Iconicity of Telugu Ideophones and Full Word Reduplications
Vasanta Duggirala, Osmania University, India
Lalita Murty, University of York, UK
This paper reports results of analysis based on morpho-syntactic and semantic properties of
ideophones, and Full Word Reduplications (FWRs) occurring in 710 written constructions in
Telugu. It was observed that most of the adverbial ideophones with high semantic specificity, less
grammatical integration, and expressing semantic function of intensification manifest higher levels
of iconicity compared to FWRs. Some of the FWRs involving nouns, adjectives and derived
adverbs whilst expressing the semantic function of augmentation, manifest moderate to low levels
of iconicity because they also encode context-bound meanings of attenuation and approximation.
Drawing on the observation that ideophones and FWRs complement each other in expressing
properties of events, attributes and objects, a tentative scale of iconicity is proposed for Telugu
with a suggestion that it should be validated through future research by making use of spoken and
written corpora of constructions containing ideophones, full and partial word reduplications as
well as echo-words.
Key words: Full Word Reduplication, Iconicity, Ideophones, Semantic functions, Telugu
1.0 Introduction
Iconicity defined as a relationship of resemblance or similarity between linguistic forms and their
meanings has a larger presence than being confined to limited number of onomatopoeic words
expressing imagistic iconicity. Ideophones denoting non-auditory eventualities connected to
vision, movement, inner-bodily feelings emotions (interoception) and other cognitive states are
semantically specific and have been shown to manifest both imagistic and diagrammatic iconicity
(e.g. Beck, 2008; Dingemanse 2012). Cross-linguistic surveys on reduplicated expressions across
languages of the world revealed that by encoding meanings of plurality, collectivity,
intensification, diminution, attenuation etc., reduplications also manifest some degree of iconicity
(e.g. Dingemanse 2015; Kouwenberg and LaCharite 2015; Mattes 2014). In the past, researchers
have suggested that both imagistic and diagrammatic (relational) iconicity play a role in word
formation processes. For example, Waugh (1994:66) stated that many word affinity relations in
English exhibit different degrees of relational iconicity, and that all of them constantly maintain a
delicate and dynamic balancing act that produces a continuum of iconicity and non-iconicity. She
felt that lexical polysemy could contribute to impoverishment of iconicity in some cases. Fischer
(1999: 348) commented that creative (iconic/depictive) and symbolic (descriptive) modes of
2
signification always co-exist in everyday language even as processes of grammaticalization
(involving metaphor or metonymy) work to regulate communication. The idea that Iconicity co-
exists with arbitrariness and systematicity in every language has received empirical support in a
more recent investigation (Dingemanse, Blasi, Lupyan, Christiansen and Monaghan 2015).
Currently, consensus seems to center around the notion that iconicity is indeed a graded
phenomenon involving many lexical classes and hence efforts should be made to operationalize
iconicity empirically for individual languages. In this chapter, we describe one such effort in
relation to Telugu.
Telugu is one of four major literary languages belonging to the Dravidian family primarily spoken
in the two Telugu-speaking states of Andhra Pradesh (A.P.) and Telangana. Telugu has
agglutinative morphology. It’s canonical word order is SOV, although other orders are permitted
in creative writing, and newspaper reportage. Ideophones are abundant in spoken and written
Telugu, both classical and modern. All word classes in Telugu can be reduplicated for conveying
semantic functions of continuity, distributivity, intensification, affection or even pejoration (see
Krishnamurti and Gwynn 1985; Abbi, 1992; Sailaja 2015 for Telugu examples). However,
published works relating to ideophones are limited to one or two dictionaries or compilations (e.g.
Bhaskararao 1977; Usha Devi 2001). There are no on-line corpora of either ideophones or
reduplicated expressions in Telugu. It is against this context, we report this exploratory study based
on morpho-syntactic and semantic analysis of Telugu ideophones and FWRs in 710 constructions
from a Personally Collected Corpus (PCC)1 of printed Telugu texts. This study seeks answers to
two specific questions: (1) Are ideophones more iconic than Full Word Reduplications? (2) Does
iconicity gets distributed across different lexical categories (other than ideophones) in creative
writing? The rest of this chapter is structured along the following main headings: Review of
previous research on iconicity and reduplication (2.0); Iconicity of Telugu ideophones (3.0);
Iconicity of Telugu FWRs (4.0); Summary and Conclusion (5.0).
2.0 Review of select research
Gasser, Sethuraman and Hockeman (2005) have offered operational definitions for the concepts:
absolute iconicity (e.g. onomatopoeia), relative iconicity (form similarity relations positively
correlate with meaning similarity relations as happens in most expressives or ideophones),
arbitrariness (absence of relative iconicity), and anti-iconicity (negative correlation between form
3
similarity and meaning similarity as with most concrete nouns). They proposed a metric termed,
Iconicity Quotient (IC) that takes into consideration phonetic and semantic similarity in the
linguistic forms and their corresponding similarity / dissimilarity in meanings. They applied this
metric to a set of ideophones and concrete nouns extracted from Japanese and Tamil dictionaries
and concluded that in both these languages, ideophones are more iconic than concrete nouns.
Subsequent research briefly reviewed below has highlighted the fact that the interpretation of
meanings of ideophones is context-specific or even construction-specific.
Akita’s (2009) exhaustive treatment of Japanese mimetics resulted in a three-way typology:
• PHONOMIMES including onomatopoeia, and other sound-based expressive elements,
• PHENOMIMES denoting physical action / movement and /or vision based ones, and
• PSYCHOMIMES denoting internal bodily experiences or mental states.
Akita presented extensive empirical support for what he termed, the Lexical Iconicity Hierarchy
(LIH) that posits onomatopoeic words as super-expressives lying on the most iconic end of the
hierarchy, and non-mimetic words on the least iconic side shown below:
Super-expressives > Phonomimes > Phenomimes > Psychomimes > Non-mimetics
Super expressives and phonomimes were considered as highly iconic because sound directly
motivates sound unlike in phenomimes and psychomimes that depend on more than one modality
or encode more abstract psychological and emotional experiences connected to non-auditory
eventualities. Akita also claimed that the highly iconic auditory mimetics tend to belong to
adverbial class while poorly iconic mimetics representing internal bodily feelings tend to be
realized as verbs, adjectives and nouns. Further, abstract concepts that visuo-psych mimetics
encode are often metaphorically or metonymically understood through more concrete physical
experiences that are linked to them. In a detailed discussion of the role of constructions in
determining iconicity (or expressivity) of Japanese mimetic verbs, Akita and Tsujimura (2016: 47)
stated that the creative aspect of mimetic verbs results from a combination of imagery that mimetic
roots induce and the argument structure types in which a given mimetic appears. Further, the most
iconic phonomimes are adverbs and cannot become independent verbs, whereas about half of the
phenomimes and psychomimes in their corpus functioned as verbs. More recently, Dingemanse
and Akita (2017) argued that greater expressiveness goes with less grammatical integration and
4
vice versa in Japanese as well as ten other languages spoken in Africa, Eurasia, Australia, North
and South Americas. They offered evidence for the expressivity of Japanese ideophones using
‘predicate integration hierarchy’:
Quotative < collocational < predicate verbal / nominal
While defending this hierarchy, Dingemanse and Akita (2017:14) pointed out that (1) quotative
ideophones occur in preverbal position; they are syntactically and semantically separated from their
host predicates. They are most expressive since they are not syntactically obligatory. (2) In
collocational constructions, adverbial ideophones which appear in close association with verbs
without quotative markers in different sentential positions, form a tight unit with their host
predicates and therefore are more morpho-syntactically integrated than quotative ideophones, and
(3) the two types of predicative constructions are tightly integrated with the syntax of the sentence,
being syntactically obligatory, they are considered the least expressive. According to their
‘syntactic optionality’ criterion, predicative verbal or nominal ideophones are more obligatory
(therefore less iconic) compared to quotatives and collational ideophones (see Akita 2017 for more
details about typological implications of these findings).
Turning to published research on the topic of reduplication relevant to present discussion,
reduplication has long been considered as a means of word formation that manifests some measure
of iconicity, i.e., forms and meanings resemble each other in a quantitative respect. Kajitani (2005)
analyzed expressions containing reduplications from 16 genetically and geographically distributed
languages for their semantic properties. The results suggested that some aspects of universal
semantic properties of reduplication do indeed reflect iconicity and common perceptual
experiences of speakers across all 16 languages. He proposed a hierarchy of meaning properties
for reduplicated expressions:
Augmentation Intensification [Attenuation Diminution]
Referring to the bold arrows in this implicational hierarchy, Kajitani (2005:94) stated that
Augmentation (AUG.) defined as increase in quantity of objects, participants, events is preferred
over Intensification (INT.) or increase in quality or degree of an action/feeling etc. in all the
languages investigated. If Intensification is expressed by means of reduplication, then
5
Augmentation can also be expressed via reduplication. Similarly, in all languages, if Diminution
(DIM.) or decrease of quantity is expressed by means of reduplication, then attenuation (ATT.) or
decrease of quality or degree can also be expressed through reduplication. The light arrow in the
middle indicates (language-specific) statistical preference. These results suggest that cross-
linguistically, iconicity of reduplicated expressions is determined primarily by the application of
‘quantity principle’ that holds, ‘more of the same form motivates more of the same content’.
Lai (2006: 486) pointed out one of the fundamental and unresolved problems relating to iconicity
of reduplicated forms: how can reduplication be associated with both intensification and
diminution / attenuation? Using data from Hakka reduplications, Lai argued that reduplication
should be treated as one of the strategies speakers adopt to share their conceptualization of the
world with others in the community, and that it is not an all-or-none phenomenon, instead, it
constitutes a scale. Lai pointed out that seemingly contradictory senses of intensification vs.
diminution / attenuation in reduplicated forms can be interpreted as iconic by drawing on
contextual factors. More recently, Rozhansky (2015) proposed two semantic patterns / principles
underlying iconicity, viz., (1) Similarity principle (expression of likeness, change of lexical class,
and pejorativity), and (2) Quantity principle (expression of plurality, continuity, distributivity). He
argued that these two patterns get combined in reduplicative expressions that encode attenuation
and emphasis typically found in adjectives. He cites Telugu data in support of this point. Kallergi
(2015) presented evidence from Modern Greek that supports the need to distinguish Total
Reduplication (TR) from emphatic or pragmatic repetition in theorizing linguistic expressivity. He
stressed that TR as an expressive device has three-fold function: creating a pragmatic effect,
encoding emotive/affective meanings, and serving social function. Kouwenberg and LaCharite
(2015) discussed iconicity of reduplicated expressions in Caribbean Creole (CC) languages and
identified certain gaps and constraints that lessen the level of iconicity expressed by different
grammatical classes depending on how they denote properties of events (verbs), attributes
(adjectives) and objects (nouns). Published works on this topic in relation to languages of India
are confined mostly to description of their phono-semantics of ideophones, and morpho-syntactic,
semantic and areal-typological features of reduplications (see for e.g. Apte 1968; Abbi 1992;
Bhaskararao 1977; Sailaja, 2015; Usha Devi 2001). Recently, Duggirala (2016) reported a study
involving 50 native speakers who rated Telugu ideophones and reduplicated forms occurring
inside sentences on a 5-point scale. They gave higher ratings to auditory, vision and movement
6
based ideophones and reduplications that expressed quantity changes compared to interoception
ideophones and reduplications expressing similarity relations. The differences in these ratings
were statistically significant.
3.0 Iconicity of Telugu ideophones
Of the 710 constructions analyzed in the PCC, there were 400 ideophone tokens belonging to 70
types. The ideophone tokens were classified into two major classes based on their structure and
function: In the first class, there were 90 ideophonic roots with the following syllable shapes: CVX
where X stands for a nasal /n/ or a coda final obstruent or sonorant in gemination (CVC / CVCC /
CVCVCC). The syllable structure of this class of ideophones clearly violates Telugu phonotactic
rule of not having coda-final consonants other than /m/ (either in singleton or as geminate form).
These ideophones represent natural sounds from the environment, punctual actions, feelings and
states lasting for a very brief duration. They appear inside sentences with the help of a vowel plus
an increment -ma, always followed by quotative markers, -anu or -na ‘say’ as in examples 1-3
(with ideophone type, # of tokens, source and construction # in parentheses):
(1) talupu kirru-mandi (Auditory ideophone: 35 tokens, JB stories, C #4)
door IDEO-creak.PST-AGR
The door creaked
(2) loolaakulu taɭukku-mannaayi (Vision ideo., 40 tokens, PYR stories, C # 134)
ear-rings IDEO-flash.PST-AGR
‘There is a bright flash of the earrings’
(3) naaku curruna manɖindi (Intero. Ideo., 15 tokens Dilawar stories, C # 306)
I-DAT IDEO burn.PST-AGR
‘I got angry’
These ideophones are akin to the non-reduplicated (extra-linguistic) ideophones in Marathi
discussed by Apte (1968) or the super-expressives mentioned by Akita (2009). One of the
characteristic feature of this class of ideophones is, they always appear in preverbal position
accompanied by quotative marker, ‘say’ and hence manifest highest level of iconicity.
The second class consisted of 310 tokens of CVCV ideophones. They were classified under three
groups based on concrete vs abstract nature of experiences they encode. Thus, auditory ideophones
7
depict single modality-based experience of hearing, whereas Vision + Movement ideophones draw
on both vision and hearing (movement usually accompanies sound) and interoception based
ideophones are even more abstract since they represent inner bodily feelings and emotions.
Table -1 presents ten examples for each of the three categories of CVCV ideophones. There were
20 types of Auditory ideophones, 30 types of Vision + Movement ideophones, and 20 types of
Interoception or inner bodily feelings and emotions. In the last category, there were ten tokens
depicting inner bodily sensations, and ten denoting emotions.
Table-1: Examples of CVCV ideophones in Telugu
Auditory
Ideophones
Types: 20
Tokens: 65
Vision + Move-
ment ideophones
Types: 30
Tokens: 160
Interoceptive
Ideophones
Types: 20
Tokens: 85
paʈa paʈa ‘biting’ bhaga bhaga ‘burning’ gadza gadza ‘shiver’
wala wala ‘crying’ mila mila ‘shine’ guba guba ‘anxiety’
kara kara ‘chewing’ taɭa taɭa ‘shine’ dzila dzila ‘itch’
ʈaka ʈaka ‘knocking’ misa misa ‘shine’ naka naka ‘hunger’
pheɭa pheɭa ‘breaking’ boʈa boʈa ‘tears’ wila wila ‘pain’
gala gala ‘gushing water’ gora gora ‘dragging’ rusa rusa ‘anger’
ʈapa ʈapa ‘falling rain-
drops’
caka caka ‘fast walk’ taha taha
‘enthusiasm’
paka paka ‘loud laughter’ guna guna ‘slow walk’ Wela wela
‘embarrassment’
kica kica ‘birds chirping’ dzala dzala ‘falling’ cura cura
‘angry look’
gusa gusa ‘whispering’ repa repa ‘flutter’ guba guba ‘sadness’
It may be noted that vision and movement ideophones with 160 tokens have out-numbered the
auditory ideophons (N=65) and interoceptive ideophons (N=85) both in types and tokens. There
was only one smell ideophone (ghuma ghuma) and one touch ideophone (gara gara) which are
included in the second category. The CVCV ideophones in Telugu do not possess simplex forms
and the reduplicant is always a syllable, not a lexeme. Meanings of these ideophones are highly
specific to the actions /states being denoted. For example, the vision-based ideophones, mila mila;
taɭa taɭa, misa misa denote shininess associated with different objects: mila mila refers to ‘shining
stars’ which can get metaphorically extended to ‘shiny eyes’; taɭa taɭa denotes ‘shining pots and
8
pans’ or ‘shiny jewels’ whereas, misa misa denotes ‘shiny skin’ associated with youthful looks.
misa misa is never used for denoting shine associated with pots and pans, jewels or stars. Since
high semantic specificity prevents an ideophone from participating as an independent predicate,
we treated those ideophones that cannot occur as independent predicates as less iconic than those
that occur in pre-verbal position and have some degree of syntactic freedom.
Bhaskararao (1977:12) stated that the following forms or sequence of forms enable CVCV
ideophones to participate in sentences:
• Vowel length
• Plural suffix, -lu followed by the verb, -aaɖu ‘do’
• Increment -ma followed by the verb -anu ‘say’
We have determined iconicity levels of 400 ideophone tokens by applying Dingemanse and
Akita’s (2017) grammatical integration criterion by first observing the category to which a given
ideophone belongs inside each construction. The results are displayed in Table -2:
Table -2: Grammatical categories of ideophone tokens
Ideophone type
Category
Auditory
Tokens: 100
Vision+Movt.
Tokens: 200
Interoception
Tokens: 100
Quot.-Adverbial 55 55% 50 25% 20 20%
Bare Adverbial 30 30% 95 47.5% 20 20%
Verbal 10 10% 50 25% 45 45%
Nominal-adjectival 05 5% 05 2.5% 15 15%
Looking at the percentage values listed in Table-2, it is evident that more than 80% of auditory
ideophones occur as quotative-adverbials and bare adverbials (in collocational constructions); over
60 % of vision + movement ideophones appear as quotative-adverbials or bare adverbials, whereas,
only 40% of interoception ideophones belong to these two categories. Since interoception
ideophones mostly belong to verbs (45%), they serve as predicates on their own depicting specific
experience of emotions or psychological states. Based on these facts, we argue that interoception
ideophones being grammatically well integrated into the syntax of the sentence are not as iconic
as the other two types. Auditory, vision + movement ideophones in Telugu are more iconic going
also by the quantity principle of iconicity, that is, together they encode semantic functions of
9
Intensification (168 tokens) and Augmentation (45 tokens) compared to interoception ideophones
that encode only intensification (75 tokens), but not augmentation. Diminutivity is conveyed by
very few ideophones. The fact that higher level of iconicity associated with the first two types of
CVCV ideophones can be accounted for by quantity principle is illustrated through examples 4
and 5 below (with information about the category, source and construction number in that order):
(4) komma pheɭa pheɭamanʈuu wirigindi (Quot.-adv., Dilawar stories, C # 60)
branch IDEO-say break.PST-AGR
‘The branch broke making the sound, pheɭa pheɭa’
(5) kotta ruupaayi nooʈulaa pheɭa pheɭa-laaɖutunnaaɖu (Pred.verb., JB stories # 98)
new rupee note like IDEO-crisp.PST-AGR
‘He is looking crisp like a new rupee note’
The semantic function of augmentation manifests in different languages in terms of repetition /
iteration, frequentativity, plurality, variety, distributivity, quantification, collectivity and so on.
Auditory ideophones in Telugu typically express augmentation only with respect to repetition or
iteration of punctual action as evident in example-4 involving the ideophone, pheɭa pheɭa (sound
made when a dry branch breaks). It can be omitted without making the sentence ungrammatical.
However, when it gets extended metaphorically to the crispness of a new rupee note or to the
stiffness of the starched shirt worn by someone as in example-5, it becomes obligatory because it
is in a predicate-verbal construction. The next two examples of vision-based and interoception
based ideophones respectively express intensification meanings also manifest higher level of
iconicity going by quantity principle of increase in the quality or degree of an action (e.g. 6) or
feeling (e.g. 7):
(6) dzuʈʈu paʈʈukoni gora goraa iiɖcukoccindi (bare adv., JS stories, C # 208)
hair holding IDEO-forcefully drag.PST-AGR
‘(she) dragged her by holding her hair’
(7) bharta waipu cura curaa cuusindi (bare adv., IJB stories, C # 337)
husband towards IDEO look.PST-AGR
‘She looked at her husband with intense anger’
10
In example -6, the vision/movement based bare-adverbial ideophone, gora goraa is used to express
manner of motion caused when a person is dragged forcibly. The interoceptive ideophone, cura
curaa in example- 8 depicts intense anger. Both are occurring in collocational constructions.
These ideophones are syntactically optional in that they can be omitted without making the
sentences ungrammatical (since they are functioning as bare adverbs). The results of our analysis
of Telugu ideophones with respect to syntactic optionality as a criterion for determining the level
of iconicity of ideophones can be expressed as follows:
Quotative –ani & Collocational – aa (optional) < Predicative verbal – aaDu (obligatory)
Auditory ideophones < Vision + Movement ideophones < Interoception ideophones
Overall, the results based on ideophones in Telugu offer support to the cross-linguistic
generalization reported in Dingemanse and Akita (2017) and Akita (2017). There are very few
Telugu ideophones that serve as nouns or adjectives. When ideophonic nouns (e.g. gala-gala-lu
with a plural marker) appear in sentence final position in newspaper language, their meanings
become conventionalized and hence they tend to lose their iconicity. Lessening of iconicity also
results when the ideophones are used metaphorically as headlines in newspapers (e.g. nagaram
gadza gadza ‘the city is shivering’). Interoception ideophones appeared in figurative language
involving metonymy and metaphors (20 tokens in our corpus).
4.0 Iconicity of FWRs in Telugu
There are at least four types of reduplicated expressions in Telugu: (1) Total Reduplication
(complete copy of the base form nela ‘month’ in reduplicated form in nela nela ‘every month’),
(2) Partial Word Reduplication involving a portion of the base being copied into the second lexeme
(e.g. iʈuu aʈuu- ‘this way-that way’), (3) Reduplicated Compounds (e.g. kuuraa-naaraa
‘vegetable-fibers’ to mean vegetables in general), and (4) Echo-words (illu-gillu ‘house and stuff).
Partial word reduplications and echo-words tend to occur more commonly in colloquial spoken
registers carrying evaluative-pragmatic (general) meanings (See Abbi 1992 for details). For the
purposes of this study, we have selected only Full Word Reduplications and not the other 3 types.
Researchers have suggested that to be fully iconic, the meaning of the simplex form must be
preserved completely in the reduplicated form. In Telugu, even total reduplication in some cases
11
involves certain morpho-phonemic changes due to presence of affixes or sandhi. Therefore, we
have considered FWRs with three degrees of transparency between the base and the reduplicant.
(a.) FWR (W1=W2) type (128 tokens): Here the reduplicant is a complete copy of the base with
absolutely no change happening due to reduplication. For example, color adjective nalla-nalla
‘black-black’ or derived adverb, parigetti-parigetti ‘having run for a long time’.
(b.) FWR (W1-W2 + affix) type (132 tokens): here the affix can come immediately after the base
(e.g. waanalee-waanalu ‘excess rains’ with emphatic marker -ee sitting between the base and the
reduplicant) or more commonly, it occurs as a suffix following the reduplicant (e.g. weeɖi
weeɖigaa ‘very hot’ with adverbial suffix -gaa).
(c.) FWR (W1-W2 + sandhi (50 tokens): In this type, the process of reduplication alters the base
form thereby introducing some degree of opacity to it (e.g. appuɖappuɖu ‘once in a while’ in which
appuɖu ‘then’ when reduplicated loses the final vowel /u/ in the base; The expression ikkaɖikkaɖee
‘somewhere right here’ formed when ikkaɖee ‘here’ undergoes total reduplication. In this item, in
addition to sandhi process involving vowel change, an emphatic marker -ee appears as a suffix at
the end of reduplicated form).
In terms of form-classes, the 310 FWR tokens comprised of nouns (62) adjectives (69), adverbs
(36), verbal adverbs (47), and derived adverbs (82). There were only ten pronouns and four verbs
in the data. The first (a) type of FWR (W1=W2) has more verbal adverbs and adjectives; the second
type has derived adverbs and nouns and the sandhi type has more nouns and some adjectives.
We have attributed higher levels of iconicity to FWRs dealing with the semantic functions of
augmentation or intensification. FWRs with evaluative/pragmatic meanings associated with
diminution, attenuation and approximation were considered to have lower (diminished) levels of
iconicity since they combine both quantity and similarity principles for interpretation in addition
to involving peripheral meanings as argued by Rozhansky (2015).
The four major semantic functions of FWRs are indicated as number of tokens with % values in
parenthesis in Table -3. It is clear from this table that irrespective of the type of reduplication, most
FWRs in Telugu contribute to augmentation function (183 tokens out of 310). Less than 50 FWR
tokens express intensification of meanings suggesting that the first two types of FWRs in Telugu
12
are iconic by Quantity principle. Diminution is expressed by limited number of FWRs (W1=W2)
types only (e.g. cinna cinna ‘small-small’ or koddi-koddi ‘little-little’ which are also used for
conveying politeness serving social rather than referential function). FWRs with affix (b-type)
and sandhi (c-type) also express attenuation function to some extent, and yet manifest somewhat
reduced iconicity as we show through the specific examples and discussion following Table -3.
Table -3: Major semantic functions of FWRs
Sem. function
FWR Type AUG. INT.
DIM.
ATT.
FWR (W1=W2)
128 76
(59.3%)
25
(19.37%)
23
(18.60)
4
(3.1%) FWR + Affix
132 81
(61.3%)
20
(15.15%)
4 (3%) 27
(20.4%) FWR with sandhi
50 26
(52.0%)
0
0 24
(48%)
The FWRs in examples 8, 9 and 10 below manifest high to moderate levels of iconicity because
they are encoding semantic functions of augmentation and intensification supported by the
quantity principle. It should be noted that in each example, the FWR is optional in that its omission
does not make the sentence ungrammatical.
FWR (W1=W2) Augmentation (plurality) – 76 tokens
(8) aa roodzu leestuunee pedda pedda keekalu weesaaɖu (JB stories, C # 433)
that day wake up-PTCP FWR big-big shouts do.PST-AGR
‘He started shouting loudly as soon as he woke up that day’
FWR (W1=W2) Augmentation (distributivity)
(9) waaɭɭu waaram waaram wastaaru (ORB stories, C # 460)
they FWR week-week come.FUT-AGR
‘They are going to come week after week’
FWR (W1=W2) Intensification – 25 tokens
(10) airhostess weeɖi weeɖi ʈii iccindi (ISD stories, C# 481)
Airhostess FWR hot-hot give-PST-AGR
‘The airhostess gave very hot tea’
13
However, if the meaning of intensification is expressed through derived adverbs instead of
adjectives as in example-10, the level of iconicity is likely to get diminished due to change of
grammatical class. Same is true when the reduplication involves grammaticalized aspectual uses
or metaphors, an observation that calls for more in-depth investigation.
Turning to evaluative-pragmatic meanings associated with FWRs, we present below FWRs that
encode attenuation and approximation meanings manifesting low level of iconicity. Examples 11
& 12 contain FWRs with low level of iconicity because reduplication is contributing to
attenuation of meaning.
FWR (W1-W2) with suffix, Attenuation – 27 tokens
(11) wanʈloo cali caligaa anipincindi (IJB stories, C # 643)
body-LOC. FWR cold-cold feel.PST-AGR
‘My body felt somewhat cold’
FWR (W1-W2) with sandhi, Attenuation – 24 tokens
(12) naanna akkaɖikakkaɖee woɭɭu teliikunɖaa paɖipooyaaru (PYR storis, C# 688)
father FWR there-there unconscious fall_PST-AGR
‘Father dropped down unconscious right at that very place’
The FWRs in the above two examples are somewhat obligatory in that while deletion of
reduplicated form does not make the sentence completely ungrammatical, it calls for a secondary
interpretation of meaning. To elaborate, in example 11, reduplicated form cali cali-gaa an adverb
derived from the noun cali ‘cold’ could mean ‘very cold’ or ‘somewhat cold’ depending on the
context relating to the event being described. There are many such forms involving derived
adverbial marker -gaa that seems to reduce the quality of given property (color) or experience.
Constructions containing simplex forms (for e.g. caligaa) is not as ambiguous as the one with
reduplication. The FWR in example 12 expresses ‘exactness’ meaning that combines both
semantic and pragmatic features amenable for analysis in radial category models (e.g. Jurafsky
1996).
The fact that a majority of FWRs are encoding augmentation function as opposed to ideophones
which for the most part seem to be expressing intensification function requires some explanation.
14
This difference could be due to differences in the inherent semantic properties that promote or
curtail their application to events, attributes and objects (see Table-4):
Table – 4: Semantic properties encoded by Ideophones and FWRs
IDEOPHONE
TYPES
EVENTS
Punctual Non-
punctual
ATTRIBUTES
Gradable Non-
Gradable
OBJECTS
Count Non-
count
Auditory + + -- -- -- -- Vision + Movt. + + -- + -- + Interoception + + -- -- -- -- FWR
(W1=W2) + + + + + + FWR Affix + + + + + + FWR Sandhi + + + + + +
One striking fact evident from Table -6 is, while all three types of ideophones encode punctual and
non-punctual events, there are many gaps with respect to attributes and objects. Vision ideophone
taɭa taɭa can describe the shininess attribute of vessels; non-count object denoting vision
ideophone, kiʈa kiʈa applies to crowds encoding augmentation via increase in participants.
However, there are aspects of augmentation involving distributivity, variety etc. that ideophones
are incapable of depicting. In contrast, lexical items undergoing full word reduplication in Telugu
express a greater range of semantic properties tied to events, attributes and objects when compared
to ideophones which seem to be confined mostly to events.
It should be noted that there were as many as 183 of the 310 FWRs in our corpus expressing
augmentation relating to properties of events, attributes and objects. Among the first type of Total
reduplication, augmentation is carried out mostly by verbal adverbs and adjectives. In
reduplication with suffix category, augmentation involves derived adverbs and nouns and the third
category of reduplication with sandhi includes numeral adjectives. In the sandhi category, adverbs
of place and time express attenuative meanings. Some adverbs of place such as mundu ‘front’ have
been grammaticalized to denote future in their reduplicated form, mundu-mundu ‘near future’ or
wenaka ‘back’ in its reduplicated form, wenaka-wenakki denotes past. In these cases, the simplex
forms and the corresponding reduplicated forms convey different, and somewhat unrelated
meanings manifesting lower levels of iconicity. This is another topic that calls for more research.
15
The iconicity levels we have attributed via quantity principle is distributed across the three FWR
types such that going by number of tokens, FWR (W1=W2) with 123 tokens and FWR Affix with
105 tokens manifest relatively high iconicity in contrast to the third type, FWR -Sandhi which has
only 26 tokens. Only FWR (W1=W2) can encode events (46 tokens) of which 30 tokens express
punctual events (e.g. koʈʈi koʈʈi ‘repeated hitting’) and the rest non-punctual events (e.g. eeɖci
eeɖci ‘having cried a lot’). FWRs (W1=W2) express both gradable properties (34 tokens; e.g.
weeɖi weeɖi ‘very hot’) and non-gradable attributes (15 tokens; e.g. maɭɭi maɭɭi ‘again and again’).
The object-denoting FWRs across all three types express countability (59 tokens) thereby
expressing augmentation through plurality. These results lend support to the idea that there is a
degree of complementarity with respect to the way relational iconicity is shared by different lexical
items in Telugu. We propose that he level of iconicity is distributed among FWRs as shown in the
hierarchy (from Highest to Lowest iconicity):
FWR (W1=W2) < FWR (W1-W2) Affix < FWR (W1-W2) Sandhi
5.0. Summary and Conclusion
This chapter described an investigation undertaken to examine how iconicity gets distributed
among 400 Telugu ideophones and 310 FWRs embedded in written constructions. Drawing on
cross-linguistic generalizations offered by previous researchers (Akita, 2009, 2017; Dingemanse
and Akita 2017 and Kajitani, 2005), we have performed morpho-syntactic and semantic analysis
to find answers to two specific questions: (1) Are ideophones more iconic than Full Word
Reduplications (FWRs)? (2) Does iconicity gets distributed across different lexical categories in
creative writing? The results reported in sections 3.0 and 4.0 suggest that the answers to both these
questions are affirmative. The operationalization of iconicity levels in Telugu ideophones and
FWRs in this study made use of the parameters mentioned below:
Categorial status and grammatical integration: The auditory and vision + movement
ideophones are more iconic than the Interoception ideophones because, majority of them belong
to quotative adverbial or bare adverbial classes exhibiting more syntactic freedom than
interoception ideophones majority of which are verbs and therefore deeply integrated into the
syntax of the sentences (in literal and figurative language) in which they occur.
16
Quantity vs. Similarity principles: The three types of ideophones expressing semantic function
of intensification received higher levels of iconicity since they adhere to quantity principle alone.
Whereas, interpretation of context-bound meanings encoded by most of the FWRs with affix and
sandhi require combination of quantity and similarity principles and hence manifest lower levels
of iconicity.
Semantic functions: Among the 310 FWR tokens, it was type-a in which the base is fully copied
without any changes that was found to be most iconic since they express semantic function of
augmentation. FWR + Affix and FWR + sandhi categories were shown to encode meanings of
attenuation and approximation with somewhat diminished levels of iconicity.
Depiction vs. description: By encoding intensification of meanings, the ideophones seem to
function in a depictive mode of signification, whereas, the FWRs with a wide range of distribution
of semantic properties relating to events, attributes and objects have the potential to function in a
descriptive mode as well, a finding that offers support to Dingemanse’s (2015:961) observation
that reduplication could bridge depiction and description. As a way of conclusion, we offer a
tentative scale of iconicity (see Figure-1) that needs to be validated through future research.
Figure-1: Tentative Scale of iconicity for Telugu
Note 1: The Personally Collected Corpus (PCC) of ideophones and reduplicated forms in Telugu was developed as
part of an investigation funded by the University Grants Commission, New Delhi under their ‘Emeritus Fellow’
scheme sanctioned to Vasanta Duggirala during 2015-2017. E-mail of corresponding author: [email protected]
ICONICITY: HIGH
ICONICITY: MEDIUM ICONICITY: Ideophones
• Auditory
• Vision + movement
• Interoception FWR
• W1=W2
FWR
• W1~W2+Affix FWR
• W1~W2+Sandhi
17
Abbreviations used:
AGR agreement
DAT dative case
FUT Future
LOC Locative
PST past
PTCP participle
REFERENCES
Abbi, A. 1992. Reduplication in South Asian Languages: An Areal, Typological and Historical
Study. New Delhi: Allied Publishers.
Akita, K. 2009. A grammar of sound symbolic words in Japanese: Theoretical approaches to
iconic and lexical properties of mimetics. Doctoral thesis, Kobe University Repository. URL:
http://www.lib.kobe-u.ac.jp/handle_kernel/D1004724
Akita, K. 2017. “The linguistic integration of Japanese ideophones and its typological
implications”. Canadian Journal of Linguistics 62:2, 314-334.
Akita, K. and Tsujimura, N. 2016. “Mimetics”. In Handbook of Japanese Lexicon and Word
formation, T. Kageyama and H. Kishimoto (eds), 133-160. Mouton: DeGruyter.
Apte, M.L. 1968. Reduplication, Echo Formation, and Onomatopoeia in Marathi. Poona:
Deccan College.
Beck, D. 2008. “Ideophones, adverbs and predicate qualification in Upper Necaxa Totonac”.
International Journal of American Linguistics 74:1, 1-46.
Bhaskararao, P. 1977. Reduplication and Onomatopoeia in Telugu. Poona: Deccan College.
Dingemanse, M. 2012. “Advances in cross-linguistic study of ideophones”. Language and
Linguistics Compass 6, 654-672.
Dingemanse, M. 2015. “Ideophones and reduplication: Depiction, description and the
interpretation of repeated talk in discourse”. Studies in Language 39(4): 946-970.
Dingemanse, M., Blasi, D. E., Lupyan, G., Christiansen, M.H. and Monaghan, P. 2015.
“Arbitrariness, iconicity, and systematicity in language”. Trends in Cognitive Sciences 19:10,
603-615,
18
Dingemanse, M. and Akita, K. 2017. “An inverse relationship between expressiveness and
grammatical integration: On the morphosyntactic typology of ideophones with special reference
to Japanese”. Journal of Linguistics 53: 501-532.
Duggirala, V. 2016. On the iconicity of ideophones in Telugu. Poster presented at the
International symposium on ‘Mimetics in Japanese and other languages of the world’ held during
Dec. 17-18. Tachikawa city, Japan.
Fischer, O.C.M. 1999. “On the role played by iconicity in grammaticalization processes”. In
Form Miming Meaning: Iconicity in Language and Literature O.C.M. Fischer and M. Nanny
(Eds.), (PP. 345-374). Amsterdam: John Benjamins.
Gasser, M., Sethuraman, N., and Hockema, S. 2005. “Iconicity in expressives: An empirical
investigation”. In Experimental and Empirical Methods, S. Rice and J. Newman (Eds.). Chicago:
CSLI.
Jurafsky, D. 1996. Universal tendencies in the semantics of the diminutive. Language 72:3, 533-
578.
Kajitani, M. 2005. “Semantic properties of reduplication among the world’s languages”. LSO
working papers in Linguistics 5: Proceedings of WIGL 93-106.
Kallergi, H. 2015. “Total reduplication as a catergory of expressive: (counter) evidence from
Modern Greek. Studies in Language 39 (4): 873-904.
Kouwenberg, S., and LaCharite, D. 2015. “Arbitrariness and iconicity in total reduplication:
Evidence from Caribbean creoles”. Studies in Language 39:2, 971-991.
Krishnamurti, Bh., and Gwynn, J.P.L. 1985. A Grammar of Modern Telugu. New Delhi: O.U.P.
Lai, H-L. 2006. “Iconic coding of conceptualization: Hakka reduplicative constructions”.
Language and Linguistics 7 (2): 483-499.
Mattes, V. 2014. Types of Reduplication: A Case Study of Bikol. Berlin: Mouton de Gruyter
Rozhansky, F.I. 2015. “Two semantic patterns of reduplication: Iconicity revisited”. Studies in
Language 39 (4): 992-1018.
Sailaja, P. 2015. “Telugu”. In Edinburgh Handbook of Evaluative Morphology, N. Grandi and L.
Kortevelyessy (eds), 321-332. Edinburgh: Edinburgh University Press.
Usha Devi, A. 2001. A Dictionary of Onomatopoeic Words in Telugu. Hyderabad: P.S. Telugu
University.
Waugh, L. 1994. “Degrees of iconicity in the lexicon”. Journal of Pragmatics 22, 55-70.