Evans, N & D. Wilkins (2000) - in The Mind's Ear. The Semantic Extensions of Perception Verbs in Australian Languages
Evans, N & D. Wilkins (2000) - in The Mind's Ear. The Semantic Extensions of Perception Verbs in Australian Languages
Evans, N & D. Wilkins (2000) - in The Mind's Ear. The Semantic Extensions of Perception Verbs in Australian Languages
In the Mind's Ear: The Semantic Extensions of Perception Verbs in Australian Languages
Author(s): Nicholas Evans and David Wilkins
Source: Language, Vol. 76, No. 3 (Sep., 2000), pp. 546-592
Published by: Linguistic Society of America
Stable URL: http://www.jstor.org/stable/417135
Accessed: 07-03-2017 12:15 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/417135?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Linguistic Society of America is collaborating with JSTOR to digitize, preserve and extend access to
Language
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR:
THE SEMANTIC EXTENSIONS OF PERCEPTION VERBS IN AUSTRALIAN
LANGUAGES
* Evans wishes to thank the University of Melbourne for study leave, MPI Nijmegen for furnishing an
ideal work environment for writing the first draft of this paper, the Alexander von Humboldt-Stiftung for
supporting his research in 1997-98, and the Australian Research Council (Large Grants: Polysemy and
Semantic Change in Australian Languages; Analysing Australian Aboriginal Languages) for its financial
support of relevant fieldwork. Wilkins would like to thank the Max Planck Society for the funding of
annual field trips in the period 1992-97. Earlier versions of the work were presented at the Institut fUir
Sprachwissenschaft, U. K61n, the Institut fUir Romanistik, U. Freiburg, the Laboratoire de Dynamique du
Langage, Universit6 Lyon 2, the departments of linguistics at the Australian National University, University
of New England, U.C. Berkeley, and U. Melbourne; we thank participants in those seminars for their useful
comments. We are also grateful to Felix Ameka, Melissa Bowermann, Gavan Breen, Eve Danziger, Bob
Dixon, Nick Enfield, Murray Garde, Cliff Goddard, Jean Harkins, John Haviland, Penny Johnson, Sotaro
Kita, Mary Laughren, Steve Levinson, Bill McGregor, Andrew Mirtschin, David Nash, Nick Reid, Eva
Schultze-Berndt, Peter Sutton, Eve Sweetser, Anna Wierzbicka and Diana Young for useful discussions,
comments and data, as well as to two anonymous referees for Language. Most importantly we thank the
speakers who have taught us about various Australian languages mentioned here: the Arrernte speakers
affiliated with the Yipirinya School and Intelyape-lyape Akaltye project in Alice Springs (esp. Margaret
Heffernan); Netta Loogatha, tDarwin Moodonuthi, and Paula Paul (Kayardild); Alice Bohm and tJack
Chadum (Dalabon), tDavid Karlbuma (Dalabon and Kune), and Mick Kubarkku (Kuninjku).
546
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 547
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
548 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 549
2. POLYSEMY AND SEMANTIC CHANGE: SOME ASSUMPTIONS AND METHODS. It has be-
come a standard assumption that semantic change from meaning A to B normally
involves a transitional phase of polysemy where a form has both meanings (Wilkins
1981, 1996, Sweetser 1990, Heine 1997:82). Less often articulated is that this phase
of polysemy (what Heine calls the stage of overlap) is typically preceded by a phase
where meaning B is only contextually implicated but not yet lexicalized as a distinct
2 We are not the first to make this observation. Hercus (1992:42), for example, writing about the Wemba-
Wemba verb nyernda 'to know, to understand', formally related to nyerna 'to sit, to listen, to hear, to
remember', notes that 'this derivation, implying that 'hearing is knowing' is common in Australian languages
and contrasts with the Indo-European method of expression "I have seen", "I know"'
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
550 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 551
In contrast, the 'see' verb reconstructable for proto-Australian as *na- (with develop-
ment to the laminal-initial form *NHaa- in proto-Pama-Nyungan, Evans 1988) has a
clear development to 'know' in only one language in the extreme south, Kaurna; the
development to 'think' in Guugu Yimidhirr may be mediated by the 'hear' meaning it
also develops there. Elsewhere *na- retains its visual sense or develops in the direction
of such meanings as 'find'.3 See Table 2.
Table 3 shows the etymological set for pIE *k^leu-, *k^leu-s- 'hear'. Although there
are many individual examples in Australia where 'hear' extends to 'think' and 'know'
(see ?4.3), we have not yet identified a 'hear' etymon with wide attestation in Australia,
and so cannot show a fully comparable etymological set demonstrating the different
pattern of extension. However, examination of proto-Pama-Nyungan *pina 'ear' and
3 Sources for the languages cited, and their geographical locations on the continent, are given at the end
of the article.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
552 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
its derivatives, which are often verbs meaning 'hear/listen', illustrates the frequency
with which these cognitive meanings develop across the etymological set. See Table
4.
Our discussion of bridging contexts predicts that such systematically different patter
ings in polysemy and etymology would reflect differences in cultural traditions. He
we face the broader task of gathering, and contextualizing, attestations in differe
languages and language areas; this is particularly important for typological work tha
depends on a large data base to show recurrent regularities and implicational relation
ships. We know from studies of other lexical domains that polysemy exhibits stro
areal patterning in Australia--sometimes at the level of the whole continent as oppo
to elsewhere in the world, and sometimes at more local levels, such as the Lake Ey
Region (Austin et al. 1976) or the Cairns Rainforest (Sear 1995). Where relevant we w
discuss the areal distribution of patterns, to avoid the pitfall of projecting an 'Austral
pattern', which may in fact be more local. Nonetheless, it turns out that most of th
patterns we discuss are Australia-wide rather than being found in specific areas, exce
for the 'see-hear' polysemy, which is largely confined to Cape York.
As well as polysemy proper, we will also investigate semantic extensions accompan
ing derivation, such as change of gender or reduplication. Strictly speaking this is
HETEROSEMY (Lichtenberk 1991)-a relation in which related (often identical) form
and their different, but related, senses belong to different morphosyntactically dete
mined grammatical categories. In polysemy, there is one lexeme with several relat
senses, in heterosemy there are two or more related lexemes each with a sense tha
clearly shows semantic affinity. An example of PURE (zero or underived) HETEROSEM
is found in Yidiny (Dixon 1991). The root bina as a nominal means 'ear, gill on fish
but as a particle it means 'I thought something was the case, but it is not'. In additio
there is a verb bina (in the N-conjugation) which means 'hear, listen to, think abou
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 553
Against this background, he sets up a five-by-three grid arraying the five main percep-
tual modalities against three general EVENT-TYPE representations of perception: as con-
trolled activity ('she looked at the painting', 'he felt his daughter's brow for signs of
fever'), as noncontrolled experience ('she saw the painting', 'he felt blood running
down inside his shirt'), and as a source-based copulative (state) construction from
which the perceiver is omitted ('the painting looked very old', 'his daughter's brow
4 Viberg did use a few published sources to glean some unsystematic lexical data for a couple of Australian
languages, but he did not gather any information on full systems, and does not count such languages in his
typological base of 53 languages. He acknowledges (1984:124) that, 'although this is a fairly good sample,
it is not satisfactory, since European languages are overrepresented and some areas, such as North and South
America and Oceania, are highly underrepresented'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
554 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
Needless to say, the above set contains only the most basic verbs, and these may
have a considerable number of hyponyms: 'look at', in English, has the hyponyms
'peer at', 'peep at', 'stare at' and many others. Basic perception verbs in Australian
languages also often have many hyponyms. Thus in Kayardild kurrija 'see, look at'
has such hyponyms as walmurrija 'look up in the sky', warayija 'look back', rimarutha
'look eastwards at', among others (Evans 1992b:326). We restrict our focus, as did
Viberg, to the basic set of general superordinate verbs; what Dixon (1972), on the basis
of Australian data, has identified as nuclear (as opposed to non-nuclear) verbs (cf. ?5).
Another limitation on the data, in our study as in Viberg's, is the simplifying assump-
tion that there are merely five sensory modalities. In fact, a good case can be made
for at least one further modality: PROPRIOCEPTION, or internal feeling, as opposed to
external touch. This sixth modality is expressed distinctively in many Australian lan-
guages. Thus, among the set of basic perception verbs in Arrernte we find welheme
'have a (proprioceptive) feeling, feel (cold, sick, hot, and so on), feel something doing
something to you'. This verb is clearly distinct from the verb anpeme 'touch, feel by
touch, feel (rough, smooth, etc.)'. Historically, the verb welheme 'feel (propriocep-
tively)' appears to have its origins in the reflexive form of the verb 'to hear' (aweme).
In Warlpiri 'feel (proprioceptively)' is synchronically an extension of 'hear', again
using the reflexive, whereas 'feel by touch' uses another verb (?3.4). We refrain from
adding this sixth modality merely because too few sources discuss it to make a compara-
tive study possible. We make a second and more radical simplification in the data by
collapsing the three event types. This is because Australian languages systematically
fail to make a lexical distinction between them, using constructional differences and
context to make the semantic distinction where necessary.
Consider the Kayardild verb marrija, which is used in the translation of English
5 As a referee pointed out, there are of course particular contexts in which the progressive is acceptable
with such verbs, in particular in registers of commentary, such as sports commentary ('she's hearing her
teammate call to her for the ball') or stage directions in theatre ('as she crosses the floor, she is seeing the
painting in a new light'). It is likely that these possibilities arise from the special aspectual structure that
these registers have, and we ignore them here.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 555
(2) KAY
malangarrba-ya ngada marri-ja dathin-ki dangka-y.
drunk-oBJ 1SG.NOM hear-ACT that-oBJ man-OBJ
'That man sounded drunk to me.'
Given the typically 'derived' nature of the source-based set, and the lack of consistent
differences between the sets denoting controlled vs. noncontrolled perception, we will
henceforth restrict ourselves to considering just the five basic perception verbs (for a
fuller discussion of how the different event types are expressed, see Evans & Wilkins
1998). In glossing these verbs we will treat the noncontrolled sense as basic, but in
our actual translations we will employ the English verb reflecting the event type impli-
cated by the context.
We now turn to the question of semantic extensions across modalities.
3.2. SEMANTIC EXTENSIONS ACROSS SENSORY MODALITIES. On the basis of his survey of
more than fifty languages, Viberg (1984:136) sets up the following simplified modality
hierarchy based on attested semantic extensions and polysemies across sensory modali-
ties in the domain of perception verbs (Figure 1). Essentially the hierarchy indicates
that a verb originally referring to 'sight' can extend its meaning to refer to 'hearing',
and a verb originally referring to 'hearing' can extend its meaning to refer to 'touch' and
so on. The pattern of extension is, however, unidirectional. A verb originally referring to
6 The abbreviations used to identify the languages in example sentences are listed at the end of the article.
7 The following abbreviations are used in the glosses: ABL(ative), ABS(olutive), ACC(usative), ACT(ual)
BEN(efactive), cAUs(ative), CHAR(acteristic), COMIT(ative), COMP(lementizer), CONTR(astive), c(hanged) s(tate),
DAT(ive), DU(al) EMPH(atic), EP(enthetic), ERG(ative), FUT(ure), GEN(itive), IMP(erative), IMPERS(onal),
INCH(oative), IRR(ealis), ITER(ative), Loc(ative), MA(sculine), NEG(ative), NEG.ACT (Negative actual),
NoM(inative), NOMZR (Nominalizer), N(on) P(ast), oBJ(ect), PAss(ive), PL(ural), P(ast) c(ompletive), P(ast)
P(erfective), PRES(ent), PRIv(ative), PROP(rietive), P(a)sT, PURP(OSiVP(lication), REive), RDu(lication), REL(exive), REL(ativ
REP(etition), R(eflexive)R(eciprocal), s(ame) s(ubject), SEQ(uential), SER(ial), SG (singular), suBJ(ect). Roman
numerals I to IV refer to noun classes in Mayali and Kuninjku. Arabic numerals refer to person values;
divalent prefixes of the form 1/3 mean 'first person acting upon third person', with the number to be understood
as singular unless otherwise marked. An equals sign indicates a clitic boundary.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
556 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
smell
'touch' never extends to cover 'hearing', and a verb originally referring to 'hearing'
never extends to cover 'sight'. The above hierarchy obscures the fact that patterns of
extension do not always operate contiguously. While shifts always preserve the pattern
of extension from HIGHER modality to LOWER modality in the domain of perception
verbs, the extensions may skip certain intermediate modalities. Viberg (1984:147) pre-
sents the complete network of attested shifts in a refined version of the hierarchy (Figure
2).
SIGHT
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 557
KURTJAR ak 'perceive, (esp.) see, find out', also 'meet, hear, smell'
rdengkarr.ingk ak 'hear' [ear-ergative/locative see/perceive/hear]
oongk ak 'smell' [odor see/perceive/hear]
DJABUGAY ngundal 'see, watch, look at'
bina ngundal 'hear, listen' [bina: ear]
MAYALI/KUNINJKU bekkan 'hear, listen; feel'
manjbekkan 'taste' (lit. 'taste-hear')
kukbekkan 'touch' (lit. 'body-hear')
WARLPIRI nyanyi 'see, look at'
purda-nyanyi 'hear, listen', etc.
parnti-nyanyi 'smell (trans.)'
TABLE 7. Extension across sensory modalities by adding a morpheme, typically denoting organ or
stimulus.
Yimithirr, Djabugay, and Kurtjar-are from the region around the southern half of
Cape York, which suggests that the extension of 'sight' to 'hearing' could be an areal
phenomenon there.
Other examples of the shift of 'sight' to 'hearing', outside of the Cape York region,
include Jaru, Ngaliwurru and, perhaps, Wardaman. Along with Warlpiri, these lan-
guages are part of a northwestern areal block, characterized by having a small, well-
defined set of monomorphemic verb roots. In this case, extension correlates with the
fact that there is a reduced set of lexicalized distinctions in the verb class.8 For Jaru,
Tsunoda (1981) notes how under most conditions a verb compound (VC) involving
the verb 'to see' is used to render the notion 'hear, listen', while in the imperative the
'see' verb on its own can be used in the sense of 'listen'. The relevant form, nyang-
'see; look' is clearly a descendent of the Australian proto-verb for 'to see' mentioned
earlier, and Tsunoda writes (1981:184):
Djaru has very few verbs-only about 40 ... [b]ut ... more than 290 preverbs and in many cases what
is expressed by a single verb in Djirbal is expressed by a VC of a preverb and verb in Djaru, even
basic notions such as 'hear/listen to'-bura nyang- ... (bura preverb 'listening', nyang- Vtr 'see/look
at') ... [A]t least in the imperative, i.e. nyang-ga, this verb alone (without the preverb bura 'listening')
can mean 'listen' . . . It appears that when nyang-ga 'see'-IMP is used in the sense of 'listen', the sentence
consists of just this word and no other words (e.g. subject, object) at all. This 'marked' use of the verb
'see' is syntactically extremely limited.
In Ngaliwurru (Schultze-Berndt p.c.), a language with only about thirty verb roots,
there is a simple verb for 'to hear', -malangawoo, but this is almost certainly based
historically on -ngawoo the verb 'to see'.9 Finally, with respect to Wardaman, Merlan
(1994:174) speculates that, 'The few verbs which end ... in -rna are: jomarna- "to
8 It is well known that there is a linguistic area in the northwest part of Australia in which languages
have small closed class sets of monomorphemic verb roots (see, for instance, Dixon 1980). This area cross-
cuts the distinction between Pama-Nyungan and Non-Pama-Nyungan. Among the Pama-Nyungan languages,
for example, Warlpiri has only 120 verb roots, Warumungu 53, Warlmanpa 43, and Walmajarri and Djaru
have about 40. Among the non-Pama-Nyungan languages, Wardaman has about 130 (with 8 used with a
very high frequency), Jaminjung about 30, and 'some languages of the Kimberleys and the Daly River area
have only about a dozen roots to which can be added verbal inflections' (Dixon 1980:280). In all the instances
we examined of languages with limited sets of verbs, if a language has a perception verb, it will be 'see'.
There is no language with a 'hear' verb that does not have a 'see' verb. As seen in Warlpiri and Djaru,
'hear, listen' is often derived by virtue of a preverb added to the verb 'to see'. However, the verb for 'hearing'
is also often derived on the basis of an addition to the verb for 'take' or 'do' (e.g. Walmajarri).
9 In Jaminjung, Ngaliwurru's closest relative, the verb for 'see' is -ngawoo, but 'hear; listen' is an extended
meaning of the verb -ooga, which is glossed as 'TAKE'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
558 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
(3) DAL
manjh kah-bob-mu ngah-bob-na-n
meat 3-smell-INCH.NP 1/3-smell-see-NP
'I can smell the meat.' (lit. 'the meat smells, I smell it')
'See' is not attested with extensions, whether direct or indirect, to the senses involving
direct contact: touching and tasting.
3.4. EXTENSIONS FROM 'HEAR' TO OTHER SENSE MODALITIES. 'Hearing' is attested with
extensions to all three lower senses. In the Kuninjku dialect of Mayali bekkan 'hear,
listen' can extend to 'feel by touch' without formal marking, as in 4, or it may incorpo-
rate the noun kuk 'body, physical presence' to give kukbekkan, which can only mean
'feel (by touch)'.
(4) MAY
La 0-wurlebmeng 0-yawam ku-rrulk-dulk-kah
and 3P-swam 3P-searched LOC-REDUP-tree-LOC
0-ngimeng kanjdji wurrno-kah, 0-yawam
3P-entered inside hollow.log-Loc 3P-searched
kure 0-wurlebmeng kun-kudji 0-bekkang 0-karrmeng,
LOC 3P-swam IV-one 3/3P-heard 3/3P-grabbed
0-bekkang 0-karrmeng.
3/3P-felt 3/3P-grabbed
'Again he went down and searched for it, this time feeling inside a hollow
log in the water, he searched around under the water and he felt it and
grabbed it.'
In Warlpiri purda-nyanyi 'hear, listen to' (itself extended from nyanyi 'see' by preverb)
will have a 'feel (proprioceptively)' reading when used reflexively with a complement
of evaluation (Laughren 1992:222). For 'feel by touch' another verb (e.g. marnpirni
'feel with hand') will be used.
(5) WLP
wati-ngki ka-nyanu purda-nya-nyi murrumurru
man-ERG PRES-REFL hear-perceive-NP sore.ABS
'The man is feeling sore.' (lit. 'the man hears himself (to be) sore').
10 The etymology of wo- is unknown. Unlike bob 'smell' and dolng 'smoke' it is not a productive incorporat-
ing noun, but comparison with roots in neighboring languages (e.g. Mayali (-wok 'language') suggests it
may have originally meant 'words, language'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 559
(8) KAY
banyji-ja diya-ja ngada barrngka-y
smell-ACT eat-ACT 1SG.NOM lily.root-oBJ
'I tasted the lily roots.'
Worms (1942:124) mentions extension from 'smell' to 'taste' in Bardi, attributing the
extension to the noun nyaar, but since his example involves a sentence it may also be
interpreted as polysemy of the preverb plus verb combination nyaar i-nen 'it smells/
tastes'. In Gugu Yalanji nyumal means 'smell or taste (transitive)'; comparative evi-
dence points to an original 'smell' meaning for this verb-see Evans & Wilkins 1998:
18. There are no examples of 'taste' extending to 'smell'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
560 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
The fact that a verb meaning 'try' in the context of food and eating will be interpreted
(via this particular bridging context) as meaning 'taste' is not unusual and is attested
in many languages of the world. Dixon (1991) presents Yidiny examples of banja-L,
in the sense of 'taste', which have that meaning only in combination with 'eat' and
which he explains as meaning literally 'try eat'. Other languages have 'taste' as an
extension of 'bite': Lardil betha 'to bite; to taste, have a taste of, eat a sample of.
Similarly, Warlpiri paja-rni 'to taste; savor' is almost certainly descended from an
original proto-Pama-Nyungan verb *paja- 'to bite; chew' (cf. O'Grady 1990:220).
In Ngiyampaa (Donaldson 1994; 1980), both 'taste' and 'feel' are complex forms
premised on the notion of 'testing' (or 'trying') with a certain body part: nga-thali
'taste', literally 'test-with mouth', and nga-mali 'to feel', literally 'test-with hand'.
Although there is often evidence that 'try' is the primary meaning of a verb, and 'taste'
a secondary meaning, in some cases, e.g. Ngalakan many-ngu 'taste, test', the etymology
shows the 'taste' meaning to be original (the form is identical to Mayali manj-ngu
discussed above).
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 561
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
562 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
The perception verb data, then, suggests that within the domain
'nature's basic plan' may be a stronger force than cultural condi
to lexicalization patterns and direction of semantic shifts. Wheth
transfield metaphorical shifts from the domain of perception to
be explored in the following section.
4. TRANSFIELD MAPPING OF PERCEPTION ONTO COGNITION. We turn now to transfield
semantic extensions from the sensory to the cognitive domain, and here we will find
a radical departure from the Indo-European pattern. We will demonstrate that in Austra-
lian languages it is 'hear' rather than 'see' that regularly maps into a large set of
cognitive verbs, including 'knowing', 'remembering' and 'thinking' as well as the more
familiar 'understanding' and 'heeding'. 'See' only rarely extends into the cognitive
domain (usually via 'recognizing visually', thence sometimes to 'know (esp. by sight)'),
and more commonly denotes interpersonal emotion and communication such as 'meet
with', 'look upon with desire', 'choose' and so on. 'Smell', 'taste' and 'feel' also have
limited sets of extensions into the cognitive domain.
We will first examine the way in which syntactic frames can be used to distinguish
cognitive and perceptual senses of such verbs, at least in some languages. This is
relevant to the question of whether we are dealing with a clear distinction between
perceptual and cognitive senses in the languages in question. Then we anticipate the
lines of development of 'hear' and 'see' by examining the semantic extensions of the
associated body parts, 'ear' and 'eye' in a typical language, Kayardild. From there we
pass through semantic extensions of the verbs themselves, starting with 'hear' and
moving on to 'see', 'smell', 'taste' and 'touch'. We conclude by summarizing the
overall pattern of mappings from sensory modalities to cognition and emotion, and
discussing the extent to which there is a recognizable geographical patterning.
4.1. DISTINGUISHING PERCEPTION AND COGNITION SENSES OF POLYSEMOUS VERBS. In a
language with a single verb for 'hear' and 'think' (or 'see' and 'think', for that matter),
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 563
(9) a. PTJ
Ngayulu alatji kulini, 'tjinguru-la...'
I like.this think.PRES maybe-we
'I think this about it, "maybe we ..."'
b. Ngayulu anangu-ngku wangkanytjala kulinu
I people-ERG talk.NOMZR.LOC hear.PsT
'I heard people talking.'
c. Wati katjangku mamangka kulintja wiya
man son.ERG father.Loc heed.NoMzR no
'The son won't heed his father.'
12 Wierzbicka (1994:455-56) writes that Pawley 'insists that the meaning of ny is unitary (in the name
of the general methodological principle that "semanticists and lexicographers should first seek a unitary
meaning for a word") ... but again, he doesn't say what this supposedly unitary meaning is'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
564 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
13 We thank Cliff Goddard for supplying the following examples, and suggesting the lines of the argument
here.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 565
(13) PTJ
Ka tjaka lirungku nguurpa kulira ngunti kulilpai
CONTR typical snake.ERG snoring hear.sER wrongly think.CHAR
tjulpu kulunypa wangka-nya-ngka palku
bird young talk-NOMZR-LOC mistake
'And as one might expect, the snake, hearing the snoring, wrongly thought
it was baby birds making sounds.'
(d) When kulini takes an auditory object, it can have the 'hear' meaning only if the
sound is a nonspeech noise, as in 14a. If the sound denoted by the object is represented
speech, it has to mean 'think', as in 14b; to get the 'hear' reading one has to say
something like 14c, with an explicit participle naming a speech act. If 'hear' and 'think'
were not conventionally different senses, but were merely coerced by context, the
unavailability of the 'hear' sense in 14b could not be explained.
(14) PTJ
a. Munu kulinu, "Rrrrrrr ..."
and hear.PsT rrrr
14 One referee wondered whether an example like 15 might represent a case of stylistic repetition ('Now
you each must hearingly hear'). We can safely reject this hypothesis. While Arrernte hortatory discourse,
of the type this sentence comes from, frequently employs a type of stylistic repetition, it is invariably repetition
via synonyms rather than exact repetition. Moreover, such repetition by synonymy works across, not within,
phrases (Wilkins 1989:31-32). Cases where the same content word appears repeated within the same clause
or utterance are extremely rare and really only happen in the clearest cases of polysemy, where each form
has its own sense.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
566 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
KAYARDILD
Similar bifurcations in the patterns of extension of 'eye' and 'ear' are widespread
in Australian languages, and have been discussed so many times (Schebeck 1978,
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 567
Such extensions are indeed also common in Australian languages. We have already
encountered uses of Pitjantjatjara/Yankunytjatjara kuli- with this sense (see ex. 11).
Other languages with this semantic range are Wik Mungkan ngeeyan 'listen, understand,
hear (and obey)' and also aak ngeeyan 'obey, listen, understand' (aak 'place, home,
camp, ground, country'), and Lardil merri 'hear, listen to; obey, pay heed to', for which
17 is an example.
(17) LRD
Kuba mangarda kiin, merral-kub-u. Warngelani merri danga-n.
good child that ear-good-PROP instantly hear person-OBJ
'That child is good, and obedient. He obeys people instantly.' [lit. 'That
good child has good ears; (he) instantly hears people.']
There are also, of course, languages with a distinct form, for example Arrernte akangkw-
irreme 'pay attention to someone; heed; obey'; Walmajarri, where mapunikanu 'obey,
take notice of, believe' is based on mapun 'true', and Burarra, where yagurrma has
the range 'agree to, obey, give assent to'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
568 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
15 In the traditional multilingual setting over much of Australia, it is quite normal for people to have
complete passive knowledge of a language they do not speak, and these people are said to 'hear' the language,
while those who have productive competence as well are said to 'speak' it.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 569
In many other languages 'think' is derived from 'hear, listen; understand' by reduplica-
tion (24-27), reflexivization (28-29) or incorporation (30).
(24) Wik-Ngathan: ngeethe- 'hear, listen'
ngeeth-eche 'think' (reduplication of ngeethe)
(25) Oykangand: aliya- 'listen, hear'
aliyiya- 'think, recall'
(26) Watjarri: ngangkunmanja 'listen, hear' (tr.)
ngangkungangkunmanja 'think' (intr.)
(27) Dalabon: wonan 'hear, listen; understand'
wonarrvn 'think about'
(28) Mayali: bekkan 'hear, listen'
bekkarren 'consider, think about before
making a decision'
(29) Dyirbal: ngamba-l 'hear, listen'
ngamba-yirri-y 'think'
(30) Ngandi: nga- 'hear' (tr.)
yic-nga- 'think' (intr.), yic- 'thinking, truth'
In Yukulta marrija means 'listen, hear' when used transitively, and 'think, feel' when
used intransitively (Keen 1983:276); the reduplicated form marrinymarrija has a middle
case frame and means 'to dream of/think of someone (i.e. to tune into their vibrations)'.
This gloss is interesting, suggesting that 'thinking of' is conceptualized less in terms
of generating an internal representation and more in terms of tuning in to an object
with an external existence.
16 Sources on some languages do not include 'understand' as a sense of this lexeme, but give no translation
equivalent for English 'understand'; Wik-Ngathan (Sutton 1995) is an example, as is Nunggubuyu wawangki-
'listen, pay attention, think'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
570 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
(32) GYA
mari doctorangka jiba-bu nyajil yina jalbu wulay
man doctor.ERG liver-with perceive that woman die
'The doctor man knows by instinct that woman will die.' (Oates 1992:
103)
17 In traditional Aboriginal societies there is a widespread belief that certain types of information and
knowledge can be gained by extrasensory perception. Certain powerful individuals may be specially clairvoy-
ant. In addition, many Australian languages have a large set of expressions for different types of TELAESTHESIS,
which Douglas (1977) defines as 'the supposed ability to acquire information about distant happenings or
forthcoming events through the interpretation of certain physical disturbances in the body'. The little data
we have suggests that such extrasensory perceptions tend to be treated as hyponyms of one of the basic
perception verbs (usually 'hear' or 'feel (proprioceptively)'), but a proper treatment is beyond the scope of
this article.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 571
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
572 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
look after with the ear'. It is also worth reiterating at this point
'ear' is glossed as 'remember' when used as a particle. Many oth
to do with memory are also typically based on 'ear'-Kayardi
'forget', literally 'ear become useless', marraldurldiija 'forget',
the many Nyulnyulan languages in which one says, for exampl
Bardi alamar i-nen djen) for 'I remember him' and 'my ear it i
Nimanburru nalebab inan djen williwilli-en) for 'I still remember
(Bill McGregor p.c.).
D; K; A; Alyawarre + +
Wik Mungkan + + +
Ngaliwurru, Banjalang, Warluwarra + + +
Nunggubuyu + + + +
Kuk; Luritja + + + +
Pitjantjatjara + + + + + +
Warlpiri + + + +
Yawurru; Wakaya + + +
Ngarluma + + +
Yidiny + + +
Wemba-Wemba; Gugu Yalanji + +
Lardil + +
TABLE 12
This pat
as well a
such as g
tion, an
recollect
mental c
as well as names and sounds.
We will now show how this pattern of extensions contrasts with the extensions
'see', and, less importantly, 'smell' and 'taste'.
4.9. EXTENSIONS OF 'SEE' TO THE COGNITIVE AND SOCIAL DOMAINS. Most extensions o
'see' in Australian languages lead into the domain of human interaction: desire an
sexual attraction, supervision, and aggression. Such extensions are of course not unc
mon in European languages, but make up a greater proportion of the extensions
'see' verbs in Australian languages.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 573
Extensions to aggression are not common with the verb 'see' itself, but in l
that combine a 'see' auxiliary (or light verb) with an uninflected lexical
collocations can denote a range of aggressive social acts. In Tyemeri, for exam
auxiliary nginnyinggin, which on its own means 'see', participates in the fo
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
574 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
There are also idioms based on 'eye' indicating negative and aggressive social interac-
tion. Thus in Arrernte we find alknge-uthneme (eye-bite) 'be jealous of someone'.
Similarly, in Yidiny we find jili-guba-N (eye-bum) 'feel jealous towards someone',
and also jili-gunda-L (eye-cut) 'make someone look away (by staring at them and
making them ashamed)'. Finally, in Pintupi, two idioms of aggression are kuru watjanu
(eye said) 'accused to face; blamed to face' and kuru panypurangu (eye spoke.against)
'belittled to his face, rubbished to his face'.
Parallel derivations based on 'eye' include Yidiny jili-budi-L (eyes put down) 'look
after', Kuku-Yalanji miyil-da kujil (eye-with keep) 'to guard something (keep one's
eyes on it)' and Pintupi kuru yutura kanyinu (eyes hiding kept) 'carefully looked after,
cared for'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 575
A few languages extend 'see' to mean 'recognize (visually)', often with an incorp
word for 'body'; sometimes this extends on from 'recognize' to 'know'. T
Mayali derivative of 'see', incorporating the root burrk- 'body', is burrknan 'rec
A related language,
understand', although theNgalakan,
one exampleextends
sentence the sense
in the of(Merlan
source the cognate
1983:192)verb
could bur 9.a- to 'know,
equally well be translated with 'recognize (visually)' (39). Warray na- 'to see' gives
rise to the compounds let-na 'to look after' and mitj-na 'to know, to recognize'.
(39) NGK
rju-bur;nani-koro nugunbiri bigur
I/him-knOW-PRES.NEG that man
'I don't know that man.'
The 'see' auxiliary in Tyemeri occurs in two collocations concerned with recognition:
yilil nginyinggin 'to be able to recognize something', but the only available example
involves visual recognition (more specifically, looking but not recognizing), and miyilil
nginyinggin 'recognize someone or something'.
In Warlpiri 'see' can take on a judgment or evaluation sense, with state-of-affairs
complements only (see Laughren 1992:223); this use has not been reported for other
Australian languages.
In a number of languages, we find that the verb 'see' can take clausal complements,
(direct quotes), which represent a deduction based on visual evidence. For Gooniyandi,
McGregor (1990) discusses what he terms 'projection of thoughts', and notes that the
verb mila- 'see' can enter into the same construction as verbs referring to mental
processes (like 'think'). He writes (1990:421-22) that '[i]n this case, the projected
clause represents a thought that was perceived, or which was based on perceptual
evidence'. Such constructions typically translate into English as 'X saw that "Y"
[clausal deduction]', but always entail actual visual perception at the source (i.e. visual
evidence is the source for the deduced/projected thought). A Gooniyandi example with
mila- 'see' projecting a direct quote is:
(40) GNY
yoowooloo-ngga-nyali mila.winbidda boolgawoolga-ngga
man-ERG-REP they.saw.them old.men-ERG
aa ngamoo girli boolgawarri garingi ngangbada
ah before same he.is.getting.old wife we.will.give.him
'The old men would see "he's getting old, we'll give him a wife".'
Mangarrayi and Ungarinjin have similar constructions with the 'see' verb. Given that,
in European languages, such deductions on the basis of visual evidence or visual recog-
nition are the typical precursor to extensions of 'see' into cognition uses without any
entailment of visual perception, it is significant that this relatively common construction
in Australian languages does not appear to give up its perception interpretation very
easily.
We know of only two Australian languages that have some evidence of 'see' develop-
ing to 'know' or 'think' without first passing through 'recognize', as in the Ngalakan
case (or through 'hear', as in Guugu Yimithirr). First, the Kaurna language, spoken
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
576 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
4.12. 'SMELL', 'TASTE' AND 'TOUCH'. In a few languages 'smell' has limited cognitive
extensions: Nunggubuyu yarra- 'to smell (something)' can also mean 'to detect, to
sense (something)'. Two languages that appear to have shifted the meaning of the
'smell' etyma *bany-rdi and *numa- are Paakantyi: parnta- 'to search, to look for, to
come out', presumably via 'sniff out', and Wemba-Wemba nyuma- 'to recognize, know'
and nyumila- 'to think', presumably via 'recognize by smell' with later generalization
to 'recognize' and 'know'.
The remaining two senses, 'taste' and 'touch' have no significant extensions into
the cognitive domain in Australian languages.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 577
In the next section we show that these same patterns are reflected in evidence from
other semiotic systems, and in ?6 we will provide ethnographic data to help explain
why it is 'hearing', rather than 'sight', which is linked to intellection at large. The
anthropologists of the senses are clearly right about cultural relativity when it comes
to transfield metaphorical mappings from 'perception' to 'cognition', even if they were
wrong about relativity in the intrafield ordering of perceptual modalities.
5. EVIDENCE FROM OTHER SEMIOTIC SYSTEMS. Until now we have concentrated on data
from the everyday registers of Australian languages, but one of the reasons Australian
languages are particularly interesting and important for the general study of polysemy
and semantic change is that they provide a further window on semantic relations in
the form of special auxiliary registers. Typically, the indigenous auxiliary registers used
by Australian communities have a smaller vocabulary and concomitantly more abstract
or hyperpolysemous word meanings, making them extremely useful for the study
of semantic structure (Dixon 1971, Hale 1971, 1982, Haviland 1979a, Evans 1992a,
Wilkins 1997). Evans (1992a:488) has noted that in many cases semantic associations
evidenced by other semiotic systems will parallel those of everyday language. Similarly,
Wilkins (1997:414) argues that 'everyday language is just one of a number of semiotic
systems which a speech community has at its disposal, and so one should not only
look to other everyday languages to provide independent documentation of a semantic
association, but one should also cross-compare semiotic systems'.
We will examine the extent to which data from other auxiliary registers parallels or
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
578 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
18 Initiation registers are also worth considering in this regard. We show elsewher
1998) how evidence from these also parallels the findings concerning the extension of p
Australian languages.
19 The root beng is found in a number of cognitive adjectives and verbs, such as beng
bad), bengngukme 'forget' (beng-shit), bengyirri 'be attentive' (beng-coMIT-stand), be
(beng-stand-cAUS), bengbun 'make distracting noise, annoy, disturb' (beng-hit) and so on.
languages, these would be derivatives of 'ear'; however, the only verb in this set based on
rran 'forget', a synonym of bengngukme that literally means 'ear-block-itself'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 579
Kendon (1988:171-72) discusses Warlpiri signs that involve pointing to the ear
ears, and notes that the manner of pointing varies in a motivated fashion and is reve
of semantic contrasts in the domain of cognition. He observes that many signs
point to the ear 'relate to the referent indirectly, for the ear now stands for "chann
understanding" '. Close observation reveals that in signs expressing effective, pos
cognitive functioning- 'that is, such meanings as "wise", "knowing", "understan
ing" '-the pointing shape that approaches the ear is a form of homed hand
index finger and little finger extended, and ring and middle finger drawn in. This
handshape is also used to indicate the notion of 'going' or moving freely through s
and might here be taken to indicate that information is moving freely, or that
channels of intellection are open. By contrast, 'if the meaning is negative-such m
ings as "senseless, crazy", "forget", and the like-the hand is a flat (B) which her
perhaps, suggests that the ear is blocked or covered'.
The signing of notions relating to the domain of cognition in the region of the
is very common in Central Australian communities. 'In sign language, a person
points to his ear usually with his right hand, palm forward and outstretched fin
together, is expressing that he knows what a person is speaking about or that he u
stands the matter under discussion' (Peile 1997:50, on Kukatja). Wilkins has reco
a complex Arrernte handsign in which the Arrernte verb alkngwirreme 'to forge
rendered using a sequence of three signs. The first sign is a loose hand, index f
trace around the ear, which variously signifies 'understanding, hearing, informat
The second sign is the sign for 'to leave' and the third sign is the sign for 'to disap
In other words, 'forgetting' is rendered in sign as 'understanding/information leav
disappear'. This is of special interest, since the everyday language form for 'forge
likely to have originated as a compound involving alknge 'eye' and uyirreme 'to d
pear' (i.e. alknge-uyirreme). That is to say, while both the everyday Arrernte form
the auxiliary sign form seem to be premised on the notion of 'disappearing', the fo
incorporates the 'eye' while the latter incorporates the 'ear'.
Adam Kendon kindly provided his database of Central Australian signs for us
search. This database contains approximately sixteen hundred entries and is Ken
entire collection of verified signs collected during fieldwork in 1978, 1981,
1984-86 at Yuendumu (Warlpiri), Ti Tree (Anmatyerre), Neutral Junction (Kayte
Tennant Creek (Warumungu and Warlmanpa), and Elliott (Djingili and Mudbura).
searched first for signs enacted in the ear region and the eye region. Our purpos
to gather any body-part, perception, cognition, social interaction and emotion read
associated with these signs (other meanings, such as animal names, were ignored
Signs enacted in the region of the ear had the following meanings:
(41) ear hear understand
wise, knowing ponder, solve, thin
deaf without understanding crazy, senseless, temporarily
insane
unaware, ignorant of be unknowing heedless
lose forget
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
580 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
20 One of the referees noted an additional use, again paralleling an extension in the o
the use of the V-fingers shape in neutral space to refer to flirting or sexual advance.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 581
But ethnographies of communication for Australian languages frequently stress the role
of individual choice in selectively directing attention in hearing:
The strong tendency in Aboriginal conversations is to turn the communication channel (talk) on and
leave it on; it is continuous ... In the Aboriginal setting, WHERE I AM SAYING THE LISTENER HAS MORE
CONTROL, MEMBERS OF THE GROUP CAN TUNE IN AND TUNE OUT OF THE ONGOING (CONTINUOUS) COMMUNICA-
TION AT WILL ... The Aboriginal pattern of interaction can be viewed as a coping strategy: it enables
an individual to opt for privacy but preserve the option to re-engage at any time. Since there are no
suitable means of using the built environment to ensure personal privacy, the members of the remote
Aboriginal community manipulate the pragmatic environment, keeping the communication channel
continually open but only directly engaging when it is appropriate or when they choose to. (Walsh
1991:3-4, emphasis ours)
typical Aboriginal social conditions of rather exposed camp life and highly developed etiquette of
selective orientation and attention to others at any given time ... (Merlan 1989:230-31).
Compared to seeing, the act of directing attention with hearing is internal: directed
visual attention can be noted from outside, through movements of the eyes or head,
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
582 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
For a Yolngu to hold a person with his gaze can be a sign of power or can signify a bid for power.
Yolngu children are discouraged by their parents from doing this. Some ceremonial rituals demonstrate
one figure claiming power over another through open and direct staring. Such direct staring is sometimes
thought of as a sign of madakarritj ('anger, belligerence'), and sometimes balanda [i.e. Europeans -
NE & DW] who want to be 'open' and friendly can be misunderstood, through the directness of their
eye contact, to be claiming authority or power.
Clearly gaze, and even facing to 'look' or 'see', are highly loaded in this communicative
setup. Such a context would strongly favor extensions of 'see, look' into certain types
of social interaction, and concomitantly limit their extension into cognition and intellec-
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 583
6.3. DIFFERENT PROTOTYPES FOR PERCEIVING OBJECTS ABSENT FROM THE IMMEDIATE
SCENE. Another way of viewing the difference between Australian and Indo-European
patterns is to see the two cultural groups as placing different bounds on when 'see'
and 'hear' can be used in a nonliteral sense. English and other Indo-European languages
readily relax the reality requirement,21 allowing the use of 'see' for 'mental vision' in
sentences like 'I can still see my grandmother's wrinkled old face looking at me the
day before she died'. Australian languages are not reported as being able to relax this
requirement for 'see', but do it for 'hear' as with many of the 'remember' and 'know'
examples discussed in ?4. In other words, if vision implicates physical reality (as
evidence) whereas audition does not, 'I heard X', vis-a-vis 'I saw X' will therefore
fail to implicate the presence or real status of X. This would then favor the use of
'heard' as a metaphor for those types of perceptionlike behavior where X is apprehended
to consciousness despite its physical absence. A revealing semantic parallel is the not
uncommon occurrence of demonstratives in Australian languages with semantics like
that of Dyirbal ngala- 'not visible; either audible or remembered'.
6.4. DIFFERENT COMMON SCRIPTS: KNOWING THE WAY, KNOWING THE COUNTRY. Another
possible explanation is that particular patterns of lexicalized polysemy reflect the fre-
quency of textual exemplars allowing the corresponding contextual extensions. In the
Australian context we might appeal to the frequency both of the practice of learning
about country, tracks and routes, and mythological knowledge by hearing them re-
counted in stories and 'songlines'. Ex. 45 contains a representative quote.
(45) 'Tywerrenge and songs come out of the body of the country ... We're not
like whitefella who can take a photograph and say what pretty country it is;
we've got the song to sing for that country.
The country has got sacred sites, that stone, that mountain has got dreaming.
We sing that one, we've got the song.
... Country is nothing else but culture.' (Wenten Rubuntja in Green 1988)
The frequency of this cultural practice then engenders a second-order frequency of
texts in which knowledge and memory are reported in terms of 'hearing (+ > names
of)22 places', 'hearing (+ > names of) ways' and so on, making utterances furnishing
bridging contexts, along the lines of exx. 33 and 34 above, common enough to serve
as templates for lexicalizing this extension.
Further, it is especially in the context of relations to country that Australian Aborigi-
nal belief systems do not emphasize seeing as giving understanding or knowledge. In
discussing Aboriginal art, Sutton (1988:19) argues that for Aboriginal Australians 'there
is no geography without meaning or without history ... The land is already a narra-
tive-an artifact of intellect-before people represent it'. Knowledge of country is
21 Crosslinguistic studies of evidentials (e.g. Willett 1988) reveal an exceptionless ranking of visual evi-
dence as the most reliable indicator of an event's real status. In the light of this evidence the Indo-European
pattern can be seen as anomalous rather than normal, since extending 'see' to intellection means it no longer
has a guaranteed association with the depicted event being real.
22 Following standard practice we use the symbol + > to mean 'implicates'.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
584 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
23 It is well known that in many parts of Australia 'new' songs, 'new' dances and their musical accompani-
ment, and 'new' painting designs are spoken of as being 'dreamt'. That is, individuals in certain circumstances
have dreams which access dreamtime knowledge, and this knowledge can be integrated into current tradition.
The more general point here is that dreams are not regarded as primarily visual experiences, but have a
strongly auditory component, and sometimes are only spoken about in aural terms.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 585
6.6. LITERACY VS. ORACY. It is significant that the founding text for the anthropolo-
gists of the senses to whom we referred at the beginning of this paper was Ong's
seminal piece on the role of literacy in privileging sight as opposed to hearing, which
assumes greater dominance in a purely oral culture. Ong (1969:634) argues that 'Oral
or nonwriting cultures tend much more to cast up actuality in comprehensive auditory
terms, such as voice and harmony. Their "world" is not so markedly something spread
out before the eyes as a "view" but rather something dynamic and relatively unpredicta-
ble, an event-world rather than an object world'. One might argue that extensions from
'see' to 'think' and 'know' are therefore more likely to develop in literate cultures,
and, conversely, that developments from 'hear' would mark cultures with a basically
oral tradition, reflecting the unchallenged role of spoken transmission in acquiring
knowledge.
If this were so, Australian languages should not be the only ones displaying the sorts
of extensions discussed here: they should be common in languages spoken in other
preliterate cultures. Although some of the examples reported in Howes 1991 indicate
that 'hear' can extend to 'think' in other parts of the world as well-Hausa and the
Ommura examples that have already been discussed, and Seeger (1981) reports similar
patterns in the Brazilian language Suya.24 A widely cast crosslinguistic study is needed
to test this hypothesis carefully.
6.7. CONCLUSION. Our survey of Australian languages has shown that in one large
language family there is a consistent pattern of deriving cognitive verbs from 'hear',
encompassing both 'expected' cognitive processes like 'understand' and 'heed/obey'
and less expected ones like 'think', 'know' and 'remember' (?4). This is in spite of
the general patterning of perception verbs in a way that confirms the well-known domi-
nance of 'see' as the source of semantic extensions to other sensory modalities (?3).
The transfield mapping of perception to cognition, it seems, is much more plastic and
amenable to different cultural interpretations than the intrafield extensions of perception
verbs. We have demonstrated that the same domain can have its universal and relativistic
sides, a foot in nature and a foot in culture.
Using evidence from direct extensions (polysemy) and indirect extensions (derivation
and heterosemy) we were able to establish clear patterns of intrafield and transfield
change for the Australian region. As far as 'hear' and 'see' are concerned, these patterns
of change are replicated by extensions involving 'ear' and 'eye' respectively. For in-
stance, while 'hear' and 'ear' most commonly have transfield extensions to intellection
at large, 'see' and 'eye' tend to remain removed from the domain of cognition and
instead typically have transfield extensions into the domain of social interaction. The
24 In Suya the same verb, ku-mba, is used for hearing, understanding and knowing. 'When the Suya have
learned something-even something visual such as a weaving pattern-they say, "It is in my ear"' (Seeger
1975:214).
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
586 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
MAY BUR 12 AY
L DJP
PN WRR A
JAWNGK
WAG NN WNG
JAM WRD MAN WNR YYO GYI
NLW LRD D OYK GAJAB
BEA PMG ICY GYA M
UUNG
NYG G
KRJ W KUK
WLP
JlW NNr
PTJ YI
KKAU
MIY MUR N
PAA GAM
NGI
KAU
G= ~ WEM
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 587
Arrernte (Eastern and Mparntwe/ ARR Wilkins 1988, 1989, fieldnotes; Van Valin &
Central dialects) Wilkins 1993; Henderson & Dobson 1994
Alyawarr ALY Green 1992, Yallop 1977, Wilkins fieldnotes
Bardi BRD Worms 1942, McGregor (pc)
Burarra BUR Glasgow 1994
Dalabon DAL Evans field notes
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
588 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
REFERENCES
ALM-ARVIUS, CHRISTINA. 1993. The English verb see: A study in multiple mean
borg: Acta Universitas Gothoburgensis.
ALPHER, BARRY. 1991. Yir-Yoront lexicon: Sketch and dictionary of an Australian
Berlin: Mouton de Gruyter.
. 1993. Out-of-the-ordinary ways of using a language. Language and culture in
nal Australia, ed. by Michael Walsh and Colin Yallop, 97-106. Canberra: Ab
Studies Press.
AMERY, ROB, and JANE SIMPSON. 1994. Kaurna. In Thieberger & McGregor, 144-72.
AUSTIN, PETER. 1992. A dictionary of Jiwarli Western Australia. Bundoora, Vic.: La Trobe
University.
; R. ELLIS; and LUISE HERCUS. 1976. Fruit of the eyes: Semantic diffusion in the lakes
languages of South Australia. Papers in Australian Linguistics 10.57-77. Canberra:
Pacific Linguistics.
BAIN, MARGARET S. 1979. At the interface: The implications of opposing views of reality.
Monash University master's thesis.
BECHTEL, F. 1879. Uber die Bezeichnungen der sinnlichen Wahrnehmungen in den indoger-
manischen Sprachen. Ein Beitrag zur Bedeutungsgeschichte. Weimar: Hermann Bohler.
BENVENISTE, EMILE. 1966. Problemes semantiques de la reconstruction. Problemes de lin-
guistique generale. Paris: Gallimard.
BERLIN, BRENT. 1992. Ethnobiological classification. Principles of categorization of plants
and animals in traditional societies. Princeton: Princeton University Press.
, and PAUL KAY. 1969. Basic color terms: Their universality and evolution. Berkeley:
University of California Press.
BLACK, PAUL, and ROLLY GILBERT. 1986. Kurtjar dictionary. Batchelor, N.T.: School of
Australian Linguistics.
BUCK, CARL DARLING. 1949. A dictionary of selected synonyms in the principal Indo-
European languages: A contribution to the history of ideas. Chicago: University of
Chicago Press.
CLASSEN, CONSTANCE. 1993. Worlds of sense: Exploring the senses in history and across
cultures. London: Routledge.
COATE, HOWARD H. J., and ADOLPHUS P. ELKIN. 1974. Ngarinjin-English dictionary. 2 vols.
(Oceanic Linguistic Monographs no. 16.) Sydney: University of Sydney.
DENCH, ALAN CHARLES. 1995. Martuthunira: A language of the Pilbara region of Western
Australia. Canberra: Pacific Linguistics.
DIXON, R. M. W. 1971. A method of semantic description. In Steinberg & Jakobovits,
436-71.
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 589
--, -
mins. (eds.) 1983. Handbook of Aus
DONALDSON, TAMSIN. 1980. Ngiyam
Cambridge University Press.
. 1994. Ngiyampaa. In Thieberger
DOUGLAS, WILFRED. 1977. Illustrat
Canberra: Australian Institute of
- 1981. Watjarri. In Dixon & Blak
. 1988. An introductory dictionary
Language Studies, Western Austr
ECKERT, PAUL, and JOYCE HUDSON
language learner. Underdale, Sout
Education.
EVANS, NICHOLAS. 1988. Arguments for Pama-Nyungan as a genetic subgroup, with particu-
lar reference to initial laminalization. Aboriginal Linguistics 1.91-110.
. 1991. Draft Mayali dictionary. Melbourne: University of Melbourne. [http://
coombs.anu.edu.au/Special Proj/ASEDA/ASEDA.html].
. 1992a. Multiple semiotic systems, hyperpolysemy and the reconstruction of semantic
change in Australian languages. Diachrony within synchrony: Language, history and
cognition, ed. by M. Morrissey and G. Kellermann, 475-508. Bern: Peter Lang.
. 1992b. Kayardild dictionary and thesaurus. Melbourne: Department of Linguistics,
University of Melbourne.
--. 1995. A grammar of Kayardild. Berlin: Mouton de Gruyter.
-, and DAVID WILKINS. 1998. The knowing ear: An Australian test of universal claims
about the semantic structure of sensory verbs and their extension into the domain of
cognition. Cologne, Institut fiir Sprachwissenschaft: Arbeitspapier 32 (Neue Folge).
FELD, STEVEN. 1990. Sound and sentiment: Birds, weeping, poetics, and song in Kaluli
expression. 2nd edn. Philadelphia: University of Pennsylvania Press.
. 1996. Waterfalls of song: An acoustemology of place resounding in Bosavi, Papua
New Guinea. Senses of place, ed. by Steven Feld and Keith H. Basso, 91-135. Santa
Fe, NM.: PP School of American Research Press.
GARDE, MURRAY. 1997. Kuninjku dictionary. [http://coombs.anu.edu.au/Special Proj/
ASEDA/ASEDA.html]
GELL, ALFRED. 1995. The language of the forest: Landscape and phonological iconism in
Umeda. The anthropology of landscape: Perspectives on place and space, ed. by Eric
Hirsch and Michael O'Hanlon, 232-54. Oxford: Clarendon Press.
GLASGOW, KATHLEEN. 1994. Burarra-Gun-nartpa dictionary with English finder list. Dar-
win: SIL.
GODDARD, CLIFF. 1983. A semantically-oriented grammar of Yankunytjatjara. Canber
Australian National University dissertation.
. 1992. Pitjantjatjara/Yankunytjatjara to English dictionary. 2nd edn. Alice Spring
NT: Institute for Aboriginal Development.
--. 1994. Lexical primitives in Yankunytjatjara. In Goddard & Wierzbicka, 229-62.
. 1996. Pitjantjatjara/Yankunytjatjara to English dictionary. 2nd edn. revised. Al
Springs, NT: Institute for Aboriginal Development.
- , and ANNA WIERZBICKA (eds.) 1994. Semantic primitives and lexical universals. A
sterdam: John Benjamins.
GREEN, JENNY (ed.) 1988. Pmere: Country in mind. Alice Springs: Institute for Aborigi
Development.
- . 1992. Alyawarr to English dictionary. Alice Springs: Institute for Aboriginal Devel
ment.
HALE, KENNETH. 1971. Notes on a Walpiri tradition of antonymy. In Steinberg & Jakobovits,
472-82.
. 1982. The logic of Damin kinship terminology. The languages of kinship in Abo
Australia, ed. by J. Heath, F. Merlan, and A. Rumsey, 31-39. University of Sy
Oceania Linguistics Monographs.
-. 1990. Ngarluma vocabulary, English-Ngarluma vocabulary. Wangka Maya,
Aboriginal Language Centre.
, and the Institute for Aboriginal Development. 1990. Warlpiri to English vocab
with grammatical sketch and English to Warlpiri wordlist. Alice Springs: IAD
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
590 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
IN THE MIND'S EAR 591
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms
592 LANGUAGE, VOLUME 76, NUMBER 3 (2000)
Psycholinguistics
Nijmegen, Netherlands
This content downloaded from 150.214.130.24 on Tue, 07 Mar 2017 12:15:11 UTC
All use subject to http://about.jstor.org/terms