BabyTalkin Six Languages

CHARLESA. FERGUSON
Centerfor AppliedLinguistics

BACKGROUND

O CCASIONALLYlinguists have turned their attention to the description of

marginal systems within languages, such as animal calls, hesitation forms,
or baby talk. Such phenomena have sometimes been studied because of purely
linguistic interest in synchronic description: they often have elements of
sound or form which do not occur in the "normal" central system of the lan-
guage or have unusual arrangements or frequencies of occurrence of elements
which do occur in the central system. This kind of study is of particular rele-
vance to the question of the monosystemic nature of languages versus poly-
typical analyses of "coexistent" systems. These marginal phenomena have also
sometimes been studied from a psychological point of view, in relation to ques-
tions of language acquisition or language function.
The present paper approaches the analysis of baby talk from a rather gen-
eral taxonomic, linguistic interest. The intention is to initiate cross-language
studies of marginal phenomena of this kind which will lead to a general charac-
terization of them and to a frameworkfor the characterizationof single-language
marginal phenomena in such a way that synchronic classification and historical
explanation become possible.
By the term baby talk is meant here any special form of a language which
is regardedby a speech community as being primarily appropriatefor talking to
young children and which is generally regarded as not the normal adult use of
language. English examples would include choo-choo for adult train, or itty-
bitty for little. In most cases the baby-talk item can also be used in some other
situation with special value; in some cases (e.g., peek-a-boo) the item has no
counterpart in normal language since it refers to an activity or object appropri-
ate chiefly for children.
The method used here will be the comparison of baby-talk phenomena in
six languages, selected for variety of linguistic structure and sociolinguistic set-
ting within the limits of available material: (Syrian) Arabic, Marathi, Coman-
che, Gilyak, (American) English, Spanish. The first two are major languages of
Asia with millions of speakers and strong literary traditions; the second two are
of small nonliterate communities, one New World, one Old World; the last two
are major Europeanlanguages.The primarysource materials for the first four lan-
guages are the articles of Ferguson (1956), Kelkar (1964), Casagrande (1948),
and Austerlitz (1956); the material on English and Spanish was compiled from
informantsfor this study."
103
104 The Ethnography of Communication

ASSUMPTIONS
Before proceeding to examination of the material, certain assumptions of
this study should be made explicit since they are not in agreement with general
views of baby talk. Here it is assumed that baby talk is a relatively stable, con-
ventionalized part of a language, transmitted by "natural" means of language
transmission much like the rest of the language; it is, in general, not a universal,
instinctive creation of children everywhere, nor an ephemeral form of speech
arising out of adults' imitation of child speech. Like other marginal systems
such as animal calls, however, baby talk tends to show somewhat different pat-
terns of diffusion from the normal language: for example, particular baby-talk
items are often present in contiguousbut genetically unrelatedlanguages.
The assumption of relative stability as opposed to ad hoc creation is sug-
gested by such cases of historical documentation as in Arabic where there is a
record of Arabic baby talk used at the beginning of the nineteenth century
which is very much like Arabic baby talk today.2 An even more impressive case
is the persistence of baby talk words for food, drink, and sleep for some two
thousand years in the Mediterranean area. The Roman grammarian Varro
(116-27 B.C.)3 cites Latin bua and pdpa or pappa as baby talk for 'drink' and
'food' respectively, and the use of Latin naenia 'dirge, lament' in the baby-talk
meaning of 'lullaby' is attested.
At the present time the general Arabic baby talk for 'drink'is mbu or mbuwa.
The baby-talk word for 'food' is papa throughout the Spanish-speakingworld;
this is regarded by some speakers of Spanish as a special use of the adult word
for potatoes, but it is attested in Spanish before the introduction of potatoes.
The modern Moroccan Arabic baby-talk word for 'bread' is bappa (or babba, or
pappa). A common Arabic baby-talk word for 'sleep' or 'lullaby' is ninni or
ninni, which occurs also in Italian. The details of diffusion are quite unclear,
but there can be little doubt of a historical connection between the Latin words
and the contemporary Arabic, Spanish, and Italian ones.
The assumption that baby-talk items are conventionalized and culturally
transmitted, not universal, can be appreciated from a glance at Table I, below.
There are similarities in the structure of these items, which will be commented
on below, but any simple notion of universality is refuted by such contrasts as
the Syrian Arabic and Spanish baby-talk items for 'father' (bdba : tata), 'baby'
(bubbu : nene), 'food' : papa), 'little' (nfinu : tiquitito).4
The assumption that(.mamm
most baby talk is taught as such by adults to children
can be validated in an impressionistic way by simple observation. Adults in-
form the baby that a train is a choo-choo and a dog a bow-wow and in effect
drill the child in such items until he produces his version of them. The alterna-
tive explanation, that millions of children independently create items like choo-
choo and bow-wow instead of the hundreds of equally satisfactory onomatopoe-
ias that could be imagined, is clearly unsatisfactory. It is, of course, true that
adults sometimes do imitate an item of child speech and it gets accepted in a
family; it is also true that there are resemblances between features of child
 Baby Talk in Six Languages 105
speech and features of baby talk and that adults often feel that baby-talk items
are imitations of child speech, but the general assumption seems safe that
adults usually initiate baby talk, using the material familiar to them as appro-
priate for this. There are instances of baby-talk words becoming incorporatedin
normal language, e.g. English tummy, several Gilyak items (Austerlitz 1956:
271-2), Spanish pininos.
MATERIAL
Baby talk includes at least three kinds of material: (1) intonational and
paralinguistic phenomena which occur with normal language as well as with
other baby-talk material; (2) morphemes, words, and constructions modified
from the normal language; and (3) a set of lexical items peculiar to baby talk.
Intonational features have been noticed by many authors, and even casual
observersmay notice the higher overall pitch, preferencefor certain contours, and
special features such as labialization which occur in baby talk in a number of
languages. Much of this is subsumed under the term Ammenton. Very little sys-
tematic description of this kind of baby-talk material has as yet been attempted5
and it will not be discussed further here.
The baby-talk material derived from normal language shows considerable
variability in the six languages, but a number of patterns of modification,phono-
logical or grammatical,are sufficiently common to be of interest.
MODIFICATIONS OF NORMALLANGUAGES

Phonology6
Simplification of consonant clusters (e.g., English tummy for stomach) is
attested for all except Arabic and may well occur there too. There is an inter-
esting variation in this: Gilyak has many final clusters and, even though it sim-
plifies them, its final clusters in baby talk are more complex than those of baby
talk in the other languages.
Replacement of r by another consonant (e.g. English wabbit for rabbit),
either by a liquid 1, y or w or by an apical stop t or d, occurs in all six langauges.
The replacement by 1 in several languages is surprisingsince some linguists feel
that trills are more "basic" than laterals in that there are many languages with
trills and no laterals but few the reverse.
Replacement of velars by apicals (e.g., English tum on for come on) is at-
tested for all except Arabic and Gilyak, and considering the frequency of velars
in the Arabic and Gilyak baby talk it seems likely that this replacementdoes not
occur in these.
Some kind of interchange among sibilants, affricates, and stops (e.g., Eng-
lish soos for shoes) occurs in all but Comancheand Gilyak, but is of three differ-
ent types: (a) hushing sibilants replaced by hissing sibilants (Arabic, Marathi,
English); (b) sibilants replaced by [E]J(Marathi, Spanish); (c) affricates re-
placed by stops (Marathi). The most interesting of these is probably the re-
placement of [s] by [~] (e.g., Spanish becho for beso) since the latter is felt
by some linguists to be a less "basic" sound and this replacement seems very
106 The Ethnographyof Communication
unnatural for English speakers. In Spanish baby talk the use of [c] for [s] is
widespread and in fact is an identifying feature of baby talk; of the languages
discussed here it occurs also in Marathi, and it is attested for Japanese baby
talk as well.
Distant nasal assimilation is attested for Marathi, Gilyak, and Spanish
(e.g., Spanish mamoch for vamos), and may also occur in the others.
Examples of loss of unstressed syllables occur in English and Spanish (e.g.,
Spanish tines for calcetines).
Grammar
At least one diminutive or hypocoristic affix is of frequent occurrence in
each language. This may be a regular diminutive form (as Spanish -ito, -ita or
Comanche -ci) or a form used chiefly in baby talk and only infrequently in
normal language (e.g., Gilyak k/q, Marathi -[k]ula/ -ukla, Arabic -o, English
-ie).
Greater use of nouns rather than pronouns and verbs is general: equational
clauses without verbs replace normal construction with copula or verb (e.g.,
English dollie pretty for the doll is pretty), and third person constructions re-
place first and second person ones (e.g., English daddy wants for I want).
In two of the languages, Arabic and Marathi, a shift in gender is used as a
mark of endearment; i.e., a feminine noun, pronoun, adjective, or verb form is
used in reference to a boy or vice versa. For example, in Arabic win ruiti yd

Notesto TableI
1. Marathiand Englishhave many baby-talkwordsfor 'mother'and 'father.'Marathimai,
ai (regular adult word), mami (English loan), 'mother'; baba,
1~iji, dada, tatya, tata, appa,
nana, aba, bhau; papa, dedi (English loans), 'father.' English mom, ma, momma, mommie;
an.na,
dad,daddy,dada,pop.Casagrandesays "Thereis no specialbabyword[formother]in common
use," the regularadult pid being used (1948: 12); mamdtPis includedin the alphabetized
listing,however,identifiedas English.Spanishbabytalk mama(alsomami)is stressedon the
firstsyllable,;mamdwith finalstressis a somewhatinformaladult word.
2. ComanchePap?i?also means 'father'sbrother'and 'father'sfriend';baby talk tok6-2,
tot'-?
'grandfather'(adult counterpart'mother'sfather') is sometimesused for 'father.'Spanish
tatamay also be used for 'grandfather.'
3. Arabicbubbuand Comanchenii~? are also usedfor 'doll.'Marathibal is also an adult word,
but is usuallyin baby talk with a specialintonation,and otheradult wordsfor 'baby'arenot
used in baby talk. The Gilyaknenais glossedonly 'doll.'A femininenenaoccursin Spanish,
althoughin Chilenenemay be used for both sexes.
4. Englishapparentlyhas no commonbaby-talkwordfor 'food'; yum-yum'delicious'is some-
times used.
5. Marathipapa also means'kiss.'
6. Arabic has variants such as ?a??d ninni (-e), ?o??j ninni (-e). Spanish hacer tuto is attested
for Chile,hacermeme(or mimi) for Mexico.
7. Spanishpichi, chichiare not attestedfor Mexico.
8. Chile:pop6is 'anus,'sometimes'vagina,'never'defecation'or 'feces'; kakais not attestedfor
Mexico.
9. Arabicgullujullu is from McCarus.
10. Spanishteneruna yaya (oryayita)is attestedfor Chile,hacersecocofor Mexico.
11. Gilyakamqamqis 'walk'; tonk (variantsVon,uonuon,mono)is 'legs and feet.' Spanishpatita
'foot' is attested for Mexico in the sense of 'walking,taking steps' Chile has
andandopatita,Mexicokacerpininos.
and,.Chile;
 TABLEI

Arabic Marathi Comanche Gilyak

KIN
1. mother mama (m)ai - yma
2. father bAbab baba ?api.? da(j), dyj
3. baby bubbu bal (nena)
nini'?
BODY
4. food mamm mammam tatA-? mama, fiafia
5. drink, water mbii (wa) papa pap-/ -
6. sleep ninni, ninni nini, 3030 - qoq
gai (gai)
7. urination ?ahh mumu, ?u - hisa, cisa

8. defecation kaCqo (i)gi ?a?h, [a?a]

-?asi" ypyp
9. bath Aullu Hullu toto
10. hurt wawa bau ykyk
11. ?an'-,- nan'-?
walk, foot dade calcal jonk, amqamq
12. breast, milk zeze, zizz pipi cici-? myfik, myfiy
13. penis - nuni, nunu wf?asI coc(k)
14. vagina 6imi tA?sI bew, pelrja

Table I (Continuedon nextpage)

 TABLE I (Continued)

Arabic Marathi Comanche Gilyak

QUALITIES
15. nice dahh han 6han ?um.ia ulak
16. bad, don't! didde, (hu)mm ha(?) ?a?h-? -
17. dirty kixx, kacq yakk, isi? ;ax alqalq
18. hot ?uhh hay ?iti? -
19. cold hia gar gar -
20. nothing left bahh koko ?ici'? - ap(k)a
21. little nfinu pitukla - -

ANIMALS & GAMES

22. dog ?aw'aw bhubbu gyck, gycy
p"/op6"?
23. cat nawnaw, biss mau, mini -
wa?6"?
24. bird kfiku 6iu kaki ? bic-(U)aq
25. goblin bagul-bua muki'? humk
26. going out buTbu,
tiSS bhur --
27. peek-a-boo naww, ba ??;no kukk, bua, --
28. carry on back ha?ha? kokru ho- mama"? aci, (b)apu
29. kurr kurr -
noise,ear
30. goodies, candy nahh khau kok6? -
 Baby Talk in Six Languages 109
binti? 'Where did you go (fem.), little girl?' said to a boy; inta iu'an? 'Are
you (m.) hungry (m.)?' said to a girl.8 In Marathi the examples are with the
use of a feminine ending on a boy's name and vice versa.

LEXICON
The number of lexical items given in the references varies from about 25 to
over 60. The commonest topics reported are: kin names, nicknames and the like;
body parts and bodily functions; basic qualities like "good," "bad," "little,"
"dirty"; and the names of animals and nursery games. About 30 such items
common to most of the six languages are listed below, classified under four
headings; in several cases attested items modified from adult words are entered
when there is no special word.

CHARACTERISTICS
Baby-talk words either as modificationsof normal words or as special lexical
items show certain general characteristics.In the first place, baby-talk items con-
sist of simple, more basic kinds of consonant, stops and nasals in particular,
and only a very small selection of vowels. One would expect that the rarer,more
peculiar consonants or the consonants which tend to be learned later would not
be found in baby talk, and generally this is true but there are some exceptions.
Gilyak, for example, uses four phonemically distinct nasals in baby talk, and a
variety of velars as mentioned above. Arabic has many baby-talk items with
pharyngeal spirants although these are often assumed to be learned late in Ara-
bic. The best example is the fact that labial emphatics exist in Arabic baby talk
and may well be the first emphatics learned by Arabic children even though-
they are marginal in the adult language.
A second phonological characteristic is the predominance of reduplication,
both of parts of words and of whole words, in the baby talk of all six languages.
For several of these languages reduplication plays a grammatical role of some
sort in the adult language, but the reduplication in baby talk is generally sepa-
rate and unrelated to the use in the normal language. Reduplication can prob-
ably be regardedas a feature of baby talk throughoutthe world.
Each of the six languages has a typical ("canonical") form of baby-talk

15. Marathi than than is a reduplicated form of adult than. Spanish nino (adult lindo) is attested
only for Chile.
16. Arabic didde means 'don't or I'll slap you, spank you'; (hu)mm means 'don't touch.'
17. Spanish chucho is a baby talk form of sucio 'dirty'; possibly fuchi is also related to this.
18. Spanish ssss is accompanied by a gesture of shaking the fingers loosely as though just burnt.
19. Marathi gar gar is a reduplicated form of adult gar.
22. Chile guau gudu, Mexico gua gud.
24. Marathi Hiuis glossed 'house sparrow.' Comanche kaka-2 is also 'headlouse.'
25. Comanche has several words for frightening children; the muki-2 is some kind of giant owl,
the mumd-? is darkness or thunder, the ?nin• is a noxious insect or small animal like a snake
or scorpion. Chilean Spanish sometimes has cuca, feminine of cuco.
27. English peek-a-boois chiefly American; the usual British form is bo-peep.
28. Marathi kokru ho- means 'play lamb,' i.e., be carried piggy-back. Comanche mamd-2 is
glossed "horse; said by a child when he wants to be carried on someone's back."
110 The Ethnographyof Communication
items. There is variation, dependent at least in part on the canonical forms of
morphemes in the corresponding adult language, but the commonest form is
CVC, i.e. a monosyllable beginning and ending with a consonant, with CVCV
as next most common. Many items have CVCCV with a double consonant in
the middle even if this is not common in the adult language. As an example of
the variation conditioned by normal canonical forms we may cite Spanish: in
adult Spanish, monosyllabic words of the shape CVC are extremely rare, and this
form seems not to occur in Spanish baby talk, where CVCV is the commonest
form.
On the grammatical side, apart from the reduplication and canonical forms
already mentioned among phonological characteristics,the most striking features
are the absence of any inflectional affixes, the presence of a special baby-talk
affix and the use of words in different grammatical functions. The semantic
fields showing a special baby-talk vocabulary most commonly represented in-
clude kin, food, body parts, and animals.
It must be noted that the features listed here as characteristic of baby-talk
items are in general characteristic of the one-vocable utterances ("monoremes")
used by children at the stage of linguistic development between the stage of
call-sounds and other prerepresentationalitems and the stage of two-vocable ut-
terances where words and sentences emerge.9Commoncharacteristicsinclude re-
duplication; primitive affixes; food, animals, toys, etc., as referents.
In view of this similarity one is tempted to make the hypothesis that every
language community provides a stock of baby-talk items which can serve as ap-
propriate material for babies to imitate in creating their monoremes but which
do not interfere with the normal words of the language and can gradually be
discarded as real words emerge in the children's speech. The child may, and
often does, create his monoremes from other sources such as sound imitation or
fragments of adult utterances, but the baby-talk items tend to be one of the
principal sources. The baby-talk lexicon of a language community may thus play
a special role in the linguistic development of its children: the facilitation of
each child's acquisition of a set of monoremes from which he can go on to the
beginnings of real grammar.Experimentalconfirmationof this hypothesis would
be difficult; perhaps the most relevant data would come from societies with radi-
cally different attitudes toward child language learning. (Cf. Voegelin and Robi-
nett 1954.)
FUNCTION
Under what circumstancesand with what intentions is baby talk used? The
published material is very limited on this point. There are, however, several situ-
ations or purposes mentioned in the articles or by informants, and these may be
considered.
Perhaps the primary purpose is felt to be teaching a child to talk; that is,
people asked why or when they use baby talk will say that they use it when talk-
ing to young children to make it easier for them to learn to talk. If asked in more
detail they may explain that what they are saying in baby talk is easier for the
child to learn and that it is clearer, i.e., easier for the child to hear; also, espe-
 Baby Talk in Six Languages 111
cially in the Marathi material, whenever there is a choice between two ways of
saying something, baby talk uses the more colorful, more "marked"in the lin-
guistic sense. This feeling is obviously incorrect in details (is pussy so much
easier than cat?) and too vague in formulation, but it seems to reflect in a folk-
wisdom way the function hypothesized above. A moment's consideration, how-
ever, shows this is not the only time baby talk is used. It is used for one thing
in talking to infants who are not yet learning to talk, and it is apparently used
in talking to pets in every one of these six language communities.Obviously one
is not teaching the infant or the pet to talk.
Secondary uses of baby talk generally seem to reflect a desire on the part of
the user to evoke some aspect of the nurturant-babysituation in which the pri-
mary use of baby talk occurs. This evocation may be from the side of the baby.
For example, a child who has just gotten past the use of baby talk by his par-
ents may then revert to baby talk-in fact, even use baby talk that he has not
used before-in order to get attention or to be treated in some way as a baby.
Also, adults use baby talk in reporting children's speech; in several language
communities (e.g. Marathi, Norwegian) baby talk is often used to represent
child speech in written literature such as novels and stories.
The evocation of the nurturant-babysituation may also be from the side of
the nurturant. For example, the use of baby talk to pets or small infants seems
to show the kind of protectiveness and affection characteristicof the nurturant's
relation with the baby. The Marathi author notes that the speaker gets a sense
of pleasure from doing this.
In Marathi, English, and Spanish, lovers' use of baby talk is attested, and
in this case it may not always be clear whether it is the protectiveness of the
nurturant or the dependence of the baby that is evoked. It is worth noting that
Kelkar reports, on the basis of observation in multilingual situations, that adults
who are using baby talk with other adults do not use baby talk in anything but
their own language. It seems very likely, however, that this varies depending on
a number of factors; it is in any case related to the important general issue of re-
lationship-signaling styles in a second language.
Finally, it is clearly documented for several languages that baby talk is used
in certain kinds of songs, riddles, and word-play on the part of adults which
bear little direct relationship to the uses with children (Austerlitz 1956:272-3).

VARIABILITYAND DIFFUSION
The fact of variability in baby talk was mentioned above; it requires fur-
ther comment here. First, there is great family variation: an item gets used in a
certain family and becomes well entrenched there but does not spread beyond
that. There are also examples of items spreading from one family to another but
not becoming general.
Second, there is the areal diffusion prevously referred to. Baby-talk items
often diffuse within an area rather than according to the lines of genetic rela-
tionship followed by the great mass of linguistic phenomena. A good example
is the baby-talk word [kix] meaning 'dirty, don't touch' and the like. This word,
112 The Ethnography of Communication
with slightly different forms depending on the phonological systems of the re-
spective languages, occurs in almost every language of the Middle East. It is
attested (McCarus 1963) for Arabic, Kurdish, Persian, and Syriac although these
languages represent two different language families, Semitic and Indo-European
(Iranian branch). The word [kix] is not attested for Turkish, which has no
phoneme of the [x] type. Another good example is the use of a word like wdwa,
uwwa, or vava in the meaning 'hurt, sore, injury' throughout the Middle East
(Arabic, Syriac, Turkish, Persian, Armenian,Greek), with a [v] in languageslike
Persian and Greek that have no regular phoneme of the [w] type.
The explanation for this kind of diffusionmight lie in the fact that the baby-
talk items are not well integrated into the grammatical system of the language
even though they are fairly well integrated into the phonological system. Be-
cause of this lack of integration it is clearly easier to borrow these terms from one
language to another, but presumably social factors in addition to this linguistic
factor should be sought as explanation.
This kind of variability, being relatively independent of genetic relation-
ship, offers a chance for the study of distribution of baby-talk items on a statis-
tical basis throughout the world and the kind of analysis of statistical universals
of one sort or another that Jakobson has tried (Jakobson 1962), at least with
mama and papa, suggesting certain reasons for their occurrence with far more
than chance frequency in languages of the world. It is a rare pleasure for the
linguist to have a language phenomenon which can be studied all across the
world without need for corrections from the genetic relationships that are in-
volved.
Another way in which baby talk can vary from one language to another is
the size of the lexicon or the range of variation of a particularpart of the lexicon.
Actually one of the surprising features of the present study is the similarity of
baby-talk phenomena in the six languages considered, when one might have as-
sumed that there would be serious cultural differencesin the kinds of items that
would appear in baby talk and the situations in which they would be used. Fur-
ther study along this line, however, would be useful.
One other point of variability should be mentioned, the differences in atti-
tude toward public use of baby talk. In our society baby talk is mentioned with
an air of apology by adults talking seriously, and one feels a good bit of embar-
rassment in citing examples of baby talk. Also in our society it is quite widely
believed that the use of baby talk inhibits learning of the language. That is,
people feel that if they use too much baby talk at home, the child is not going
to learn the normal language properly. This belief is presented explicitly in books
on child development, although there seem to be no experimental data which
would substantiate it.1o In the Arab world, however, there seem to be no such
feelings. Adults may discuss baby talk perfectly easily, and they use it freely if
it is appropriate.There seems to be no trace of the notion that use of baby talk
may inhibit the acquisition of the adult language. Among both Americans and
Arabs, however, it seems to be felt that baby talk is more appropriatefor women
to use than men.
 Baby Talk in Six Languages 113

SUMMARY
Baby talk is a linguistic subsystem regarded by a speech community as be-
ing primarily appropriate for talking to young children; it consists of intona-
tional features, patterned modifications of normal language, and a special set
of lexical items. The special lexical items typically number between 25 and 60
and cover kin names and appellations, bodily functions, certain simple quali-
ties (e.g., dirty, pretty, hot, cold), and vocabulary concerning animals, nursery
games, and related items. Baby-talk words typically contain stops, nasals, and
a limited selection of vowels, have the structure CVC or CVC(C)V, are frequently
reduplicated, and often have a diminutive suffix characteristic of baby talk in
that language.
Baby-talk works are not universal, but are transmitted much like other lan-
guage phenomena in the community. Baby talk seems to serve in each language
community as a special source for children's pregrammaticalvocables, enabling
them to create items at that stage which they can discard as they acquire true
words and grammar.Baby talk in addition to this primary use is also used to talk
to infants and pets and between adults in situations with "baby" aspects.
Baby-talk items are fairly well integrated into the phonological system of the
language, but are so unrelated grammaticallyto the normal that on the one hand
they show considerablevariability within a speech community and on the other
hand tend to diffuse readily across language boundaries regardlessof genetic re-
lationships. A given baby-talk system may be characterizedin terms of internal
structure by the size of the special lexicon and the range of variability. Exter-
nally it may be characterizedby the extent of its secondary uses and the attitude
towardits public use.
NOTES
1As an additional source for Syrian Arabic, McCarus' notes were used; they also pro-
vided information on baby-talk items in Iraqi Arabic, Turkish, Kurdish, Persian, Syriac, and
Alexandrian Greek. Further checking of Arabic was done with Mr. and Mrs. Moukhtar Ani
of Damascus. Kelkar provided some additional Marathi information in a personal communica-
tion. Chief informants for the Spanish were Mrs. Raquel Saporta of Chile and Miss Yolanda
Lastra of Mexico; English items came from the author and his colleagues. Susan Ervin-Tripp
read the manuscript and made valuable suggestions.
2
Sabbagh's sketch (Sabbagh 1886) of colloquial Syrian and Egyptian Arabic, written in
1812, has five baby-talk words (voweling uncertain): bahh 'all gone,' dahh 'shiny, nice,' 'uhh
'hot,' 'goo,' said to elicit smile and first word, mnakh 'sweet, goodies.' All these are in
use in nci•,
Syrian Arabic today (modern form for the last two nkikg, nafh).
'Varr. ap. Non. 81.2 cum cibum ac potionem buas ac pappas vocent et matrem mammam
patrem tatam (Heraeus 1904, repr.: 170-172).
'This notion of universality is found even in such careful works as Lewis (1957: 80)
"In fact, baby language is an international language. If we make a short list of the earliest
words actually spoken by children, with their meanings, we have a vocabulary that every
one will recognize."
"Kelkar pays considerable attention to intonation in his study 3.2.
SThe careful account of the phonological characteristics of Norwegian baby talk in Haugen
(1942 : viii-x) includes most of the characteristics listed here.
SSurprisingly enough, Spanish baby talk shows distinctive use of stress, e.g. pipi 'bird':
114 The Ethnography of Communication

pipi 'urination.' Also, several baby-talk items differ from other adult words only in stress;
for example, baby talk mama 'mother' and papa 'food' differ from informal adult mam'd,
papd 'mother,' 'father,' and baby talk gua gud differs from adult Caribbean Spanish gudgua
'bus' and Bolivian gudi gua 'child.' Spanish baby talk has both CVCV (= CVCV) and CVCV
as canonical forms.
SArabic examples are from McCarus (1963).
SSome monoremes persist as vocables in more complex utterances, but the notion of a
monoreme stage in language development seems valid. A convenient recent account of the
characteristics of monoremes is in Werner and Kaplan (1963: 134-137). Full recognition of
the similarity between baby talk and actual items of child language is found in Jakobson
(1962: 539): "Nursery coinages are accepted for wider circulation in the child-adult inter-
course only if they meet the infant's linguistic requirements ...."
"
This notion appears even in careful reviews such as McCarthy (1954: 536): "... baby-
talk used by adults in the child's environment often makes for preservation of infantile speech
habits." A more balanced statement on this point appears in Lewis (1957:89): "But a mother
who, because of a theory that baby-language is too 'babyish'-not 'correct language'-refuses
to speak it to her child may be doing him harm, retarding his language development. On the
other hand, if baby language is spoken to a child for too long in his life he may be retarded
in another way-his speech may remain childish at a time when he should have grown out of
this."
REFERENCES CITED
ROBERT
AUSTERLITZ,
1956 Gilyak nursery words. Word 12 : 260-279.
JOSEPH B.
CASAGRANDE,
1948 Comanche baby language. International Journal of American Linguistics 14: 11-
14.
CHARLES
FERGUSON, A.
1956 Arabic baby talk. In For Roman Jakobson, Morris Halle et al., eds. The Hague,
Mouton.
HAUGEN, EINAR
'1942 Norwegian word studies, Vol. I, Part III (Baby talk, pp. vi-x). Mimeographed,
on deposit Library of Congress.
HERAEUS, WILHELM
1904 Die Sprache der r-mischen Kinderstube. Archiv fiir lateinische Lexikographie 13:
149-172. Repr. in Kleine Schriften von Wilhelm Heraeus. J. B. Hoffmann, ed.
Heidelberg, Carl Winter's 1937, pp. 158-180.
ROMAN
JAKOBSON,
1962 Why "mama" and "papa"? In Selected writings, Vol. I. The Hague, Mouton.
KELKAR,ASHOK
1964 Marathi baby talk. Word 20: 40-54.
LEWIS, M. M.
1957 How children learn to speak. London, George G. Harrap.
MCCARTHY, DOROTHEA
1954 Language development in children. In Manual of child psychology, Leonard
Carmichael, ed. New York, John Wiley.
MCCARUS, ERNEST
1963 Near Eastern baby talk. Unpublished notes.
SABBAGH, MIKHA'IL
1886 Miha'il Sabb&-'s Grammatik der arabischen Umgangssprache in Syrien und
Aegypten, H. Thorbecke, ed. Strassburg, K. J. Triibner.
VOEGELIN, C. F. and FLORENCE M. ROBINETT
1954 'Mother language' in Hidatsa. International Journal of American Linguistics
20: 65-70.
WERNER, HEINA and BERNARD KAPLAN
1963 Symbol formation. New York, John Wiley.