DIAGRAMMATIC AND IMAGIC HYPOICONS
IN SIGNED AND VERBAL LANGUAGES
Università Roma Tre
Università della Calabria
Far from being a marginal phenomenon in natural languages, iconicity is widely attested in verbal languages and is
a structural feature of signed languages. There is a difference, though, in the nature of iconicity in each code.
Whereas almost all instances of iconicity detectable in verbal languages belong to the Peircean category of
diagrammatic hypoicons, the iconicity which characterizes signed languages both at the lexical and at the discourse
level may be ascribed to the category of images. The wealth of hypoicons of images in signed languages can be
explained by taking into account not only the peculiar semiogenesis of these languages, but also the “substance of
their expression”. Compared to the phono-acoustic modality, the visual-gestural modality filters iconicity in a less
fine-grained way. In spite of this difference, signed languages are as profoundly arbitrary as verbal languages.
Iconicity is in fact governed and ruled by system constraints. The iconic phenomena detected in signed languages
do not contradict, therefore, the Saussurean principle of Radical Arbitrariness, largely considered as a design
feature of natural language.
The problem of linguistic iconicity raises the age-old question of the extent to which language is
a mirror of reality and of extent to which it is able to be so. Undoubtedly, this is one of the
research domains in which linguists benefit the most from the inclusion of signed languages
within the scope of their studies. Signed languages are quite obviously more iconic than verbal
languages. A careful comparison between signed and verbal languages does therefore guarantee,
in this particular respect, a deeper understanding of the causes, mechanisms and limits of
iconicity in language.
In this paper, we shall assume that the analogies and differences between the iconic phenomena
of signed and verbal languages can be accounted for in a semiological perspective. Iconicity
occurs, of course, in many kinds of non-linguistic signs; the iconicity in human languages,
however, displays a number of peculiarities. Our claim, therefore, is that a unified theoretical
account of linguistic iconicity must be achieved in order to allow for the comparison of its
different manifestations in different kinds of human languages. In particular, we will show that
the taxonomy of iconic signs proposed by Charles Sanders Peirce (see §2 below) may be
productively applied to both signed and verbal languages. We shall then focus on the
comparison between iconic phenomena in signed and verbal language structures (§3-§5). We’ll
argue that, although, iconicity is detectable in both codes, there is a clear difference in the way
the phono-acoustic modality “filters” iconicity in verbal languages as compared to the visual-
gestural modality of signed languages. In particular, we will argue that whereas the iconic
phenomena detectable in verbal languages can be ascribed to the Peircean category of
“diagrammatic iconicity”, when it comes to signed languages iconicity ought to be regarded in
most instances as an example of what Peirce calls “imagic” iconicity.
Finally, we shall argue that in both signed and verbal languages iconic signs are compatible
with the Saussurean notion of “arbitrariness” (§6-7).
This paper is the outcome of close collaboration between the two authors. Tommaso Russo is primarily
responsible for writing Sections 2, 4 and 7; Paola Pietrandrea for sections 3, 5, and 6. The introduction was co-
written. The authors wish to thank Elena Pizzuto for the frequent and thorough discussions on the iconicity of sign
2. A Peircean account of iconicity in verbal and signed languages
According to Peirce
every sign is embedded in an interpretation process which he describes as
follows: a representamen, which is to say a particular sign, will relate to a particular object,
namely a denotatum, through the action of an interpretant.
The interpretation of a sign can thus be seen as a threefold process: firstly, a particular
entity is selected as a sign or representamen. Secondly, a certain entity is selected as object for
the sign. Thirdly, the sign is related to the relevant features of its object which allow the
interpretation process to take place.
Interpretant is the name Peirce gave to the action of selecting those features of the object which
lead to its schematic representation. He calls this representation immediate object, whereas the
real denotatum is called the dynamic object.
Peirce distinguishes between several kinds of signs, thus designing a very complex and
comprehensive taxonomy. We shall focus especially on two kinds of signs: icons (and their sub-
kinds: images and diagrams) and symbols.
Icons are signs which resemble their objects “in some respects”, such as, for instance some
qualitative aspect of their immediate objects. Thus, a naturalistic painting of a human figure is
based on a selection of features which relate to that person: Peirce points out that a portrait only
ever resembles its object in some respects.
Symbols, on the other hand, are norm-following signs which relate to their objects through
some convention, or some other general rule. In a Peircean perspective, linguistic signs are
primarily symbolic in nature. Nevertheless, as Peirce states, it is also possible to find a particular
kind of icons among linguistic signs: ‘icons in which the likeness is aided by conventional rules’
(C.P. 2.279). In the case of linguistic iconic signs, the mediation of semantic and expressive
linguistic conventions provides the link between the sign and some relevant features of the
object of reference
Peirce draws a crucial distinction between two different kinds of icons which he calls
hypoicons: images and diagrams
. Images are hypoicons which ‘partake of simple qualities’ of
their object: for example, they have certain perceptual features in common with their objects.
Diagrams are instead hypoicons which ‘represent the relations, mainly dyadic, or so regarded, of
the parts of one thing by analogous relations in their own parts’. Diagrams, that is, are schemata
of objects or events in which the relations between features of the object are represented by
relations between features of the sign form. Whereas images are overtly referential in nature,
diagrams are strictly devoted to the representation of different kind of abstract relations.
Our claim is that these two kinds of hypoicons are to be found in both signed and verbal
languages. Nevertheless, it will be argued in the following sections that there is a substantial
difference in the balance between imagic and diagrammatic iconicity in the two kinds of
3. Traces of diagrammatic iconicity in verbal languages
The role of iconicity in verbal languages has been deeply explored over the past decades by
several functional linguists, who took the undeniable wealth of elements of iconicity in language
Our account is based on the texts written by Peirce between 1902 and 1906, a particularly productive period of his
research in the theory of signs (Peirce 1931-58, see also Proni 1990).
Peirce provides a phylogenetic explanation for the limited degree to which iconic signs are present in human
languages: “In the earliest form of speech, there probably was a large element of mimicry. But in all languages
known, such representations have been replaced by conventional auditory signs. These, however, are such that they
can only be explained by icons.”(C. P. 2.280)
Peirce distinguishes three kinds of hypoicons: images, diagrams and metaphors but we will deal here only with the
as evidence against the generativist principle of the Autonomy of Language.
The comparison between verbal and signed languages contributes to this discussion
substantially by highlighting two facts. Firstly, with the exception of a few interesting instances
(which we shall discuss later in this section), almost all the examples of iconicity of verbal
languages described in the literature fall into the class of diagrammatic hypoicons. Secondly, the
‘banishment’ of imagic iconicity is not imposed by the intrinsic nature of language, but is to be
ascribed to the nature of the phono-acoustic medium.
The analysis of verbal languages has detected traces of iconicity on three different levels: 1)
in the internal structure of linguistic systems (we shall refer to this kind of iconicity as
paradigmatic iconicity); 2) in the morphosyntactic structure of clauses (syntagmatic iconity); 3)
in various discourse-pragmatic functional domains (pragmatic iconicity).
Two well-known examples of paradigmatic iconicity are the One-Form-One-Meaning
Principle (Bolinger, 1977; Haiman, 1985a) and the Iconicity of Lexical Categories Principle
(Hopper & Thompson, 1985)
. Both principles describe instances of mapping between relations.
According to the One-Form-One-Meaning Principle, the relations among the forms of a
linguistic system mirror the relations among the meanings of the semantic domain. The
Iconicity of Lexical Categories Principle, instead, describes the relations between prototypical
representatives of a lexical category as being mirrored by the relations between their formal
apparatuses, which is to say that, depending on the greater or lesser degree to which an element
from the former class is prototypical (a more or less prototypical verb or noun, for instance) its
formal apparatus will be more or less complete
. In so far as both principles describe instances
of mapping between relations, both should be also taken as examples of diagrammatic
Examples of what we have called “syntagmatic iconicity” are the naturalness proper of
morphological constructions (Dressler, 1987), the isomorphism between conceptual distance
and linguistic distance (Givón, 1980; Haiman, 1985a), and the “Relevance Principle” guiding
the stem-inflection relations (Bybee, 1985). Let us examine these phenomena in more detail.
Dressler (1987) claims that the ideal morphological sign has to have a diagrammatic
structure; its structure, that is, has to mirror the relations between its component parts. The
diagrammatic relation between the components of the morphological sign on the one hand, and
the relation between their designata on the other, entail that the value of the sign may be
understood through a mere exploration of its structure. Building on Vennemann’s (1973) and
Behaghel’s (1923-32) intuition, Haiman (1985a) has proved that the relation holds cross-
linguistically in domains such as the expression of coordination, causation, transitivity and
possession, that the closer the two elements in the described relation are,
the more linguistically
integrated they are.
The binding hierarchy proposed by Givón (1980) goes in the same
direction. Givon has shown that the reduction of a complement clause mirrors the degree to
which its truth is implied by the matrix clause. A refinement of this iconic principle has been
proposed by Bybee (1985), who has identified a Relevance Principle which governs stem-
inflection relations, according to which ‘The more a morphological category is relevant to the
The former states that “the natural condition of language is to preserve one form for one meaning, and one
meaning for one form” (Bolinger, 1977) and the latter that “the more a form refers to a discrete discourse entity or
reports a discrete discourse event” - i.e. the more it fulfils the prototypical functions of, respectively, noun and verb
- “the more distinct will be its linguistic form from neighbouring forms, both paradigmatically and syntagmatically”
(Hopper & Thompson, 1985: 151).
According to Hopper and Thompson, in fact, prototypical verbs and nouns display all the morphological and
syntactic properties of verbs and nouns respectively. This does not hold for less prototypical nouns and verbs which
lose part of their formal properties in discourse.
i.e. the more semantic features, properties, or parts they have in common, the more they will affect each other, be
factually inseparable, be perceived as a unit, whether inseparable or not.
i.e, the lesser is the number of morpheme and word boundaries lying between the two units.
verb […], the greater is the morpho-phonological fusion of that category with the stem’ (Bybee,
. Thus, syntagmatic iconicity too is diagrammatic. The relations among the parts of a
linguistic sign, in fact, always mirror relations among their designata. This holds for the
components of natural morphological constructions; for the two terms in a relation of
coordination, complementation, causativity, transitivity, and possession; and finally it holds for
the relation of stem-inflection.
Iconicity is also a feature of the way a number of discourse-pragmatic functions are
reflected in language. Givón (1991, 1984 (2001)) has studied the principles of iconic coding
which guide the structuring of pre-grammatical systems such as pidgin communication; the
same principles are also a feature of grammaticalized languages, although at that point they are
integrated to a greater extent with the more symbolic machinery. Some of these principles focus
on the fact that linguistic relations tend not only to reflect relations between entities or events
within reality (as we have seen above), but also they reflect relations among the speaker’s
communicative intentions. For example, the relation between information that is more or less
important may be mirrored by the amount of linguistic material employed in their encoding
, or by the position it occupies in the string (Sequence Rule)
. Once again
we are faced with iconicity of the diagrammatic kind.
All the instances of iconicity examined in this review are thus diagrammatic. At this point,
the question arises whether there is any imagic iconicity in natural languages, or whether the
intrinsic nature of linguistic signs prevents them from partaking of the simple qualities of their
Indeed, imagic hypoicons have been detected in natural languages in spite of their being
relatively rare. Ever since De Saussure, onomatopoeia has been widely acknowledged as
representing one such category. Other examples are the ideophones of many African and Asian
languages (Kita, 1997), sound-symbolism (Frei, 1970), the evocative ‘vocal gestures’ occurring
in expressive usages of languages (Fónagy, 1983). In all these cases the iconic sign does not
simply establish a correspondence between relations, but it recalls its denotatum in some
We propose that the correspondence between the order of constituents in a sentence and the
order of the events or entities they depict, which are, however, frequently ascribed to the
category of diagrams, be understood as a further class of images in verbal languages. An
example of this kind of iconic sign is Caesar’s well known utterance veni, vidi, vici, which, as
Simone (1995a) pointed out, mirrors in its structure the ordo naturalis of the three events, thus
enabling the listener to reconstruct the order of the events through a simple inspection of the
structure of the utterance. We believe that these constructions lie at the boundary between
diagrams and images: as diagrams they represent the sequential relationship between the events
in their structure; as images, they actually exploit a qualitative feature of the acoustic signifier,
namely its temporal linearity. Thus the temporal sequence of the events is mapped onto the
temporal acoustic medium point by point. Indeed the focus of these constructions seems to lie
more in the representation of the temporal sequence of events, as a global semantic unit, than in
the relations between the single components. This does not hold true for the other diagrammatic
relations we have reviewed above. The difference between them is what makes us inclined to re-
categorize this kind of iconic signs as more image-like.
The relevance principle, supported by robust cross-linguistic evidence, explains the universal tendency of verbal
inflectional categories to be hierarchically ranked along the same path: stem < aspect < tense < mood < person.
Givón’s ‘Quantitative Rules’ state that: ‘predictable -or already activated - information will be left unexpressed’
and that ‘unimportant or irrelevant information will be left unexpressed’ (Givón, 1984(2001: 35).
Givón’s Sequence Rules state that ‘more important or more urgent information tends to be placed first in the
string’ and that ‘less accessible or less predictable information tends to be placed first in the string’ (Givón, 1995:
Imagic iconicity, thus, is not excluded from natural languages; on the contrary, as the next
section will show, it plays a considerable role in characterizing another kind of natural language,
namely signed languages. Simply, diagrammatic iconicity is far more pervasive in verbal
languages than imagic iconicity. We shall propose an explanation for such a tendency in section
4. Imagic iconicity as a structural device in signed languages
It is widely acknowledged that the grammatical structures of signed languages contain examples
of diagrammatic iconicity. It is not uncommon, for instance, to find noun-verbs pairs which are
characterized by regular affixation processes; the combination of movement affixes and the
stem handshape marks the distinction between such noun-verb pairs (Supalla and Newport
1978). Other regular stem-inflection relations concerning verb modality (Wilcox and Wilcox
1995), verb aspectual system (Klima and Bellugi 1979) and certain regular features in the
organization of the lexicon (Boyes Braem 1981) are also to be understood as instances of
In all these cases, though, imagic hypoiconicity co-exists with diagrammatic hypoiconicity.
As Wilcox (2004) has argued, with many of the above examples the visible features of a sign
may be easily traced back to some schematic representation of a prototypical denotatum. For
example, the iterative aspect is marked by circular movements; these movements clearly
represent a prototypical instance of the superordinate category of “iterative actions”.
Figure 1. The two signs TO MEET and TO MEET-REPETEADLY
Therefore, although we do find examples of diagrammatic iconicity in signed languages,
their pervasively iconic nature (Boyes-Braem, 1981, Pizzuto & Volterra, 2000, Wilcox 2004,
Wilcox and Wilcox 1995, Cuxac, 2000) has to be ascribed to the category of images. Two
different examples will be useful to illustrate this important point: the first one pertains to the
level of lexical organization and the second one pertains to the level of discourse.
As Penny Boyes-Braem firstly made clear (1981), signed lexemes are often made up of
formal features which are visually motivated and are thereby iconic. Their visual motivation is
not idiosyncratic; it derives from regularities at the level of the form of formational parameters.
Handshape forms, for example, often relate to features of a sign’s meaning via reference to
some particular visual form. Thus, in the LIS lexicon, the B handshape relates to meanings
which imply reference to a flat surface (as in the signs TABLE, FLOOR, CARPET; cf. figure
2.): a specific visual feature of the hand (namely that of being physically spread) has been
mapped onto a specific feature of the meaning of the signs. Other handshapes relate to other
semantic features: 5 handshapes in which the hand is open and spread out relate to flat, non-
compact, flexible surfaces (as in the LIS signs DRESS, BUTTERFLY, WATER; cf. figure 3);
G handshapes relate to compact, long, monolinear objects, etc.
It is worth noting that groups of lexemes with common formational parameters form
comparable paradigms to the paradigms of verbal language lexemes which share a common
derivative stem. Nevertheless, each of these signs is iconic and visually motivated: iconicity,
thus, coexists with the autonomous regularities of the lexical system.
Figure 2. The signs TABLE, FLOOR, CARPET
Figure. 3. The signs DRESS, BUTTERFLY, WATER
It must be noted that when signs have a common handshape, they do often also share some
feature of meaning; however their global meanings differ greatly from one another and are not
predictable from the handshape itself. This confirms Peirce’s notion that symbolic iconic signs
do abstract certain schematic features of their dynamic objects in an ordered and
Besides, the semantic spectrum of the visually motivated handshapes in these lexical
paradigms is rather wide. This feature allows for a balance between visual motivation and the
systematic semantic oppositions of the lexicon, which also turns out to be an important feature
of our second class of examples, namely that of imagic iconicity in discourse.
Alongside lexical ‘frozen iconicity’ it is also possible to find instances in discourse of what
has been called ‘dynamic iconicity’ (Russo 2004), particularly in the use of so-called “classifier
forms” or polymorphemic productive forms (cf. Schembri 2003, Brennan 2001, Cuxac 2000).
Signed language classifiers are free or bound morphemes which lack a fully specified lexical
meaning; they are used to refer to one or another member out of a class of entities which share
the same form. These forms have often been compared to the classifiers of some verbal
languages (see Aikhenvald 2003, among others)
. For our purposes, it is important to highlight
In particular, signed language classifiers are not redundant, as many kinds of verbal language classifiers are. They
have multiple lexico-grammatical functions (e.g., they have verbal functions as well as locative-adverbial meanings,
that in discourse classifiers display a particular kind of dynamic imagic iconicity. The visual
features of these forms, and especially the handshapes, can be related to a particular semantic
feature of a denotatum in many different ways. We have already noted that in the lexicon the
semantic spectrum of a particular handshape may cover different meanings. The semantic
spectrum of certain handshapes can be further extended in discourse and acquire particular
meanings. An example from a signed narrative will help illustrate this point:
Figure 4. FALLING-LEAF-(FROM-BRANCH)
In the first part of the sequence above (fig. 4) Giuseppe Giuranna, a Sicilian LIS signer
describes a leaf falling from a tree. The signer doesn’t produce the citational form of the LIS
sign LEAF, but he employs a so -called “classifier”, the 5 handshape classifier, the form of
which is reminiscent of the flat flexible form of a leaf .
It should be noted that the iconic relationship between the 5 handshape classifier and the
semantic features of the referent is not in itself sufficient to distinguish between a leaf and many
other objects that can be semantically characterized as [+flexible] and [+ flat] (other objects fit
the description of a flat flexible surface, such as ‘pieces of paper’ or ‘flags’, etc.). The audience
is able to relate the use of a handshape to the appropriate referent through the indications given
by the context and the co-occurrence of handshape, movement, orientation, and the place of
articulation. A very general semantic feature ( such as [+ flat]) may or may not be in itself
sufficient to establish which referent the signer is addressing. When it is not sufficient, the co-
occurrence of the other parameters will reinforce the relation.
The similarity between the 5 handshape and the “leaf” referent, therefore, has to be
reinforced by the additional features of the movement and locus of the handshape: indeed, in the
utterance signed above the “slow weaving downward directed movement” of the handshape is
also iconically related to the semantic representation of a ‘falling leaf’; this further element
significantly contributes to understanding the relationship between the sign and its meaning. In
addition, the left hand assumes a G handshape, a classifier generally used to address long and
thin objects: in this context the referent can be identified as the branch of the tree from which
the leaf is falling.
We have also noted (cfr. Russo 2004) that a slight modification of the context of utterance
and of the co-occurring parameters can radically change the referential meaning of the
In the following utterance (fig. 5), drawn from the same sequence of signs mentioned
above, the signer introduces a new referent whilst maintaining the same handshape classifier.
they can use as pluralizing devices and very often be used as proforms). Finally they are not a limited, closed set
because virtually every handshape can be used as a classifier.
Figure 5. The
signer exploits a 5 classifier to represent a NEWSPAPER on the ground
While his right hand is moving, the left hand suddenly assumes and retains throughout a B
handshape, which first specifies the ground on which the leaf falls, and then the part of the
newspaper which remains in place while a page turns over. Later on, the signer changes location
and movement in the production of the sign, and the change helps reinterpret the 5 handshape
classifier as representing “a newspaper page in the act of turning”. It may be argued, thus, that
the simultaneous presence of visually motivated parameters, as established in the discourse
sequence, is a major tool to communicate and to interpret the meaning of the utterance.
We also suggest that the previously available linguistic and situational context of utterance
provides additional information which helps identify the referent of the sign
. These mappings
of semantic features onto expressive forms are rather frequent in signed discourse: it can be
argued that when the signer is not relying on the standard lexicon, “imagic iconicity” is a
necessary resource for a full understanding of the utterance. The underspecified meaning of the
regular morpho-phonological units are defined at discourse level exclusively via iconic
interpretation. Thus iconicity cannot just be regarded as an accidental feature of signs; we must
acknowledge that it is a proper structural device.
It is important to underline that this kind of dynamic imagic iconicity coexists with the formal
constraints of the grammatical levels: the entire sequence we have analysed is actually “well
formed” on the grounds of the morphophonological and morphosyntactic constraints which
govern parameter selection, of the semantics of classifier handshapes and movements, the
alternation of left and right hands, and the non manual markers.
The iconic imagic phenomena described above are largely attested both in sign language
lexicon and discourse. Pietrandrea (2002) has calculated that 50% of the occurrences of
handshapes and 67% of the occurrences of body locations are iconically motivated (based on a
corpus of 1944 standard signs drawn from the LIS dictionaries). In a comparative analysis of
dynamic iconic structures in three LIS discourse registers (Conferences, Narratives and Poetry),
Russo (2004) has calculated that: a) iconic features are a fundamental trait of the citational
lexemes which occur in all three language registers (poetry being the most iconic register); b)
the analysis of non-citational forms in the signed utterances brings out an additional
layer of iconic features. The three registers differ as to the occurrence of discourse iconic
features. Although the poetic and narrative registers are more pervasively characterized by
dynamic iconic imagic structures (Narratives: 43%, Poetry: 53,4%), a remarkable percentage of
these structures (13 %) also shows up in the more formal and technical register of Conferences.
To sum up, “imagic” iconicity is a widespread structural semiotic device of signed
As, in our example, the signer is talking about trees before introducing the sign FALLING-LEAF, this
information increases the understanding of the iconic relation between the classifier handshape and the leaf (the
contextual information “+ related to trees,” is actually added to the semantic features “+ flat”, “+flexible”
reinforcing the iconic relation between “a leaf” and the particular articulatory unit we are examining).
languages integrated with the formal regularities of the language system both at the lexical and
at the textual level.
5. Why imagic iconicity prevails in signed languages
Cuxac (2000, Cuxax and Sallandre, this volume) has argued that the great incidence of image
hypoicons in signed languages is related to the peculiar semiogenesis of these languages
(Cuxac, 2000), which induces the community of speakers to rely on a massive iconization of
experience to guarantee the mutual comprehension through a code which is continuously
renewed with each generation.
We fully agree with this approach, but we are still interested in understanding the
mechanisms through which imagic iconicity finds its place in signed languages. This could help
us understand the reason why, as the first part of the present work shows, there is a different
balance between imagic and diagrammatic iconicity in verbal and signed languages.
Our claim is that the peculiar nature of the articulators and the medium employed in signed
languages play an important role in preserving imagic iconicity.
As argued by Boyes-Braem (1981), in signed languages the hands are used with a linguistic
purpose. The hands are employed in daily life in many tasks, such as pointing, manipulating
objects, counting, and, in certain cultures, (the Italian, for example) representing objects. It
makes economic sense that signed languages should make efficient use of this pre-codification
of the hands in the creation of signs.
Boyes-Braem’s argument can be easily extended to explain the linguistic use of the body.
The speaker’s body is always present in signed languages discourse. Again, it makes economic
sense to exploit this presence to express meanings that are related to parts of the body.
Besides, as Wilcox (1999) has noted, just as the words of verbal languages are made up of
sounds, signs consist of hand movements, which are both actual objects in the world as well as
representations of objects in the world. Unlike sounds, hand movements are visible objects in
the world and they are visible objects about the world. As visible objects, hand movements
exploit the world’s four dimensions (Hockett, 1978). The fact that visible gestures share the
same substance of the physical world (at least that part of the world most frequently referred to
in language, i.e. visible entities) somehow explains why gestures represent it in a more detailed
and “qualitative” fashion, as Peirce would put it, than sounds do. Gestures do not entail the
switch from a visual world to an acoustic representation of it: there is no need to adjust a four-
dimensional world to the monolinearity of the acoustic channel (Hockett, 1978). Therefore
gestural representations partake of the physical qualities of the world to a greater extent than
acoustic representations do. They are, in other words, more suited to becoming imagic
That the nature of the medium has to do with the incidence of images in signed languages is
highlighted by the fact that most instances of image hypoicons in verbal languages are either
representations of sounds through an acoustic medium - e.g. onomatopoeia, ideophones, and
less directly, phono-symbolism - or representations of temporal sequences of events in a
temporal medium - e.g. the correspondence between the order of the constituents in a sentence
and the order of the events or entities they depict. This proves that when there is no synaesthetic
switch, imagic iconicity arises more readily.
If we look at it the other way around, a large number of instances of imagic hypoiconicity
in signed language discourse seem to exploit the simultaneous combinatory properties of signs
and their visuo-spatial features. Indeed, as we shall see in §7, the possibility of assigning a
particular semantic value to each formational parameter is frequently exploited in signed
utterances by means of the spatial/simultaneous nature of their medium.
6. Radical arbitrariness in signed languages
In spite of the pervasive presence of imagic iconicity, a principle of radical arbitrariness is
preserved in signed languages. Firstly, in a large part of signed language lexicon (i.e. the part
that Cuxac (2000) defines “standard lexicon”), the iconicity is “phonemized”. That is, the
depiction of referents can only be conveyed by the few formational parameters the language
allows. Every sign language, as well as every verbal language, selects its “phonological”
patterns arbitrarily. So, for example, the handshape used in the ASL sign for BATHROOM) -
figure 6- is not part of the formational parameter inventory of LIS.
Figure 6. The ASL sign for BATHROOM
Secondly, the selection of the aspects of articulators and referents to be considered as
relevant at the linguistic level is arbitrary.
Figure 7. The signs STONE, CAR, PRISONER
Figure 7 shows three signs articulated with the same handshape, the A handshape. This
handshape is iconic in all the three signs. Indeed, in the first sign (STONE), the A handshape
conveys the meaning roundness (of a stone). In the second sign (CAR), the A handshape
conveys the meaning grasping a thick object (for example a steering wheel). In the third sign
(PRISONER), it conveys the meaning of closed hands (like those of a prisoner in chains). It is
easy to notice that in spite of the fact that all these usages of the A handshape are iconically
motivated, different aspects of this handshape are selected as linguistically relevant in the three
signs. In the first sign, it is the form of the handshape to have been selected, in the second sign,
it is its grasping function; in the third sign the handshape is selected in its physical salience of
closed hand. The same holds for each iconic usage of handshapes and locations (see Pietrandrea,
2002). Thus, given a formational parameter, it is not possible to predict its meaning because it is
not possible to predict which aspect will be selected as relevant for linguistic purposes.
In the same way, it is not possible to predict which sign will be used to express a given
referent (Klima and Bellugi, 1979), because it is not possible to predict which aspects of the
referent will be selected for linguistic purposes. As shown by Caselli et al. (1994), both the ASL
and the LIS sign for sheet (figure 8) are iconic. Nevertheless they are radically different. In the
ASL sign, in fact, it is the flatness of the object that has been selected as salient, whereas in LIS
sign it is the form of the hand grasping the object that has been selected.
Figure 8. The ASL and the LIS signs for SHEET
Finally, all quantitative analyses highlight the fact that a consistent part of signed language
systems is totally arbitrary. The analysis of the standard sign corpus, for example, shows that
50% of the occurrences of handshapes and 33% of the occurrences of locations have no evident
motivation. The analysis conducted on discursive iconicity shows that opaque forms widely
occur in lectures and narratives (respectively 52,6% and 35,7%) and to a lesser extent in poetry
(23% of signs). Arbitrariness, thus, seems to co-exist at different levels with imagic iconicity in
7. The coexistence of iconicity and radical arbitrariness
How can highly iconic language phenomena coexist with the formal and structural needs of a
linguistic system? We claim that the iconic phenomena in signed languages, as well as those in
verbal languages, are not just an incidental feature of the surface form of signs. Indeed the
coexistence of iconicity and arbitrariness must lie at the heart of the complex interplay between
the formal requirements of the linguistic system and the pragmatic constraints which guide the
interpretation of a linguistic utterance. On the one hand, each particular linguistic unit responds
to certain formal requirements at the phonological, morphological and syntactic level (rules for
the combination of single units in a meaningful proposition included). On the other hand,
textual and situational context always add additional information to the interpretation of the
Ferdinand De Saussure argued in his posthumously published Course de Linguistique
général (1967) and in his notes on the theory of signs (De Saussure 2002), which have
been recently made available, that a language system must guarantee a balance between the
formal requirements of the language system (i.e. the formation rules and the semantic
regularities, De Saussure used to refer to with the name of langue) and the residual space
assigned to the freedom of interpretation of an utterance in a particular context (parole).
De Saussure’s point was that both the autonomous regularities of a language system and its
freedom to vary along the diachronic dimension could be explained by a unique principle: the
Radical Arbitrariness of the linguistic sign (De Saussure 1967:104-113, De Saussure
2002: 28-29). On the one hand, if understood as radically arbitrary, linguistic systems may
display formal regularities both at the expressive-phonological level (the signifiants) and at the
semantic level (the signifiés) without being determined by external reality. On the other hand,
radical arbitrariness is crucially related to the underspecification of linguistic units, which must
be regarded as one of their design features (De Mauro, 1991). Because they are underspecified,
linguistic units can be interpreted in context and can change meaning and form along the
diachronic dimension: this is a precondition for the semantic openness of natural languages and
in general for linguistic variation. We may say, in fact, using De Saussure’s terminology, that
language systems are open to those changes which arise from the variability of individual
linguistic utterances (parole), but they are also always determined by previous arbitrary
regularities concerning their signifiants and signifiés (langue). The Saussurean principle of
Radical Arbitrariness has often been understood as stating that the linguistic signs must not be
iconic, i. e., that they cannot display perceptual features of their referents (see Givón 2002). We
claim instead that this principle basically states that languages obey systematic formal
constraints which are autonomous, i.e. arbitrary, to the extent that they fulfil two distinct
requirements: (a) they are not predictable on the grounds of external reality and (b) they do not
exhaustively determine utterance meaning, allowing pragmatic interpretation and, consequently,
(see also Russo 2004, ch. 6).
Iconic features can thus characterize linguistic signs to the extent that they do not contrast
with these two requirements. This entails that iconic constraints on linguistic systems may also
be admitted, to the extent that they do not contrast with the regularities that can be consistently
detected in the language system at different levels. Futhermore, iconicity can characterize
linguistic signs at the level of pragmatic interpretation. Being underspecified, linguistic units do
not afford a complete and exhaustive interpretation of an utterance and need some pragmatic
prop for the interpretation to take place.
This is consistent with our analysis of dynamic iconic structures in signed texts
. As shown
above, discursive iconicity acts as a pragmatic prop to ease the understanding of signed
utterances. Diagrammatic and imagic hypoiconicity, in general, are useful hints to the
interpretation of the linguistic sequence. In signed languages the balance between formal
regularities and openness to pragmatic interpretation is obtained not only in the standard-lexicon
but also, as shown by Cuxac and Sallandre (this volume) in the so-called “structures de Grande
Iconicité”, i.e., the highly iconic structures which characterize signed discursive units (Cuxac,
2000), such as, for example, Role Taking Devices (Wilbur 1987, Ajello 1997, Pizzuto, Giuranna
e Gambino 1990, Engberg Pedersen 1989).
In Role Taking, posture, facial expressions and eye movements all contribute in
determining an iconic relation between the signer and the character he is embodying. While
these structures display a high degree of imagic iconicity, they are also constrained by regular
patterns governing the selection of the articulatory forms, the relation between the selected
forms, the features of meaning, the timing of the signs and their co-occurrence with other kinds
of sign (Pizzuto, Giuranna e Gambino 1990, Cuxac 2000). As noted above, the same point can
be made in relation to so called classifiers. In spite of the fact that signed language classifiers
are best characterized as forms that can be productively combined in discourse for a range of
lexico-grammatical functions, they do obey a great deal of formal constraints. Furthermore,
while the single classifier may be regarded as semantically underspecified when out of context,
iconicity adds a semantic determination to its contextual occurrence in a systematic way.
Iconicity is thus a major structural resource of signed languages permeating every level of
the language system and acting as a major pragmatic constraint in utterance interpretation.
We may conclude our analysis by saying that, while verbal and signed languages are both
characterized by iconic features at every level of their structure, signed languages seem to
exploit imagic hypoiconic devices to a greater extent. This is probably due to the semiogenetic
conditions of these communicative devices and to the particular features of the linguistic
community they are used by. In addition, a major structural drive to maintaining highly iconic
imagic forms in sign language systems seems to derive from their visuo-gestural modality and
from the multi-linear combinatorial features of sign components.
In addition the formal requirements of the linguistic system also leave space for the variation of the form of the
linguistic units, but we are not dealing with this point here.
The point is tenable also for the iconicity of verbal texts as in the case of the Givón’s Quantitative and Sequence
rules mentioned before in paragraph 2.
By contrast, verbal languages mostly exploit linear combinatorial devices
. This may have
contributed to the submergence of part of the imagic hypoiconic features of verbal signs and to
the prevalently diagrammatic character of their iconicity.
Iconicity and arbitrariness, thus, coexist in verbal and signed languages although they take
on different forms and reach a different balance in the structures of the linguistic system.
Aikhenvald, Alexandra Y.
2003 Commentary: Classifiers in spoken and in signed languages: How to know more.
In Perspectives on Classifier Constructions in Sign Languages, Karen Emmorey
(ed.), 87-90. Mahwah, NJ: Lawrence Erlbaum Associates.
1997 Lingue vocali, lingue dei segni e «l'illusion mimétique». In Miscellanea in Onore
di Enrico Campanile, Filippo Motta (ed.), 1-14. Pisa: Pacini.
1923-32 Deutsche Syntax. vols 1-4. Heidelberg: Carl.
1977 The Form of Language. London: Longmans.
1981 Significant features of the handshape in American Sign Language. Ph. D. diss.
University of California, Berkeley.
2001 Encoding and capturing productive morphology. Sign Language and Linguistics
Bybee, Joan L.
1985 Morphology. Amsterdam/Philadelphia: John Benjamins.
Caselli, Maria Cristina, Simonetta Maragna, and Virginia Volterra
2006 Linguaggio e sordità. Gesti, segni e parole nello sviluppo e nell’educazione. 2nd
edition. Bologna: il Mulino. Original edition by Caselli, Maria Cristina,
Simonetta Maragna, Laura Pagliari Rampelli and Virginia Volterra, Linguaggio e
sordità. Parole e segni per l’educazione dei sordi. Firenze: La Nuova Italia,
2000 La Langue des Signes Française (LSF) - Les Voies de l'Iconicité. Faits de
Langues 15/16. Paris: Ophrys.
Cuxac, Christian and Marie-Anne Sallandre
this volume Iconicity and arbitrariness in French Sign Language - Highly Iconic Structures,
degenerated iconicity and diagrammatic iconicity.
De Mauro, Tullio
1991 Ancora Saussure e la semantica. Cahiers Ferdinand de Saussure 45: 101-109
De Saussure, Ferdinand
1967 Corso di Linguistica Generale edited and translated by Tullio de Mauro, Bari:
Laterza [french Edition Paris, Payot, 1922].
2002 Ecrits de Linguistique Générale. Paris: Gallimard.
Dressler, Wolfgang U.
1987 Word Formation (WF) as part of natural morphology. In Leitmotifs in Natural
Morphology, Wolfgang U. Dressler, Willi Mayerthaler, Oswald Panagl and
Wolfgang U. Wurzel (eds.), 99-126. Amsterdam/Philadelphia : John Benjamins.
Indeed multilinear features seem to be prevalently exploited, in verbal languages, at the level of subphonemic
units and at the level of the suprasegmental traits, including the co-occurrence of meaningful gestures.
1989 Proformes en morphologie, syntaxe et discours. In Etudes Europeennes en
Langues des Signes, Filip Loncke and Sophie Quertinmont (eds.), 35-52.
1983 La vive voix. Essais de Psychophonétique. Paris: Payot.
1970 Cinquante onomatopées japonaises. In David Cohen, Mélanges Cohen, 359-367.
The Hague: Mouton
1980 The binding hierarchy and the typology of complements. Studies in Language 4:
1991 Isomorphism in the grammatical code: cognitive and biological considerations.
Studies in Language 15.1: 85-114.
2001 Syntax. A Functional-Typological Introduction, I. 2nd ed. Amsterdam /
Philadelphia: John Benjamins. Original edition, Amsterdam/Philadelphia: John
2002 Biolinguistics: the Santa Barbara Lectures. Amsterdam/Philadelphia: John
1985 Natural Syntax. Iconicity and Erosion. Cambridge: Cambridge University Press.
Hockett, Charles Francis
1978 In search of Jove’s brow. American Speech: A Quarterly of Linguistic Usage
Hopper, Paul J. and Sandra A. Thompson
1985 The iconicity of the universal categories ‘Noun’ and ‘Verb’. In Iconicity in
Syntax, John Haiman (ed.), 51-183. Amsterdam/Philadelphia: John Benjamins
1997 Two-dimensional semantic analysis of japanese mimetics. Linguistics 35: 379-
Klima, Edward S. and Ursula Bellugi
1979 The Signs of Language. Cambridge Mass.: Harvard University Press.
Peirce, Charles Sanders
1931-58 Collected Papers. Cambridge Mass.: Harvard University Press.
2002 Iconicity and arbitrariness in Italian Sign Language. Sign Language Studies 2(3):
Pizzuto, Elena and Virginia Volterra
2000 Iconicity and transparency in sign languages: a crosslinguistic-crosscultural view.
In The Signs of Language Revisited: An Anthology to Honor Ursula Bellugi and
Edward Klima, Karen Emmorey and Harlan Lane (eds.), 261-86. Mahwah,
NJ:Lawrence Erlbaum Associates.
Pizzuto, Elena, Enza Giuranna and Giuseppe Gambino
1990 Manual and non-manual morphology in Italian Sign-Language: Grammatical
constraints and discourse processes. In Sign Language Research: Theoretical
Issues, Ceil Lucas (ed.), 83-102. Washington, D.C.: Gallaudet University Press
1990 Introduzione a Peirce. Milano: Bompiani.
2004 Iconicity and productivity in sign language discourse: an analysis of three LIS
discourse registers. Sign Language Studies 4(2): 164-197.
2003 Rethinking “Classifiers” in signed languages. In Perspectives on Classifier
Constructions in Sign Languages, Karen Emmorey (ed.), 3-34. Mahwah, NJ:
Lawrence Erlbaum Associates.
1995 Iconic aspects of syntax. A pragmatic approach. In Iconicity in Language,
Raffaele Simone (ed.), 153-170. Amsterdam/Philadelphia: John Benjamins.
Supalla, Ted and Elissa Newport
1978 How many seats in a chair. The Derivation of Nouns and Verbs in American Sign
Language. In Understanding Language through Sign Language Research,
Patricia Siple (ed.), 91-132, New York: Academic Press.
1973 Explanation in syntax. In Syntax and Semantics vol 2, John P. Kimball (ed.), 1-
50. New York: Academic Press
Wilbur, Ronnie Bring
1987 ASL: Linguistic and Applied Dimensions. Boston: College Hill Press.
1999 The invention and ritualization of language. In The Origins of Language: What
Nonhuman Primates Can Tell Us, Barbara King (ed.), 351-384. Santa Fe NM:
School of American Research Press.
2004 Cognitive iconicity: conceptual spaces, meaning, and gesture in signed language.
Cognitive Linguistics 15-2: 119–147
Wilcox, Sherman and Phyllis Wilcox
1995 The gestural expression of modality in ASL. In Modality in Grammar and
Discourse, Joan L. Bybee and Suzanne Fleischman (eds.), 135-162.
Amsterdam/Philadelphia: John Benjamins.