origin, as other authors have suggested. It seems that all the authors
agree that there is no Latin etymon for Spanisn -ote. The ultimate origin may be -ottus, a
non-Latin hypocoristic and probably analogically related to ittu (etymon of Spanish -
ito), originally. Many authors agree in the direct genetic relationship of this -ottus,
Vulgar Latin suffix of foreign origin, and the Catalan ot. Pharies rejects the French
hypothesis because the bases of the few -ote words that were borrowed from French into
Spanish were not transparent to the Spanish speakers of the time, unlike the Catalan
loans. The Catalan bases for the suffix -ot were indeed understood by Spanish speakers
(e.g., animalot), and the suffix in Catalan has the same evaluative function as the Spanish
-ote today. Apart from this, there is a parallel borrowing from Catalan to Spanish of the
DIM -ete. All these arguments, according to Pharies, are enough to sustain the Catalan
hypothesis. The final vowel (e) in -ote is paragogic to satisfy phonotactic constraints in
Spanish. It is not uncommon in Spanish since there are common schwa insertion
processes (e.g., speak espeak) that satisfy Spanish phonotactics in foreign words;
probably a simple case of phonological adaptation.
Because of this origin and the modern uses of-ote, Pharies labels this suffix as an
augmentative-pejorative, which was essentially the use in the Catalan etymon; except in
North East Catalan, where it was a diminutive-pejorative suffix. There seems to be
primarily a pejorative and an augmentative use, and secondly a diminutive function.
Pharies indicates that this uncertainty between DIM and AUG, always with the
pejorative, is also observed in Spanish, even though 95% of the cases it is the AUG the
one that is realized (p. 456). Some have argued that pejorative connotations may be
inherent in augmentative functions. Pharies cites Latorre (1956) who thinks that the AUG
has the particular property of communicating a mocking, joke, and cartoon-like tone.
Latorre thinks that what is great is never expressed via AUGs, but only what is
extraordinarily out of shape or proportions (modern -azo, an exception).
This should be taken with caution. What is undesired, funny or joke in one culture
may be the opposite in another. Admittedly, as Pharies (in personal communication)
suggests, anything that is smaller or bigger than the norm can inspire negative or
pejorative connotations because it is too small or too big. Thus, if bigness is fine or
positively viewed, then the AUG would not create pejorative connotations (and if
smallness is positive, then DIMs would not produce pejoration). For some, what is
extraordinarily big may be what is desired and smallness may be undesired or funny; in
which case, DIMs would very likely become pejoratives. AUG -azo in modem Spanish,
for example, may mean "great" (car+azo).
Regarding this suffix, Nebrija asserted that the Castilian (Spanish) language of his
time did not really have superlatives (as explained by Alvar and Pottier, 1987: 378).
Then, we may ask, technically, does modern Spanish have SUPERLs? If superlative is
considered the grammatical function of "most" as observed in comparative adjectives
such as positive (good), comparative (better), and superlative (best), then as in Nebrij a's
times, -isimo is not really a superlative (at least not one with the meaning of "the most")
in Modern Spanish. Unlike Latin, whose SUPERL meant "most" and "very", Modern
Spanish SUPERL means mostly "very". There are still a few cases with relative
superlative functions but mostly for honorific reasons and mostly in learned words
(Excelentisimo .ilr Presidente or "Excellent+SUPERL Mr. President", for example).
However, this suffix does exist in modern Spanish, which is apparently a difference
between contemporary Spanish and Nebrija's Spanish, in which this suffix was not part
of Spanish morphology. SUPERL does exist in contemporary Spanish but with a
somewhat different connotation from its ancestor or etymon in Latin. If SUPERL simply
means "very", as my study and many others suggest, then this function is performed in
Spanish both analytically/periphrastically and synthetically. Synthetically is performed
via prefixes such as re-, requete, and others, as Alvar and Pottier (1987) suggest and via
the SUPERL suffix -isimo, which these authors do not account for.
Jornvig (1962) indicated that this synthetic elative, in its origin, represented a
learned suffix of late introduction to Spanish (from Latin). This Latinism in Spanish was
caused by the Latin Renaissance, very influential during the John II of Castile reign
(1406-1454). Because of this, many Latinisms invaded the Spanish language; and this
SUPERL was one of them. Jornvig cites the XV century as the first time in which this
suffix appeared in Spanish. However, the spread of this suffix did not occur significantly
until the second half of the XVI century, when it was also commonly observed in
informal settings; thus, not more a learned suffix necessarily. He criticized the theory that
the origin of this suffix in Spanish was due to the Italian language. He concluded that the
appearance in Spanish emerges from its Latin etymon, as mentioned before, but its fast
spread later on was definitely influenced by Italian but also by the influence of Catholic
preachers in Spain, who used this suffix extensively. Recall that one of the most powerful
tools of conquest and colonization of the Roman Empire was the Catholic Church.
Catholic preaching and teaching was a constant in Roman Empire times, thus, listening to
Catholic preachers was a very common activity in the cities where this Empire ruled.
Therefore, Catholic preachers' language may have easily influenced the language of their
audience. These preachers were not only Romans (the ruling class) but also those who
spoke in the name of God, so they may have enjoyed an important reputation for a long
time. It is not strange, then, that this elite class of "good, ruling, holy and powerful'
preachers had linguistically influenced the masses.
Evaluativeness, as a primitive linguistic category, involves both semantic features
(such as "littleness"/"bigness", "approximation", "insignificance", "intensification" and
the like) and pragmatic features (such as "attenuation", "admiration", "endearment",
"modesty", and others related.) This is a very common feature cross-linguistically, which
is manifested in child-related language and language acquisition observations. Thus,
categories such as littleness, childness, and endearment may have been elevated to a
postulate. Languages often mark this linguistic category via diminutives, augmentatives,
pejoratives, and other related morphemes. Evaluatives may have different morpho-
syntactic behaviors from language to language, even though they tend to preserve similar
semantic and pragmatic connotations crosslinguistically. Marking this feature analytically
(as in English, for example), or synthetically (as in Spanish or Fula) depends on either the
history of the language or the degree of significance of such a feature in a particular
culture and cognitive aspects. Evaluative morphology, the main focus of this paper, is
precisely the marking of this feature at the synthetic level in some languages. However,
evaluativeness may imply cross-linguistic semantic and pragmatic connotations at
different grammar levels. This category crosses boundaries of grammar levels and
1 References on languages cited: for Japanese, Suzuki, R. (1999). Language socialization
through morphology: The affective suffix -CHAU, in Journal ofPragmatics 31, pp.
1423-41; for Greek, Alexopoulos, E. (1994). Use of Diminutives and Augmentatives in
Modem Greek. In Amsterdam Studies in the Theory and History of Linguistic Science,
IV. pp. 283-88. Irene Philippaki-Warburton (ed.). Amsterdam: Benjamins; for Swahili,
Frankl, P.J., and Omar, Y. (1994). Diminutives and Insignificance, Augmentatives and
"Monstrosity": Examples of Class Reassignment in Swahili, in S.u,,l African of African
Languages, 14:3, pp. 113-116; for Dutch: Robinson, O. (1980). Dutch Diminutives Over
Easy. In Dutch Studies, 4, pp. 139-157; and for German: Schneider, K. (1993).
Pragmagrammar and the Case of German Diminutives. In Wieviel Grammatik braucht
der Mensch? pp. 158-73. Theo Harden (Ed.) Munich: ludicium.
A complete analysis of Spanish EVALs' functions and uses must start with
fundamental semantic considerations. This chapter answers two interrelated questions: 1)
What are the propositional meanings of Spanish EVALs?, and 2) How can we account
for the diversity of meanings and uses? This discussion takes us to the core semantics of
Spanish EVALs and to an explanation of their polysemy, which is an important
characteristic of this type of Spanish morphology (unlike other Spanish morphological
processes). This chapter first addresses some general and fundamental semantic issues
regarding the model adopted here and then focuses on each of the three Spanish EVALs,
in the following order, according to their degree of polysemy: DIMs, AUGs, and
Semantic vs. Pragmatic Polysemy
We can look at the case of polysemy observed in Spanish EVALs, especially DIMs
and AUGs, from the perspective of cognitive semantics, since it has to do with the way
speakers of Spanish organize this type of concept or category: evaluativeness. Here
polysemy is defined, according to Taylor (2003) and from a cognitive semantic
perspective, as the association of two or more related meanings with a single
phonological form. The term polysemy normally refers to semantic senses, but it can
obviously extend to pragmatic forces. In the case of Spanish EVALs, we observe a type
of polysemy more at the pragmatic level, which can be labeled as "polypragmy". When
we say that Spanish DIMs are "polypragmous", then, we mean that DIMs have many
(connected but different) pragmatic functions, which may be distinct from pure semantic
polysemy. Ambadiang (1997) suggests that the very complex morphology of DIM
formation (e.g., various allomorphs, various options for the same base) may be due
precisely to the fact that diminutivized words may receive multiple interpretations.
Reynoso (2002) referred to this polysemy as "uno de los aspects mds caraterizadores
del uso del diminutive" (one of the most distinctive features of the use of DIMs) (2002:
937). For example, she referred to at least seven different connotations of this suffix (all
of which were observed in the data analyzed as shown in the pragmatic analysis
presented in the next chapter): affection ([dear]), pej oration, littleness, intensification
("very"), euphemism, emphasis (a type of intensification in the present analysis), and
subjective expressive diminution of base identity.
This type of polysemy of EVALs (especially DIMs and AUGs) is complementary
in Nerlich and Clarke's (2003) terms, since all the various senses analyzed here are
connected. This constitutes a fundamental principle for this chapter. This principle goes
back to Wittgenstein's (1974) "family resemblances" used in prototype theory.
Wittgenstein indicated that
What a concept-word indicates is certainly a kinship between objects, but that
kinship need not be the sharing of a common property or a constituent. It may
connect the objects like the links of a chain, so that one is linked to another by
intermediary links. Two neighboring members may have common features and be
similar to each other, while distant ones belong to the same family without any
longer having anything in common. The relations between the members of a
concept may be set up by the sharing of features which show up in the family of the
concept, crossing and overlapping in very complicated ways. (1974: 35)
This may explain why Jaeggli (1980) refers to diminutivization (and EVALs in general)
as one of the most productive morphological processes of Spanish.
Two distinct types of functions or connotations of Spanish EVALs clearly exist:
semantic and pragmatic ones. These should be kept separate even though they are related.
Reynoso (2002), in one of the most recent semantic-pragmatic accounts of DIMs in the
Spanish language, makes a difference between semantics (referential) and pragmatics
(non-referential) also, but she includes both under the general cover term of semantics.
Pure semantic aspects in her analysis are under the semantic-referential category; the
pragmatic aspects are under the semantic-pragmatic category.
This is precisely where Reynoso's study and my study converge but diverge at the
same time. Like Reynoso's, my study shows both types of effects. However,
theoretically, the non-referential aspects are not semantic here precisely because of their
non-referential nature. This difference may be more a conceptual that a practical one,
however. My study discusses the neutral sense ([little]) mostly in this chapter, where the
semantic connections of DIMs are shown to grow out of this basic littleness notion. In the
pragmatic chapter, this neutral or non-referential use is briefly discussed, with examples
from the data. The majority of the next chapter focuses, however, on non-referential or
pragmatic functions, which synchronically and functionally have little or nothing to do
with the core sense of littleness. That is one of the main reasons these two areas are
conceptually and organizationally kept separate in my study.
One more important difference of my study to Reynoso's analysis is that my study
places the DIM within a broader study of Spanish evaluatives. In this way, we can
observe some important aspects that are true not only to DIMs but also to Spanish
EVALs in general. DIMs are just part of a broader phenomenon: Spanish evaluativeness.
Admittedly, it is at times very difficult to separate semantic from pragmatic
functions. However, there are many instances where it is very clear that a non-referential
(pragmatic) use is at play. We will see later that context is a crucial aspect in this respect.
For example, DIMs are normally associated with the meaning of littleness but also with
endearment (Jaeggli, 1980; Hualde et al., 2001). The latter is a more pragmatic function
whereas the former a more semantic one. Based on different contexts, there are many
other pragmatic functions of such affixes: irony, euphemism, intensification, and
augmentativization, among others. Now, the main question is: What is the connection, if
any, between the pure basic semantic denotation and these other pragmatic functions of
the DIM; and the other EVALs? The next section discusses these issues.
Cognitive Semantic Model: Radial Categories
One possible explanation for the multiplicity of pragmatic functions is extension or
association (some may also argue that we see semantic shift as well, at least in the case of
the endearment notion). This present analysis shows, based on the theoretical framework
used here (Jurafsky, 1996), that all the pragmatic functions observed grow out of or
emerge from the basic littlenesss" notion. This chapter is, in essence, a cognitive
semantic one with the purpose of answering the first research question of my study: How
can we link the diverse meanings and uses of Spanish EVALs? Even though we need to
refer to many pragmatic features, they all are connected to a basic semantic-referential
Jurafsky's model builds on Lakoff s (1987) radial categories, the first to formally
and overtly apply Rosch's (1983) psychological model of prototypes to linguistics, and
cognitive linguistics in particular. We also discussed above how this model emerges from
the field of philosophy in the works of Wittgenstein (1974). As mentioned, one of
Wittgenstein's major findings is that some categories do not express a single concept or
meaning. Categories may be instead characterized by family resemblances (or related
features). These resemblances are widely shared among the different nodes or members
of this semantic mapping or network (or "category members") in an overlapping fashion,
such that no one feature is common to all.
This framework suggests that these members have an internal structure. There are
members that are typical, there are others that are exemplary, and yet there may be others
that are anomalous. In many cases, Spanish EVALs for example, we see what Rosch
(1983) and Lakoff (1987) call a "radial structure", since there are core meanings. In
Rosch's prototypical model, a prototype (an element in a category used to represent the
category as a whole) is used as a cognitive construction to perform some kind of
reasoning. It basically functions as a cognitive reference point. The central subcategory
(e.g., littleness for DIMs, as suggested below) of this network provided the basis for
extending the category in new ways and for defining variations. Lakoff suggests that at
the cognitive root or core of the formation of categories, we find image schemata and
their metaphorical tokens. That is the reason one of the most useful cognitive tools in this
type of semantic mapping are general extension mechanisms such as the metaphor or
In summary, the cognitive semantic approach applied in my study builds upon
Wittgenstein's notion of family resemblance categories based on a complicated network
of similarities or associations that may overlap and criss-cross. This criss-crossing allows
in category members with hardly any element in common, but it is crucial that each
overlap with certain other members of the category. Such is the case of seemingly
opposing connotations as endearment vs. pej oration and littlenesss" vs. "very" (in the
case of DIMs), attenuation vs. intensification (in the case of AUGs), and formality vs.
informality (in the case of SUPERLs). Wittgenstein (1974) proposed then a certain level
of tolerance for fuzzy boundaries or blurred edges. Rosch (1983) and Lakoff (1987), on
the other hand, observed a certain automatic and unconscious tendency in humans to
perceive categorically and to base those categories on prototypical examples. Radial
categories are composed of chained elements that radiate out from the central
(prototypical) examples, and this chaining is frequently a matter of metonymic links as
well as shared features.
Cruse (1986) observes similar principles, more from a lexical semantic perspective.
In Cruse's model, related meanings of a word blend fluidly into one another, and
different aspects of a word's meaning may be emphasized or de-emphasized depending
on the context in which it occurs. This framework has even been applied to computer
models for Natural Language Processing (NLP) in cases of polysemy. Dolan et al. (2000)
for example, show how their computer model's (MindNEt) "processing of the discrete
senses in machine-readable dictionaries yields a representation of lexical semantics with
the continuous properties of Cruse's model" (2000: 182). This all emerged from many
instances of polysemy and the practical task of word sense disambiguation in natural
language processing. Thus, the principles my study adheres to have been useful in
philosophy, cognitive sciences, computer sciences, and linguistics. The following is an
example of its linguistic application, our main concern.
Jurafsky's Model: DIM's Radial Category: Let us now consider more
specifically the issues concerning evaluative morphology and the Radial Category Model
described above. The only EVAL that has received more formal consideration in this
approach is the DIM. Jurafsky's (1996) Radial Category Model argues that despite the
crucial dependence of synchronic meaning on both historical and cognitive context,
researchers have traditionally used different tools for capturing synchronic and diachronic
generalizations in modeling a complex semantic category like the diminutive. In the case
of the diminutive, this is partly caused by the extraordinary, often contradictory range of
senses synchronically (small size, affection, approximation, intensification, female
gender), and the difficulty of proposing a coherent historical reconstruction for these
Jurafsky (1996) synchronically tries to explain the varied and contradictory senses
of the diminutive. Diachronically, the radial category acts as a kind of archaeology of
meaning, capturing the generalizations of the classic mechanisms of semantic change
(metaphor, abstraction and inference). He claims to have compared DIMs and their
origins in more than 60 languages, particularly in Indo-European where the theory
suggests a new reconstruction of the proto-semantics of the PIE suffix *-ko-. Jurafsky
shares with Lakoff the fundamental intuition that the body is a central site in grounding
interpretations of the world, including those that involve power and dominance issues.
Thus, much of his reasoning in the linking chains for DIMs lies on this body-world
In summary, Jurafsky's (1996) Radial Category Model accounts for both the
synchronically and diachronically diverse semantics of the diminutive. This is a type of
structured polysemy, which clearly binds the various DIM senses. From a synchronic
approach (the main focus of my study), the model accounts for the various and apparently
contradictory senses of the diminutive, for example in cases when a DIM has AUG
functions (see the "Intensifying DIM" below). Historically, Jurafsky also presents a
binding of these diverse senses with a common original source. He concludes that "the
origins of the diminutive cross-linguistically lie in words semantically or pragmatically
linked to children" (1996: 533). The following is a summary of the diagram Jurafsky
proposed. In his diagram, numbers indicate the sequential order or semantic associations.
For example, he assigns 1 to the [child] sense, and 2 to [affection]. This simply means
that first is the sense of childness (the core), and then it expands to have the notion of
affection (2) and then the idea of smallness (3). From the sense of smallness, several
other connotations emerge, again in order of sequence: contempt, female, resemblance,
and approximation. These other nodes also may originate other ideas or connotations. For
example, the resemblance node gives rise to the idea of imitation, and the approximation
node gives rise to the function of hedging. Other functions of DIMs also appear in
Jurafsky's diagram, but the ones summarized here are the ones found in the data.
The various uses or functions in the data are explained based on this model.
Following Jurafsky, all the functions of DIMs may be bound to some more general and
common sources. The innovations of my study, in the application of this model, are the
* Application to a synchronic naturalistic data study in monolingual Spanish
* Accounting for other uses of the DIM not accounted for by Jurafsky; namely, the
euphemistic, ironic, and commiserating functions.
* Extension to other Spanish EVALs (not only to DIMs, as Jurafsky did).
* Emphasis on the pragmatic aspect of such semantic binding or association.
One important consideration that distinguishes my study from Jurafsky (1996) has
to do with the universality of such a model. Even though the categories analyzed here
(e.g., DIMs) may be universal, it is very hard to use this type of model (Radial
Categories) to explain the connections of diverse uses of DIMs made in all cultures
(where DIMs constitute a linguistic category) and languages. Instead of universalizing
such proposal, my study uses this model to describe potential links between the core
sense of littleness and the other uses or meanings of DIMs (and other EVALs) within a
particular context: Spanish-speaking groups observed in the data. Thus, in my study, it is
not necessarily a theoretical explanatory model but a potential descriptive tool of some
specific linguistic behavior in a particular speech community.
Core Sense of DIMs
The goal in this section is to explore and propose a plausible definition of the basic
semantics of DIMs. In other words, following the Radial Category Model, it is an attempt
to answer the following question: Which sense is the one that connects, somehow, all the
other senses of this suffix (at least in the data analyzed here)? Voeykova (1998) indicates
that there are basically two formal hypotheses about the basic semantics of DIMs that
may have served as a base for the many other variations in meaning at present:
"smallness" (Dressler and Merlini-Barbaresi, 1994; Ravid, 1998) or "childness"
(Wierzbicka, 1984; Jurafsky, 1996). Because of the arguments below, my study rejects
the "childness" hypothesis and favors the more traditional perspective; "smallness", but
with some modification. My study proposes that the notion or concept of littlenesss", and
not necessarily "smallness", is what constitutes the core sense of DIMs in general, and
Spanish DIMs in particular (assuming littlenesss" is a broader term than "smallness";
"small" seems to refer only to size, but "little" also refers to amount, for example).
Below, let us see first the inconsistencies in which Jurafsky fails when trying to assign
the "childness" sense to the nucleus of this semantic category of DIM.
Jurafsky suggests that an acceptable semantic analysis of DIMs cannot rely on just
a single abstract concept based on "small". The reason for this, he argues, is that we
would need many metaphorical, inferential, or abstractive extensions in order for "small"
to be able to model senses such as individuating, imitation or exactness functions.
According to Jurafsky, the notion of "small" does not connect whatsoever with words
such as Spanish boquete ("hole") derived from boca ("mouth"), since a boquete indeed
can be larger than a boca. He furthermore cites Dressler and Merlini-Barbaresi (1994) to
indicate that the diminutive cannot simply be listed in the lexicon/grammar with the
"smallness" abstract meaning only, and that other senses are derived by contextually-
based inferences when the diminutive is used. Dressier, Merlini-Barbaresi and Jurafsky
point out that if it were the case ("small" as the core), then we would expect these same
inferences for words for "small" in each language (i.e., Italian piccolo should behave like
the diminutive -ino); this does not occur. They concluded that there must be some
additional, complex, lexicalized meanings specific to this type of suffix.
However, there are some inconsistencies in the previous arguments. First,
metaphorical and inferential extensions are an essential part of the cognitive model
Jurafsky is applying. How can they then be left out? Why not link the sense "little" to
other senses via metaphors or inferences? Furthermore, any link of the "child" sense with
other connotations will also need metaphoric and inferential abstractions. If we do not
apply this type of semantic extension, we will not be able to assign any concept to a core
sense of any category, and the model of Radial Category turns useless. Second, the only
concrete example he showed to argue against littlenesss" as the basic sense is the boquete
example. However, this word can indeed be interpreted with the approximation or
imitation function Jurafsky mentioned before. A boquete is "sort of a mouth", but not
really a mouth. A boquete is missing many features to make it a real mouth, and that is
why it is just "a little bit like a mouth". It, in fact, can be easily connected with the
littlenesss" concept. Finally, it would be even more difficult and much more abstract to
connect this DIM with the concept of "childness". Third, if we were to adhere to Dressler
and Merlini-Barbaresi's (1994) argument above, then again we could not assign to DIMs
any core sense, since the equivalent lexical items for this type of morphological marker
can always behave differently. For example, the word for "child" (Jurafsky's suggestion)
does not behave the same as the morpheme ito in Spanish.
The following is the only section in Jurafsky's paper where some argumentation is
given for the choice of childness: "My tentative conclusion is that the origin of the
morphological diminutive is the sense Child. We show that in almost every case in which
a historical origin can be determined for a diminutive morpheme, the source was either
semantically related to Child (e.g., a word meaning "child" or "son"), or pragmatically
related to Child (e.g., a hypocoristic suffix on names)" (p. 562). However, without
looking at the specific examples, these words or morphemes that probably meant "child",
"son" or hypocoristics (probably in ancestors and proto-languages), could also be
interpreted as "little". For example: Victorcito can mean "little Victor", "Victor's son",
"child Victor" or "Victor Jr.". What tools were used to determine that the meaning was
"child" and not "little" is not clear in Jurafsky's study. He also mentioned hypocoristics
and toponyms as sources of DIMs in many languages, especially Indo-European
languages. However, that does not show the original sense of "child". On the contrary,
how can we explain that names for places (e.g., cities, towns, regions) gave origin to the
notion of"childness"? It is more plausible to associate place names with the notion of
littlenesss" because one of the inherent properties in places is size; which may be related
to littlenesss" more than to "childness".
At times, there is much ambiguity regarding his proposal for a core sense. For
example: "our examination of the IE data suggests a completely different reconstruction,
in which 'child' and not 'related-to' is the proto-semantics of *-ko-, and the various
approximation and related-to senses are extensions of this core small/child? sense" (bold
and question mark added; Jurafsky, 1996: 565). One possible reason for this ambiguity is
precisely the apparent contradiction that his own data showed. For example, he shows the
different PIE's daughter languages he analyzed to reconstruct the semantics of *-ko- (p.
566). In the section that he categorizes as the "SMALL/CHILD" senses, of the eleven
examples he gave, nine have clearly the "small" or "little" sense and only two have
apparently the "child/son" sense. Even more, in the only case where this type of DIM
does not have a modified base or noun, it simply meant "small" (not "child"). It is
important to note here, however, that in Spanish (and other Romance languages), it is not
the PIE suffix ko- but -lo- the one that constitutes the etymon for many Spanish DIMs
today (Pharies, 2002). This ambivalence of Jurafsky's examples probably constitutes the
main reason for him to recognize, at times, not one but two central senses for DIMs: "...
the central senses Child and Small. Every diminutive in our database has either the Child
or Small sense" (p. 561).
One more inconsistency shown in Jurafsky (1996) was in relation to the analytical
or periphrastic DIMs (adjectives such as "little" or "small"). Jurafsky observed thatpetit
in French is grammaticalizing as a diminutive, and the former diminutive suffix -ette is
disappearing. In Spanish and Italian (with very productive morphological DIMs), these
analytical adjectives (pequeho and piccolo respectively) are not common. The French
adjective for "little/small" is more common than the DIM suffix, whereas in Spanish it is
the opposite. Jurafsky concluded, literally, that "for at least these periphrastic
diminutives, then, the original sense of the diminutive seems to be Small, and not Child.
Further study is needed to examine the origins and development of these periphrastics."
(p. 569). It is quite inconsistent and inelegant to conclude that for DIM affixes the
original sense is "child", but for DIM adjectives, the original sense is "little/small".
Another study by Voeykova (1998) on Russian DIMs has also criticized Jurafsky.
Voeykova observed that the "smallness" sense is very important for the child when
acquiring DIMs in Russian. Even though both Voeykova (1998) and my study recognize
the value of the arguments favoring the "child" meaning, it is necessary to point out that
"this meaning is less relevant for the acquisition of language by a child in comparison to
"smallness", since it demands a very high degree of abstraction in all cases ... about
inanimate objects" (Voeykova, 1998: 112).
Just by considering the inconsistencies above, we can see the need for another
proposal. As mentioned above, littlenesss" seems to fit plausibly in many more examples
than "childness". Jurafsky himself argued that "the diminutive function (for the purposes
of this paper defined as any morphological device which means at least "small") is
among the grammatical primitives which seems to occur universally or near-universally"
(1996:534). "Littleness" fits more within the category of "grammatical primitives" than
"childness"; all objects may be defined within a littlenesss" range, and not necessarily
within a "childness" range. If one of these two features is the core of the DIM, then, it
should be the most primitive category: littlenesss". Even Lakoff, in an important point of
departure for Jurafsky, describes basic-level status solely in terms of objects and
recognizes that the relatively subjective notion of littlenesss" is at the center of this
conceptual category of "diminutiveness", in part because it has many of the
characteristics and the attractiveness of basic-level terms (fundamental in body-world
connections). It is easy to use, it is the most contextually neutral term, and it is the first to
enter most readers' lexicons.
The "childness" sense, for example, finds hardly any relation to the DIM functions
in Cantonese mentioned in Jurafsky (1996). On the contrary, littlenesss" seems to be a
sense that better accounts for such diverse uses. The partitive function can be interpreted
as "a little of that"; resemblance with larger object can be interpreted as "sort of a little
X"; as a marker of approximation, this Cantonese DIM (marked by tone) may be
interpreted as "reddish or a little red"; the pragmatic hedge function may be interpreted as
"a little favor" instead of "a favor"; the DIM marker of marginalized women can be
interpreted as "just a little of a woman, not much (of a woman)". Probably because of
this, Cantonese DIMs also mark the [female] feature. Linking all these to a primitive
category of littlenesss" seems more plausible than linking those to the "childness"
function. He mentions DIMs in languages such as Nahuatl, Ojibwa, Yiddish, Ewe,
Londo, Hungarian, Boro, Kayah, Khase and Tboli. Interestingly, in all these languages
the core sense is "little", as he himself showed in the translations.
This proposal (of littlenesss" as the core sense) shows more evidence for the
common tendency in semantic extension and change. The reason for this is that the radial
category for the diminutive extends the central physical domain of size to the other non-
physical domains (a common trend in this type of semantic association) of gender, social
power and others. In this way, it also provides further widespread evidence about the
unidirectionality of semantic change from the physical to the social and conceptual
domains discussed above.
In an Amharic example given by Jurafsky, the case of "this man-teacher" vs. "this
woman-teacher", where the only morphemic difference is the inclusion of the feminine
marker with DIM functions in the second expression, is very revealing. There is a more
direct connection between the idea of "this little teacher" and "this woman-teacher" than
between "this child-teacher" and "this woman-teacher". Probably because of the inferior
status given to women in many societies, the metaphoric connection of "woman
professional" = "little professional" is at play. It is important to note here that the same
difference is observed in minimal pairs such as "book" and "booklet" in Amharic, where
the DIM form makes "book" into a "booklet" (a little book; not necessarily a child book).
Many explanations about different DIMs in different languages and different uses
of DIMs start from the littlenes sense in Jurafsky's explanations. The "word chotto,
whose central (and historically prior) meaning is something like 'a little', functions like a
diminutive in Japanese" (p. 557). His lambda-abstraction examples also emerge from this
core sense. "For the diminutive, this process takes the original concept small(x), which
has the meaning smaller than the prototypical exemplar x on the scale of size, and
lambda-abstracting it to lambda(y)(smaller than the prototypical exemplar x on the scale
y)" (1996: 555). The pragmatic hedges and politeness-marking functions he discussed are
also based on this littleness concept. He mentions, for example, that "in a number of
languages, including Tamil and Malagasy, this use of diminutives for politeness is even
more grammaticalized, and the word for 'a little' functions generally like English
'please'" (1996: 558). All his partitive and exactness examples find an explanation that
connects to the basic idea of littlenesss".
In his examples of animal offspring, we can see that all of those DIMs can also be
translated as "little". For example, "a bear cub" can be a "little bear". However, not all of
them can be translated as "child" or "son". For example, in "chicken" and "chick" (the
latter had a DIM in the language cited), can we translate "chicken+DIM" as "the child of
chicken", or better and simply as "little chicken" or "chick"? The latter seems more
Furthermore, since metaphoric speech constitutes an essential element in the model
discussed here, we need to at least superficially consider which of the two senses ([little]
or [child]) fits more in what we know about metaphoric thinking. As it happens, the
notion of size (in which littlenesss" belongs) constitutes the base for many metaphors.
The reason for this is that it has basic physiognomic and perceptual properties. In fact,
Seitz (2001) distinguishes four key aspects of early or primary metaphors: perceptual
(e.g., color, shape, size), enactive (movement, action), physiognomic (i.e., visual-
affective), and cross-modal or synesthetic experiences. It is well established that humans
exploit perceptual features such as shape, color, size and others when performing
metaphoric thinking (Seitz, 1997). Even though size is not really a physiognomic feature
(like facial features), it shares with physiognomy the visual property.
Children can exploit the physiognomic (i.e., visual-affective) basis of metaphor
(Seitz & Beilin, 1987). Indeed, physiognomic perception has been well studied and there
is an extensive literature (see Seitz & Beilin, 1987, for a review and empirical analysis of
the physiognomic basis of metaphor). This perception may be bodily-based (motion,
gesture, or bodily action; see Seitz, 2000 for a more elaborate discussion on this
cognition-perception link). In his considerations in the psychology of visual perception,
Arnheim (1988) indicates that people perceive a building, for example, because of spatial
properties (lines, volume, size) that are distinctive in the visual dynamics of such solid
structure, which is the perceived form. Thus, size littlenesss) serves better as a base for
metaphoric chaining (in our model of radial categories) than age-related properties
(childness) since it shows more basic perceptual primitives. Other studies directly
connect size with symbolism and physiognomic stimuli (Ultan, 1970; Lindauer, 1988).
Finally, and probably more importantly in support of the littlenesss" core proposal,
Lakoff (1987) listed five criteria for determining the central sense (what he also called
"proto-scene", derived from spatial scenes) of any category: 1) earliest attested meaning,
2) predominance in the semantic network, 3) use in composite forms, 4) relations to other
spatial particles (contrast sets), and 5) predictability of other senses in the network.
"Littleness" seems to be the earliest attested meaning for DIMs, even in Jurafsky's
examples. This notion of littlenesss" dominates most links among senses (more than the
"childness" notion), even though it is not necessarily the most common sense in modem
Spanish uses of DIMs. Criteria number 3 does not support one proposal or the other.
Criterion number 4 above is critical in Spanish morphological evaluativeness. As shown
before, DIMs are part of a broader system: Spanish evaluative morphology. There is
plausible symmetry and contrast in the system caused by the opposites littlenesss"
(DIMs) and "bigness" (AUGs). If we accept Jurafsky's "childness" core proposal, then
this symmetry and contrast is lost in the system, which is unfortunate from a linguistic
perspective and criterion 4 is not met. The section on AUGs below elaborates more on
this symmetric contrast in the system. The littlenesss" proposal also satisfies the criterion
of predictability (5 above), at least partially; admittedly as much as the "childness"
proposal. Criteria 1, 2, and 4 above seem more critical to support the littlenesss"
Studies on DIM's meaning such as the one carried out by Savickiene (1998) in
relation to Lithuanian L1 acquisition seem to oppose "smallness" as a semantic core for
DIMs. Savickiene concluded that "the non-semantic meaning of the earliest diminutives
disconfirms the assumption of smallness as central meaning of the earliest diminutives"
(p. 133). My study agrees and disagrees at the same time with this conclusion. This
agreement or disagreement depends on what she meant by "central meaning". If it refers
to the semantic core sense, then my study disagrees. The fact that the most important and
common use of DIMs is [dear] does not oppose the idea that this sense may have
emerged from the [little] core sense. We simply need to recall that important
metaphorical and inferential semantic extensions are at play here. On the other hand, she
does not present any argument regarding what the core sense could be. If "central
meaning" means "most common use", then my study agrees, and the next chapter comes
back to this issue.
Admittedly, many uses of DIMs in the data analyzed here may be confused
between [little] and [child] connotations, like the following two examples:
1) Male church singer: (introducing his next song):
"Tenemos un pequefio pajarlTO en mi casa"
We have a small bird+DIM in my house
2) A mother to her 6-year-old son: "iQuieres pifiITA?"
Do you want pineapple +DIM?
In both cases, the DIM may be ambiguous. A further look to the context (at times,
the only help we have to accurately interpret EVALs) seems to indicate that example 1
has the [little] function, whereas 2 has the [child] connotation. In 1 there is no reference
to the age of the bird whatsoever, thus it looks more as a reference to size. In 2, the same
mother asked the same question to her husband without using the DIM suffix. Thus, this
[little] vs. [child] debate is justified but regarding original denotations of such suffixes,
my study takes a clear stand.
Because of all these arguments, the littlenesss" proposal supported in my study
renders the radial category graph shown in figure 3. This graph shows littlenesss" as the
core sense of DIMs primarily and "childness" secondarily; which are within a circle to
signal this semantic core. All the other DIM functions/senses grow out of this core sense
via linking chains and constitute pragmatics-driven uses; all those out of the circle. The
pragmatic uses on top represent the affection category; the middle line shows the
pragmatic category of attenuation, and the bottom line contains derogation-driven uses.
Whenever two (or more) arrows point at a single function, it implies that that function
may have two (or more) paths for semantic-pragmatic extension. Some functions do not
directly connect to the core sense, which semantically distances those functions from the
core sense and causes less semantic transparency. Yet, even those relatively obscure
functions indirectly connect to the core, as this semantic network shows. Thus, this graph
connects this section (DIM: core sense) to the next section of this chapter (DIM: chains).
[irony] [female] [pejorative]
Figure 3. New proposal for DIM's radial categories
Let us keep in mind always that Figure 3 presents a description of the potential
associations of the meanings and functions of DIMs in a particular speech community
(the participants in the data collection process). We should not lose sight of the cultural
relativity of the links in this graphic representation.
The following are the (non-discrete but continuous) categories, senses or
connotations observed in the data in relation to DIMs: endearment (or [dear], according
to the type of notation used in semantic decomposition analyses), littleness ([little]),
childness ([child]), irony ([-dear], [-little]), intensification ("very"), attenuation,
euphemism, flirtation, femaleness ([+female]), commiseration, and pejoration. These
eleven uses of DIMs in the data are analyzed from a pragmatic perspective in the next
chapter (Chapter 4). That chapter elaborates on each of these and the major pragmatic
categories under which they may be classified. The present semantic analysis shows how
all these notions are related.
Let us consider first the related notions of "child", "little" and "dear". The
following may be one logical and natural connection between the [little] and [dear]
functions. As Taylor (1995) put it: "Human beings have a natural suspicion of large
creatures; small animals and small children on the other hand can be cuddled and
caressed without embarrassment or fear" (p. 145). This connection between the [little]
and [dear] notions "is thus grounded in the co-occurrence of elements within an
experiential frame" (p. 145). It is naturally embedded in human beings' perceptions and
previous experience. Without necessarily rejecting the previous proposal, my study
suggests that there can also be another possibility. The previous hypothesis is of a more
inferential nature (i.e., if then). The second possibility is of a more metaphorical
nature. The notion of littlenesss" or "smallness" may have been transferred from the size
plane to the distance plane: The more distance between two people, the less intimacy and
affection between the two. Thus, since the distance (between mother and child, for
example) is so little or small, then it may reflect a high degree of affection or endearment;
thus, there could be a metaphoric association between the [little] and [dear] meanings.
One example from the data that may show this connection between [dear] and
[little] is the one below, where some people are celebrating the high school graduation of
3) Young man: (reading a funny poem that he improvised for a teen highschool graduate)
"Ron.CITO, gracias por ser un buen primlTO"
Ron. +DIM thanks for being a good cousin+DIM
Ron. was, at that moment, 17 years old, but he has been at church since he was born.
Thus, most of the people at the party are friends or relatives that know him as a child and
care for him very much. This is a celebration for Ron. to show their appreciation for him
precisely. Even though he is not a child any more, he continues to be loved by these
people, and they continue to call him "Ron.+DIM".
Regarding the [child] and [little] connection, DIMs are normally associated with
children because they are "little". Thus, children and DIMs is a normal observed
connection cross-linguistically (Jurafsky, 1996; Melzi and King, 2003). We typically
observe in children two main features: littleness and endearment. Children are LITTLE
and they are DEAR to us. Even animals (at least most of them, especially those with
some cortical endowment or limbic functions; e.g. mammals who bear live young) tend
to love, protect and care for their offspring. According to the famous psychoanalyst
Erikson (1950), generativity (which is embodied in the need to care for, raise, and mentor
the offspring) is a crucial stage of development of many living beings. This readiness to
parent, Erikson asserts, may be viewed as naturally built into our species.
Thus, it is not absurd to think that this may be (at least) a reason for the common
connection of DIMs with endearment. Children are little and children are dear; thus, this
littlenesss" may be "dear". This is based on a basic logic syllogism of the type "if ...
then" (Aristotle's Prior Analytics in Smith, 1989).
There are two premises and one conclusion:
Premise A: If [+child] is [+little], AND...
Premise B: If [+child] is [+dear], THEN...
Conclusion: [+little] is [+dear] (henceforth, DIM = little and/or DIM = dear)
This, for some linguists, may be considered a type of semantic shift, where the word or
morpheme takes on a new meaning often related to the original one.
Under this approach, we can then conclude that even though there is nothing
"little" in the notion of "dear" (in pure semantic terms) of the DIMs, the common use of
such affixes with the endearing function comes from the experiential association of
children with features such as littlenesss" and "endearment" (in logic terms). It seems to
be an example of metonymy, since there seems to be an association by context. It is
empirically observable in many cultures and languages of the world that the language in
children-oriented environments is heavily characterized by such morphemes. This is
probably the reason why "little" and "dear" are the functions or "meanings" of the DIMs
normally accounted for in the literature.
This endearment notion discussed above is the direct source for some uses of DIMs
that my study labels as flirtatious. The "flirty" DIM is a semantic/pragmatic extension of
"dear" to the sex or the sex-related arena. This DIM means "dear" but "sexually dear".
Thus, it is a type of endearment with sexual connotations, or in simple terms, sexual
affection or interest. The following is an example:
4) "y junio (the calendar fireman for that month): Mig..MiguelITO" "Ay virgen santa!"
& June: Miguel...Miguel+DIM. "Wow, Holy Virgin!"
This DIM may mean "dear Miguel" but also "sexy/hot Miguel". A young woman looks at
firemen posing for a picture calendar. She obviously admires the physical appearances of
these firemen models. For the speaker, this fireman is so sexy that she likes him very
much. This may be a link between the [dear] and [flirt] function in a real-life context.
Let us now go to the opposite meaning of DIMs, in comparison to the three
accounted for thus far. Many of DIM's uses actually mean the contrary of [child], [dear],
and [little], especially the last two. These are examples of irony or sarcasm, which
represent the most difficult connotations or uses to argue for in this semantic connection.
Furthermore, we also see that, according to Kruez (1996), the first and primary cues that
may help signal irony are precisely the counter-factual ones. These two ideas, direct
opposite and counter-factual, imply that when meant to be ironic, the DIM actually
means "big" or more commonly "not dear or not appreciated"; the opposite of the
common notions of DIMs (little, dear). It seems that through the agency of semantic