A FIELD MANUAL FOR CROSS-CULTURAL STUDY OF THE ACQUISITION OF COMMUNICATIVE COMPETENCE. SECOND DRAFT--JULY 1967.

Survival, Water, Medical Field Manuals

Military Manuals

Slobin, Dan I., California Univ., Berkeley.

Document text

R € @ © R FT RK E S$ U M E § . 0 


ED 015 434 24 AL 999 691 
A FIELD MANUAL FOR CROSS-CULTURAL STUDY OF THE ACQUISITION OF 
COMMUNICATIVE COMPETENCE. SECOND ORAFT--JULY 1967. 

BY- SLOBIN, CAN I. AND OTHERS 
CALIFORNIA UNIV., BERKELEY 
REPORT NUMBER BR-6-8757 FUB CATE JUL 67 
CONTRACT OEC-4-7-998757-2015 
ECRS FRICE MF-$1.99 HC-$15.28 255F. 


CESCRIFTORS- PSYCHOLINGUISTICS, *CHILO CEVELOFMENT, 
SOCIOLINGUISTICS, *LANGUAGE CEVELOPMENT, LANGUAGE ABILITY, 
APPLIED LINGUISTICS, STRUCTURAL ANALYSIS, *LANGUAGE RESEARCH, 
*FIELO STUDIES, FIELO INTERVIEWS, *RESEARCH METHODOLOGY, 
GUIDES, MANUALS, RESEARCH CRITERIA, TAPE RECORDINGS, SOCIAL 
CHARACTERISTICS, RESEARCHERS. CULTURAL FACTORS, 
BIBLIOGRAFHIES. VERBAL COMMUNICATION, RESEARCH SKILLS, 
RESEARCH DESIGN, FSYCHOLOGICAL STUDIES, CHILD LANGUAGE, CROSS 
CULTURAL RESEARCH, 


THIS FIELD MANUAL WAS CEVELOFED AS FART OF A FROGRAM TO 
TRAIN INVESTIGATORS IN THE COLLECTION OF COMPARABLE 
CROSS-LINGUISTIC ANG CROSS-CULTURAL CATA ON CHILDREN'S 
ACQUISITION OF LINGUISTIC CODES AND SOCIAL RULES FOR THE USE 
OF SUCH CODES. ALTHOUGH COMFARATIVE RESEARCH REQUIRES 
FROCEDURAL STANDARDIZATION, THE FHYSICAL AND SOCIAL 
CONDITIONS IN THE VARIOUS FIELD SETTINGS INTERFERE WITH 
STANDARDS SET FRIOR TO ARRIVAL IN THE FIELD. THIS MANUAL 
CONSTITUTES A COMMON FRAME OF REFERENCE FOR INVESTIGATORS 
COLLECTING CATA THAT IS TO BE USED IN COMPARATIVE REFORTS. 
FART I OF THIS MANUAL CEFINES THE MAJOR TOFICS TO BE 
INVESTIGATED AND SETS UF A RESEARCH SCHEDULE. FART II 
CISCUSSES SUCH METHODOLOGICAL FACTORS AS CONTRASTIVE ANALYSIS 
OF LANGUAGES AND CULTURES. RECORDING AFFARATUS AND 
TECHNIQUES. AND NATIVE INFORMANTS, ASSISTANTS AND 
INTERPRETERS. THE THEORETICAL CONSIDERATIONS INCLUDED IN FART 
ITI ARE FHONOLOGY, GRAMMAR, SEMANTICS, THE SOCIAL SETTING OF 
LINGUISTIC BEHAVIOR, AND LANGUAGE USAGES AND STYLES. AFFENDECD 
ARE EXTENSIVE REFERENCES AND ADDITIONAL SUGGESTED FIELD 
TECHNIQUES. THE REACER OF THE MANUAL SHOULD HAVE SOME 
KNOWLECGE OF LINGUISTICS BUT NEED NOT HAVE CONDUCTED FREVIOUS 
FIELD WORK. REQUESTS FOR COFIES OF THE MANUAL AND FOR TAFES 
TO BE USEC IN FHONOLOGICAL IMITATION FROCECDURES, AND 
INFORMATION ABOUT RELEATED STUDIES SHOULD BE ADDRESSED TO OR. 
CAN I. SLOBIN, CEFARTMENT OF FSYCHOLOGY, UNIVERSITY OF 
CALIFORNIA, BERKELEY, CALIFORNIA 94720. (JD) 


aq 


ED015434 


AL 000 601 


A FIELD MANUAL 


for cross-cultural Study of the 
acquisition of communicative competence - 


(Second draft -- July 1967) 


Edited by 


Dan |. Slobin 


And written by 


Susan M. Ervin-Tripp 
John J. Gumperz 
Dan |. Slobin 

Jan Brukman 

Keith Kernan 

Claudia Mitchell 
Brian Stross 


University of California 
Berkeley 


U.S. DEPARTMENT OF HEALTH, EDUCATION & WELFARE 
OFFICE OF EDUCATION 


THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE 
PERSON OR ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINIONS 
STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION 


POSITION OR POLICY. 


A FIELD MANUAL 
for cross-cultural study of the 
acquisition of communicative competence 


(Second Draft -- July 1967) 


Edited by 
Dan |. Slobin 


And written by 


“DERMISSION TO REPRODUCE THIS 


waren Ms = ang Susan M. Ervin-Tripp 
BY Deters John J. Gumperz 
Oa ee MK TE US. OFFICE OF Dan |. Slobin 


FURTHER REPRODUCTION OUTSIDE 
age REQUIRES PERMISSION OF Jan Brukman 
CORRE OWNER.” Keith Kernan 


THE 
Claudia Mitchell 
Brian Stross 


University of California 
Berkeley 


With support from 


Institute of Human Learning 
University of California, Berkeley 


National Science Foundation 
Project Literacy 
Social Science Research Council 


United States Office of Education 
(Contract 4-7-008757-2015) 


; 
Lj 


Contents 


INTRODUCTION e 7 * e e * ° o * . * . . * ” 


PART ONE: 


1.0. 
1.1. 
1.2. 


1.3.6 


La 


1.5. 


1.6. 
1.7. 


PART TWO: 


2el. 


2.2. 


2.3. 


MAJOR TOPICS OF INVESTIGATION 


Research Schedule . . .« « « « « « 
Motor Development ....+-«-e««-. 
Linguistic Development: Phonology 
1.2.1. Definitions ...... 
alt oo Imitation * e e cd * o 2 
hedede Babbling oeeoeeee#e @ 
1.2.4. Speech Samples ..... 
Linguistic Development: Vocabulary 
1.3.1. Initial Vocabulary ... 


1.3.2. Vocabulary Related to Grammatical Categorie 
Linguistic Development: Grammar... . -« 
1.4.1. Collection of Basic Corpora 
1.4.2. Elicited Sentence Imitation 


1.4.3. Speech Comprehension. . 
1.4.4. Elicitation of Speech . 
Communicative Development. . . « « 
1.5.1. The Child's Lifespace . 
1.5.2. Dialog between Children 
1.5.3. Parental Baby Talk... 
1.5.4. Communicative Routines . 
Linguistic Belief Systems. .... 
Notes and Queries for Observation. 


METHODOLOGICAL FACTORS 


Contrastive Analysis . . « « « « 
21.1. Methods . .« « e e e 
2.1.2. Style vs. Speaker .. 

2.1.3. Functional Realization 

2.1.4. Linguistic Contrasts . 

2.1.5. Vocal Signals .... 

2.1.6 

2ek07 

Cc 


- intelligibility ... 
- Evaluation . . . «se. 

rding Apparatus and Techniques 
2.2 ode Apparatus * * * * a * * 
2.2.2. Recording Techniques . . 

2.2.3. Care and Servicing... 
EE EE eee 


Native Informants, Assistants, and Interpreters 


* 
e 
+ 
e 
e 
e 
e 
e 
° 
4 
a 
e 


= 
o 
° 
e 
* 
* 
e 
@ 
e 


° e e e e . + oe - 
a ® & J e . . * * 
e © * + e a * s . 


2.3.1. Factors of Social Organization... 
223-2. Problems of Social Change. ....« -« 
2.3-3- Social Criteria in the Selection of 


Informants and Assistants 


iii 


+ a e . e* e e @ * 


s * . 2 . . e » * = * e 2 s e - 


ae 3s e @ e - e e e 


e* @ e * es @ e @-.e © e* e 


eee ee eS eee eee rere eee ee ey Pe re oe ee ae 


an | 


Page 
PART THREE: THEORETICAL CONSIDERATIONS AND POSSIBLE ‘y 
SUBSIDIARY STUDIES t 
3.1. Phonology ...«-e«eseseever ete ee eee ee @ 93 
3.1.1. Pre-Linguistic Studies eeeeeee# @ 93 
3.3.8. Reirearns CYAOPR + ee re ee eo es 95 p 
3.2. Grammar ...-c«4-c-eeee ee 2 2 2 © © e 8° e860. 105 


3.2.1. Aspects of the Acquisition of English 
Grammar . e e es e * a td oe eo a o a e e & © 106 
3.2.2. Implications and Suggestions for Cross-~ 


Linguistic Research ...-+.++e-+ee- 116 N 

3 3 ° Semantics e e e * e e o i * * e . * e * © . e a e 121 ty 
3.3.1. Languege and Cognitive Development ... 121 
3.3.2. Componential Analysis .......-+.e. 12h 
3.4. The Social Setting of Linguistic Behavior ...+-+-+ 4129 


3.4.1. The Relation of Linguistic to Social “ 
Categories a e a td e se & a - eo + a e o * 129 
3.4.2. Relation of Language Development to 


Other Areas of Socialization. ....e.e- 135 : 
3.4.3. Informal Education ....-2-eceeee 137 
3.4.4. Formal Education ....-+eeeeee LY 
3.5. Usage and Styles . 2. + «e+ ee ee ee © © © © © @ 155 
3.5.0. Introduction .....2e2eercececeee 155 
3.5.1. The Acquisition of Usage Pattermns .... 157 
3.5.2. Special Languages, Verbal Art, Speech i 
Playe «© «eee ee eevee ee ee eo @ @ 166 : Nis 
3.5.3. Natural Conversation .....2e-e-ee- fe 
3.5.4. Multilingualism and Multidialectism... 176 
3.5.5. Linguistic Taboos ....+-++eee-ee- 180 
3.5.6. Speech Acts: Notes for Designing a Cross- 


Cultural Study of the Speech Act of | 
Promising e e * * . = e e e e e e e e e e 186 


APPENDICES - 
1. Techniques for Studying Grammatical Development... 193 sis 
1.1. Model Sentences for Elicited Imitation .. 193 
1.2. Comprehension Tests... 2. 2eeeeeeee 198 ) 
1.3. Elicitation Techniques .....2+e+-+e-e+ 201 Uy 
2. Transcription of Tapes .. + s«+«-eeeee ee ee 2 212 
2.1. Nature of Transcription . ...+«+e«e«e«¢ee 212 
2.2. Suggested Transcription Format for Dialog . 214 
2.3. Suggested Transcription Format for ea 
MOM cp cece eee eee eee oe) «6 
3. Techniques for Studying Multilingualism. ...... 218 | 
3.1. Picture Vocabulary Test .....-2eee+ee 218 . 
3.2. Word Association Test .. «2 e«-2eeee- 220 
3.3. Following Directions. .....+s-seeee- 220 | 
3.4. Self-Ratings. oeeee ee © © @ © © &© & @ 221 . 


iv 


eo 
a ¢ 


me ee et ee 


h, Useful Tables .. 2+ ees eeee eee e eee ee 2% 222 
oe Age. Calculation Chart «2«ee+e-eee2822% 2% © 222 
4.2. International Phonetic Alphabet eececeeeeees., 223 
5. Selected List of References on the Acquisition of 21 


Different Native Languages ..-+«e«eeeeee ee © © 22% 


REFERENCES 


eeeeeee#e 


eeeeeee#e# eee ¢ #® # #@# eeeee®*® 227 


Tables 


1. Major Research Tasks . 2.22 eee eee eee eee eevee 
2. Distribution of Children by Age, Sex, and Task .... « « 


36 12-Month Time Schedule eeeeeeegeeeeeeseeee#e# # # #® @ 


4. Developmental Milestones in Motor and Language Development 
(from Lenneberg 3 1967) e e & a @ 6 e ® e a % e ® . @ 2 oe eo 7 


5. General Scheme of Phonemic Development in Russian Children | 
(after Shvachkin, 1948) . 2.2.2 eeecrcecercee eves 96 : 


6. Fragment of Pivot Grammar of One Child . . «+ «ee +e» 107 


7a Substitution of Phrase for First Member of Open-Open 
Sentence .-«ee-ererrerevveevreevr eer eee eee # © @ @ 108 


7 Substitution of Phrase for Second Member of Open-Open 
Sentence -_ e e a e e e * * e @ ° eo a e e @ e a e s a © o * ® 109 


8. Emergence of Noun Phrases .. 22+ eeeeeeceee ee 6 110 
9. Negation Stage I... ee eee cece reer ee evo ee se ID 
10. Negation Stage II ..+e«eeere eee eee eee eee 3 
| 11. Negation Stage IIIT... 2 ceececcrccereeeececreceose Ji 
12. Negation Stage IV ..-2«2e2eeeere ee ee eee eo © © FBS 
13. Classification of Imitation Situations (from Slobin, 1964). 17 


14. A Partial Checklist of Potential Taboo Domains (after 
Kany, 1960) ...cecesceceecveerereececerserves 185 


vii 


INTRODUCTION 


This field manual is intended to guide investigators in the col- 
lection of comparable cross-linguistic and cross-cultural data on the 
acquisition of communicative competence. It thus represents a joint 
endeavor in developmental psycho- and sociolinguistics, for we are 
concerned both with children's acquisition of linguistic codes and the 
social rules for the use of such codes. 


‘ There have recently been numerous studies employing modern lin- 
guistic concepts in the study of early language development in chil- 
dren (reviewed in Ervin-Tripp, 1966). Some longitudinal studies have 
explored the sound system (e.g. Burling, 1959; Gvozdev, 1948); others 
have examined syntactic development (e.g. Braine, 19632; Brown & 
Fraser, 1963; Gvozdev, 1949; Miller & Ervin, 1964). In same cases 
cross-sectional methods have been employed (e.g. Berko, 1958; Templin, 
1957), establishing norms for age changes in a particular language 
comunity. These studies have raised many questions about how children 
learn the coniplex linguistic system of any lenguage rapidly, during the 
same years, and yet reveal relatively small variation in rates of ac- 
quisition though there are large variations in practice and exposure to 
speech. We believe that there are three major gaps in our research on 


language development. 


We need information about the learning of languages which are 
structurally different from English. There have been proposals (e.g. 
Chomsky, 1965, 1966, 1967; McNeill, 1966a,b) that certain grammatical 
structures are universal, and that certain universal cognitive abili- 
ties are likely to be involved in language learning. We may also find 
certain universals in terms of stages of development in children 
(Jakobson, 1941). All of the recent American studies of syntax have 
found that at a very early stage children have "pivot" grammrs — 
(Braine, 1963a) consisting of classes of a few words which occupy com- 
plementary positions. For example, in a given child's grammar, where 
or there may always come first in sentences, and up and off may always 
come last. English has rather variable absolute position of words-- 
certainly off and up do not always occupy sentence final position. An 
even more interesting example is the learning of Russian. The scanty 
evidence available suggests that Russian children also have pivot 
grammars at first, even though Russian word order is extremely variable, 
since it is a language which, like Latin, employs inflectional affixes 
rather than order for many basic grammatical relations (Slobin, 1966a, 
in press). Therefore the pivot grammars cannot arise from imitation. 


The variety of systems in natural languages is so great that the 
study of language acquisition in varying languages can be regarded as 
a natural experiment which it would be extremely difficult to simulate 
artificially. For example, American Indian languages have far more 
complex verb inflections and verb morphophonemics than English but we 
do not know about child language development in any American Indian 


The Introduction was written by Susan M. Ervin-Tripp. 
ix 


language. The attempt to simulate primary language learning with an 
artificial language is manifestly extremely difficult, and so far ex- 
periments of this sort have been very limited in scope, have studied 
older children,and have essentially been second language studies. 
There is reason to believe that first and second language learning may 
be quite different, and that one cannot generalize from one to the 
other. 


The second desideratum is that language acquisition studies be 
broadened to include not only the traditional formal core of language, 
but competence in the use of language. Not only do children learn the 
phonology, grammar and vocabulary which permit us to identify a lan- 
guage, but they learn when to speak, when to be quiet, when to use 
ritual language, when to use baby talk, when to use polite forms, and 
when to shift language in @ multilingual community. We know nothing 
about the relation between these aspects of learning and the learning 
of the basic code, but it is clear that instruction about when to vo- 
calize and when to be silent can begin even before language begins. 
Differentiated style according to the age of addressee can be seen in 
children under three, as shown in baby talk addressed only to infants, 
polite requests only to older children or adults, and systematic dif- 
ferentiation of multi lingual usage. 


The uses of language may vary considerably in different cultures. 
Theoretical discussion of one aspect of functional variation has be- 
come well known through the writing of Basil Bernstein (1961a,b; 1962, 
1964). In brief, he argues that members of closed social networks use 
"restricted codes” which are highly redundant. In contrast, members 
of open networks and their children can switch to an "elaborated code” 
which is less redundant, uses more complex and varied forms of sub- 
ordination, and relies on vocabulary and syntax more than on voice 
quality and paralinguistic signals for differentiating meaning. Under - 

these formal contrasts there appear to be differences in the 
usual functions of language. The acquisition of full competence in 
writing requires the acquisition of an elaborated code, so analysis of 
these code differences may pe very important in studying literacy. 


There has been increasing attention recently by anthropologists 
to the formal description of rules of speech in terms of social vari- 
ables such as rank, age, and sex of addressee, and occasion, which 
condition alternation in speech (for references, see Gumperz, 196523 
Hymes, 1964). There has,e1so been a characterization of certain 
structural relations or sequences in speech events such as greetings, 
thanks, conversation, introduction, and narration. To qualify as 4 
native speaker in a speech commmity one must learn these rules as well 
as the rules of grammar. This is to say, of course, that one must 
jearn to behave as though one knew the rules. Assessment of the ac- 
quisition of this competence should be as fundamental to the study of 
language development as the study of the formal grammatical code. 


alana braaihteiinaaiwes aes etme 6 


ner 


The study of sociolinguistic rules is in its infancy, so we are 
thrown back on ethnographic notes and queries rather than on precise 
hypotheses and methods as devices for directing field workers. One of 
our goals in this research is to obtain sufficient ethnographic data 
to provide a basis for comparative discussion among the field workers. 
This portion of the Field Manual must constitute early steps towards 
a sociolinguistic theory. 


There has been considerable controversy over the role of the so- 
cial milieu in language development (Lenneberg, 1967). The controversy 


turns on the degree to which a strong maturational component in language 


development may make linguistic competence relatively impervious to ac- 
celeration through changes in milieu. The linguistic repertoire of the 
community clearly must influence the content of the child's learning, 
but it is not clear whether the order of acquisition of granmatical or 
sociolinguistic rules might be altered either by their place in the 
adult system or by specific values and practices of the child’s care~- 
takers. These practices include prohibiting or encouraging the child's 
presence when various speech events occur, and stimulating or respond- 
ing to his speech. 


In advance, it is not clear how one would devise universal measures 
of rate of development, in order to test hypotheses about the relation 
of milieu and language development. However, if we are able to find 
situations in which natural variations exist within communities, as in 
the United States, one can provide a rough assessment of such effects. 
We regard experimental manipulation (Cazden, 1965) as desirable when 
some of the possible features of influence have been identified. But 
some kinds of manipulation are ethically impossible, thoug: they are 
found often in nature. For example, variations in family structure can 
easily be studied, including presence and absence of father or mother, 
presence of extended family acting as socializers, peer groups who are 
kin or primarily non-kin. Labov's work (1965) suggests that after five 
the peer group may be the primary source of language socialization. 


An example of social effects on language development can be found 
in the studies of sex differences in the rate of language development. 
Puerto Rican language development in general is faster than in matched 
English-speaking groups in New York, and no sex differences appeared 
(Anastasi & de Jestis, 1953) though girls were superior in the white 
sample, and boys in a Negro sample (Anastasi & D'Angelo, 1952) in New 
York; in Detroit (Thomas, 1962) also in a Negro lower-class sample, 
boys were extremely handicapped. It seems quite evident that the 
causes for these differential developmental rates are social, but we 
have too little information about relevant features of the social en- 
vironment to draw any conclusions. 


The first phase of this research is inevitably descriptive. There 
has never been a systematic study of language socialization, including 
the milieu and sociolinguistic competence. Ethnographies usually lack 
much information about family interaction, especially in urban settings, 
because it occurs relatively privately. As we indicated earlier, some 


xi 


NN  eeeeEeEeEeEeeeeeEeEeEeEeEeEeEeEeEeEeEeEEeEeEeEeeEeEeEeEeEeEeEc—eee 


studies comparing language development rates in different groups in the 4 
United States cannot be interpreted because we lack information about 
their language socialization practices. 7 


The material on language development in the narrow sense, which 
will include considerable text collection from a small group of children, ~~ 
will permit cross-linguistic comparisons of the development of struc- — | 
tured systems. In these cases, differences may derive from differences - 
in linguistic structure. 


The outcome of the initial fieldwork laid forth in this manual will * 
allow for: (1) generalization about development of given linguistic 
features in given languages, on the basis of a very small set of tests oe 
and a sufficient sample to support preliminary generalization; (2) dis- 
confirmation of assumptions about universals in development or milieu, 


and suggestions of new hypotheses about possible universals, either ed: 
descriptive or structural, hitherto unnoticed; (3) data for building | 
a sociolinguistic theory; and (4) further work in the same and carefully “ 


selected comparable field sites. Such further work will be devoted to 
the collection of larger samples for descriptive purposes, and will ex- 
plore antecedent-consequent relations in cultures where internal vari- 2 
ations in antecedent (milieu) variables are found. For example, if we 
are to study the learning of sociolinguistic rules in any detail, we 
need more work in each culture on characterizing the relevant rules. 

In the first phase of the work, outlined here, we place a different 
fieldworker at each site. In later stages, detailed cross-cultural 
testing of particular hypotheses might require collaboration between a 
traveling specialist and teams of local fieldworkers at established 
and well-studied sites. 


This field manual is the joint product of the authors, and is in- 
tended to provide a guide for them as they do their first fieldwork on 
these problems.* It is also intended as a guide for others who may 
wish to contribute to our small store of knowedge on these problems. 
In picking new field sites for the use of this manual, an important 
consideration is the availability of an adequate linguistic descrip- 
tion of the local language. In terms of training and site selection 
it would be difficult to improve on the advice offered by Margaret 
Mead over 35 years to those who would study "the primitive child": 


The student should have training in both anthropo..ogy 

and psychology, with an emphasis in psychology upon 

social, child, and abnormal psychology. Experience 

in psychiatric case work, especially among children, 

or in teaching, will prove more valuable than a 

knowledge of laboratory methods and equipment which 

it is almost impossible to use in the field. The 

anthropological training should emphasize particularly S| 
work in social organization--analyses of kinship 


+ 

Major writers of sections of the Manual are indicated on the first ~ 
page of each section. The entire Manual was extensively edited by 
Dan I. Slobin. 


xii 


systems, a sufficient knowledge of primitive religion to 
familiarize the student with the alien concepts which he 
will encounter, and a good training in linguistics. (‘The 
ability to learn language quickly and a high memory span 
for nonsense material are perhaps the two most necessary 
native abilities for this type of work.) A detailed 
kmowledge of the ethnography of the area chosen is also 
necessary. A student whose training is primarily in 
psychology [or, we might add, in linguistics] would be 
wise to choose an area which has been well described 
ethnologically . . . The choice of an area which has al- 
ready been described has also the advantage that the con- 
ditions will be known, the problem can be phrased in ad- 
vance in terms of the culture. _ 


For those, however, with a broader ethnological train- 
ing and interest, other cautions hold. . . . the final choice 
of a working location must be made from nearby, and be based 
upon exact information concerning the size of the commnity 
and the residential arrangements. For example, 4 detailed 
study of children could not be made in Dobu, where each 
village comprises only ten or fifteen people, and the members 
of each village are hostile to the members of other villages. 
Cultural attitudes towards children should be ascertained, if 
possible. For most purposes, it is better to work in a region 
where children are regarded as the most natural interest for 
adults than in a place where a man of rank is never supposed 
to speak to a child. A male investigator should try to find a 
region where the fathers play en important domestic rule. A 
woman investigator should attempt to discover the native atti- 
tudes toward barrenness and marital status in relation to their 
power of bringing good or bad luck to children, and then dis- 
close herself as single, widowed, divorced, the mother of twins 
left with their grandmother, or whatever fable is most likely 
to give the natives confidence in her (1931, pp. 685-6). 


We urge anybody planning to take the Field Manual to a field site 
to choose a location where there are current research field stations. 
Te number of such field stations is increasing, and information about 


_ them can be obtained from the Whitings at Harvard or from other anthro- 


pologists doing comparative work. The advantages of working at an estab- 
lished site are that initial work on censuses and basic ethnography will 
pe available, and local well-trained collaborators will be at hand. Work 
along the lines of this manual could also be done fruitfully within the 
highly varied speech communities in western cultures, for example, com- 
munities of immigrants, the deaf, American Indians, and so on. 


Before going to actual field sites, it is a good idea for all who 
intend to use this manual to try out all of the methods they plan to 
use with appropriate families nearby, preferably sufficiently distinct 
in culture and language as to provide a fair anticipation of later 


xiii 


ae 


field problems. Any comparative work must require some procedural 
standardization, and it is highly necessary to have prior communication 
about difficulties before fieldwork begins. Conditions in the field 
interfere with standardization in several ways. In terms of physical 
and social setting, the investigator is increasingly removed from the 
frame of reference in which he first discussed these problems; there- 
fore investigators become more unlike each other in the field setting. 
Communication with others whose data must be the basis for comparative 
reports is much more aifficult for a variety of reasons. The severe 
demands of settling into a new community and establishing rapport make 
it very likely that investigators will surrender all efforts at a com- 
mon set of aims and procedures unless very explicit prior commitments 
are made. These are likely to be especially severe problems for 
anthropologists wnose relativistic ideology may make it seem & positive 
good to depart from the common frame of reference established prior to 
departure for the field. The Field Manual is meant to be a shared com- 
mitment restraining these forces.” Pilot studies within the home 
country can permit the investigators to cope with some of these strains 
while still in the work group. For this reason we recommend that any 
investigators not in our initial group write to us about any ambigui- 
ties they find doing pilot work from the Manual, in order to make our 
studies as comparable as possible. Comments about the Manual, re- 
quests for copies and for tapes to be used in the phonological imitation 
procedures, and information about related studies should be addressed to 
Dr. Dan I. Slobin, Department of Psychology, University of California, 
Berkeley 94720. 


Throughout our work on the Manual, in 1966 and 1967, we have been 
given a home in the Institute of Human Learning of the University of 
California at Berkeley, to which we express our gratitude. Since the 
first draft of the Manual was prepared, funds from Contract 4-7-008757- 
2015 with the U.S. Office of Education have enabled us to try out 
procedures locally. Through Project Literacy, the Office of Education 
also indirectly financed a series of consultative meetings about the 
problems raised by the draft manual. Many generous colleagues have 
given freely of their time and counsel, either through visits or letters 
--or both. These include Harriet Amster, Ursula Bellugi, Noam Chomsky, 
Aaron Cicourel, A. Richard Diebold, Jr., Joe Dillard, Charles A. 
Ferguson, Robert Hess, Dell Hymes, Vera John, Edward S. Klima, Eric H. 
Lenneberg, William Labov, David McNeill, Wick Miller, Michael Moerman, 
Harry Osser, Harvey Sacks, Emmanuel Shegeloff, Elizabeth F. Shipley, 
Tatiana Slama-Cazacu, Carlota S. Smith, William Soskin, William Stewart, 


as All fieldworkers are urged to supplement this manual with the 
Field guide for a study of socialization (Whiting et al., 1966), used 
by the Harvard study of child rearing in six cultures. That manual 
contains many valuable suggestions on the study of socialization and 5 
fieldwork in general, and is referred to at various places in our 
manual. Another useful aid is W. J. Samarin's Field linguistics 
(1967), to which we also refer the reader for additional advice. 


xiv 


and Beatrice Whiting. As a result of their advice we made changes in 
the Manual, and learned from them much more which may influence the 
work of those present at our meetings. We wish we could specify more 
precisely each of the great variety of valuable contributions we have 
received. (Responsibility for the final contents of the Manual, of 
course, lies with us.) 


The Social Science Research Council, through the Committee on 
Sociolinguistics, has helped finance the fieldwork in this study with 
funds from the National Science Foundation. Recipients of such aid 
have been Jan Brukman, in a Koya-speaking village in India, and 
Claudia Mitchell, in a working-class Negro section of Oakland, Cali- 
fornia. In 1967-68 Carolyn Wardrip will study middle-class white 
families in Berkeley, and David Argoff will compare Finnish and 
Russian-speaking families in Finland and the Soviet Union. In addi- 
tion, Brian Stross in Chiapas, Mexico, and Keith Kernan in Samoa will 
carry out field manual studies under separate sponsorship. Clerical 
assistance is being provided by Rachel Holmen. The group will re- 
assemble in the summer of 1968, with others who have done similar re- 
search, to prepare a third revision of the manual for publication, and 
to work on problems of data analysis. We are grateful for the extensive 
support which has been available in these initial phases of our compara- 
tive research. We view this project, though it began in Berkeley, as an 
effort at national and international collaboration, and welcome others 
who wish to join in the various phases of our work. 


Coane Prresasssress My 
es . mee 


PART ONE ‘ 


MAJOR TOPICS OF INVESTIGATION 


1.0 RESEARCH SCHEDULE 


There are fourteen major areas of investigation which constitute 
the core of this field manual. All researchers are expected to collect 
comparable data on these central topics, following the schedule of 
activities presented below as closely as possible. In addition to the 
major topics, the researcher is encouraged to refer frequently to the 
"notes and queries" section presented at the end of Part One (1.7), and 
to be constantly alert to the theoretical considerations and possible 
subsidiary studies presented in Part Three. A number of important 
methodological factors are spelled out in Fart Two, and the Appendices 
contain examples of specific tests and procedures. The investigator 
should read through the manual at least once a month during his stay in 
the field. 


A core sample of 24 children, between the ages of six months and 
13 years, will serve as subjects and informants for the central investi- 
gation, along with a number of adults in the community, including the 
mothers of some of the children. Each of the major topics is taken up 
in detail, following an initial outline of the schedule of research 
activities to be undertaken in the field. This schedule presupposes @ 
twelve-month field visit, leaving the first month free for orientation, 
and the last month to make up for inevitable slippages in the schedule, 
transcription of materials gathered at the end of the stay, etc. 


At the time of beginning formal research (the second month in our 
schedule) the investigator should have some knowledge of the language, 
should have selected a native assistant (see Section 2.3), and should 
have designed the research materials for topics IV, VII, VIII, and IX. 
The schedule is arranged so that the first contacts with child subjects 
are with older children (3;4 - 7 years), in order to facilitate entry 
into the world of child speech. Table 1 presents the list of research 
tasks; Table 2 the distribution of children across topics; and Table 3 
the 12-month time schedule. 


Section 1.0 was written by Dan I. Slobin. 


1 


ee | 


* 
Table 1. Major Research Tasks 


I. Determine level of motor development [1.1]. 
II. Record babbling [1.2.3]. 


III. Elicit and record imitation of prosodic aspects of speech 
according to standardized tests [1.2.2]. 


crete NE GARR ERED TER LIL SOLD SOL LIL BOT 


Iv. Elicit and record imitation of phonological aspects of speech 
according to standardized tests [1.2.2, 1.3.1]. 


VY. Elicit and list initial vocabulary [1.2.4, 1.3.1]. 


VI. Record free interaction between child and mother (or mother- _ = 
surrogate) for intensive grammatical analysis [1.4.1]. 


VII. Elicit and record imitation of systematically constructed 
model sentences [1.4.2]. 


VIII. Administer standard speech comprehension tests [1.4.3]. 


IX. Elicit and record production of grammatical speech according 
to standardized tests [1.4.4]. 


X. Describe child's interaction networks and standard daily 
activities, over the course of a year, on the basis of 
observation and interview of child and parents [1.5.1]. 

XI. Record and observe dialogs between children [1.5.2). 
XII. Record speech of mother to child [1.5.3]. 
XIII. Observe communicative routines [1.5.4]. 


XIV. Interview adults concerning beliefs about language acquisition 
and use [1.6]. 


7 Figures in square brackets refer to descriptions of data-gathering 
techniques to be found in Part One of the Field Manual. 


In Table 2 each child receives a letter designation, which serves to 
identify him in the subsequent text. In noting ages, a semicolon separates 
years from months; a conma should be used to separate months from days: 
e.g. a child of age 23;10,13 is two years, ten months, and thirteen days 
old. The investigator should attempt to find children as close to the 
ages given as possible; the tolerance range should be about one month 
(in those societies where exact ages of children are known). Column XIV 


refers to the child's mother. Research months are counted from arrival 
in the fiela. (If the field stay is longer than 12 months, the re- 
searcher may wish to begin the formal schedule later than the second 
month; however, at least one free month should be left at the end for 
unforeseen circumstances. ) 


Table 2. Distribution of Children by Age, Sex, and Task 


~~ First 
Child|Age |Sex|I}IZ| IIL) IV] V| VI} VIL|VIT| X|x XI} XII|XIII|XIV|Research 
Month 
: A 1036] M [+) +) + I+ + + + }+ 3 
B 1036) F I+} +) + I+ + +) +i] + 4 
Cc *% | M I+ + i+ }+ + + + | + 6 
D * | F I+ + I+ i+ + + +] + 6 
E x | # [+ + +] + | + | th+ + + | + 3 
q F % | & I+ + +] +] +i+ + +/+ h 
G |2;0] M {+ + +] +] Fi+] + + 3 
: H {23;0| F {+ + +] + | +]+] + + 3 
| I |2;8] M i+ + +] i +i+] + + 3 
| J |2;8] F j+ + + | + | +i+] + + 3 
K 1334) M {+ + + | + | t]+] + + 2 
L |334) F I+ + +1 +) +l+l + + 2 
: M {4321 M j+ + + {| +] +i+] + ~ 2 
N |43;2] F I+ + + {+ | ++} + + | + 2 
0 {5 M +] + + 2 
P |5 | F +| + + 2 
| Q I7 M +] +1 + + | + 2 
| | R 17 |F +) +{ +] +] + 2 
. s 19 M +| + + 3 
| tT 19 | F +) + + 3 
u jn iM +) + + 3 
YY tae ig +] + + 3 
a w (13 | M +| + + 3 
X 13 | F +1 + * 3 


4 Note: The age of children C and D is indeterminate; they should be 
c} early in the one-word stage, and have a limited vocabulary. Children E 
and F are not picked on the basis of age, but mean-utterance length, as 
described in 1.4.1; their sex is not important, but they should both be 
of the same sex. 


For each child studied, the investigator should prepare a dossier 
containing such information as the following: 


name “ 

birthdate (or closest approximation) 

sex 

handedness 

general physical description (accompanied by photographs ) 
general mood description 

any relevant pathology or abnormality 

level of motor development (see Table 4) 

articulateness and quality of speech and voice 


family situation | 
parents living? | p> ts 
parents’ ages 
number, ages, and sexes of siblings 
people living in household and their relation to child 
relevant kin living elsewhere 
status and occupation of family and its members 
description, photographs, and map of dwelling 


* tn some field situations the investigator may not be able to as- 
certain the exact date of birth of a child, or even its general age. 
In such cases, try to determine the time of the child's birth in re- 
gard to a season, a festival, a special event, an historical occasion, 
or the like. It may also be helpful to ascertain the child's age 
relative to other children for whom more precise information is avail- 
able. Some guides to estimation of children's ages can be found in 
Bayler and Bayley (1959), Krogman (1941, 1958), and Watson and Lowrey 
(1951, esp. pp. 44-76). Bayler and Bayley present height and weight 
curves and graphic growth records for boys and girls (charts available 
from Medical Dept., Ross Laboratories, Columbus 16, Ohio). However, 
it should be borne in mind that such statistics vary with race and 
environment. Krogman (1958, pp. 52-67) lists many references on environ- 
mental, racial, and dietetic influences upon physical growth; and 
Krogman (1941) gives many tables on season, diet, health (pp. 704-801), 
and socioeconomic status and rural-urban contrast (pp. 802-857) in 
relation to growth measures. Behavioral indices of age are probably 
more useful cross-culturally than are morphological indices. 


x 
A more detailed outline of this part of the dossier is given in 
1.5.1. 


Table 3. 12-Month Time Schedule 


| Child 
Nonth|A BC DE F GHig«kKoL-MNOPQRS TUV WX 


1 
2 KLM No P @R 
3 la” By GHIdJ sTtTuVW xX. 
¥ * 
h B F K LM N 
* 
E 
*% 
5 Ia GHIJ 
% 
6 Bc*p*E K LM N 
¥* * 
7 IA FGHId 
* 
E 
8 RY K LM N 
* 
9 cDE GHIdJ 
: * 
10 F OP QR 
} * 
{ e 
~ 11 GHIdJ s TUVW X 
j 12 
| * wotes: Asterisk after child's initial in a given month means that mother 


is also to be studied in that month--either to record her speech to the child 
a or to interview her concerning beliefs about language acquisition and use 
XIV), as shown in Table 2. Horizontal lines indicate months; because E is to 
be visited every 6 weeks, he frequently falls on the line between months. 


1.1. MOTOR DEVELOPMENT 


There is good reason to believe that, in normal children, speech 
and motor development follow temporally related maturational sequences. 
This thesis is most clearly developed by Eric Lenneberg, in his recent 


pook, Biological foundations of language (1967): 


The onset of speech consists of a gradual unfolding of 
capacities; it is a series of generally well-circumscribed 
events which take place between the second and third year of 
life. Certain important speech milestones are reached in a 
fixed sequence and at a relatively constant chronological age. 
Just as impressive as the age constancy is the remarkable 
synchronization of speech milestones with motor -developmental 
milestones . .. (p. 127). 


In Table 4, below, we reproduce Lenneberg's Table 4.1 (pp. 128-130), 
in which this “temporal interlocking of speech milestones and motor mile- 
stones" is revealed. It is evident that cross-cultural research is needed 
to validate Lenneberg's assertion that this developmental sequence is 
universal. : 


The categories presented under "Vocalization and Language" in the 
table are clear. The investigator should be certain that his sample 
includes children demonstrating the forms of vocal and verbal behavior 
summarized there. With the exception of the very youngest infants, 
children A-N should satisfy the requirements of these categories, and 
they are marked appropriately under column I in Table 2. Neonates 
should be found for the 12-, -, and 20-week-old entries in Lenneberg's 
table. The descriptions of stages of motor development in that table 
are likewise quite clear, and the investigator should have little diffi- 
culty designing simple tests to determine the motor milestones described 
by Lenneberg. (Watson and Lowrey [1951, pp. 87-99] also present norms of 
behavioral development, including motor, adaptive, language, and personal- 
social milestones. ) 


Table 4. Developmental Milestones in Motor and 
Language Development (from Lenneberg, 1967 ) 


At the con- 7 
pletion of: Motor Development Vocalization and Language , 


12 weeks Supports head when in prone Markedly less crying than at 8 
position; weight is on ei- weeks; when talked to and nodded 


bows; hands mostly open; no at, smiles, followed by squealing- j 
grasp reflex. gurgling sounds usually called 
cooing, which is vowel-like in ). 
character and pitch-modulated; | 


sustains cooing for 15-20 seconds. 


Section 1.1 was written by Dan I. Slobin. 


Table 4. (cont.) 


Motor Development 


Vocalization and Language 


At the com- 
pletion of: 
| 16 weeks 
] 20 weeks 


a 6 months 


8 months 


10 months 


Plays with a rattle placed 
in his hands (by shaking 
it and staring at it), 
head self-supported; tonic 
neck reflex subsiding. 


Sits with props. 


Sitting: bends forward 
and uses hands for sup- 


port; can bear weight 


when put into standing 
position, but cannot 

yet stand with holding 
on; reaching: unilateral; 
grasp: no thumb apposi- 
tion yet; releases cube 
when given another. 


Stands holding on; grasps 
with thumb apposition; 
picks up pellet with thumb 
and finger tips. 


Creeps efficiently; takes 
side-steps, holding on; 
pulls to standing posi- 
tion. 


Responds to human sounds 
more definitely; turns head; 
eyes seem to search for 
speaker; occasionally some 
chuckling sounds. 


The vowel-like cooing sounds 
begin to be interspersed 
with more consonantal sounds ; 
labial fricatives, spirants 
and nasals ere common; 
acoustically, 211 vocaliza- 
tions are very different 
from the sounds of the 
mature language of the en- 
vironment. 


Cooing changing into 
babbling resembling one- 
syllable utterances ; 

neither yowels nor conso~ 
nants have very fixed recur- 
rences; most common utter- 
ances sound somewhat like ma, 
mu, da, or di. 


Reduplication (or more con- 
tinuous repetitions) becomes 
frequent; intonation pat- 
terns become distinct; ut- 
terances can signal emphasis 
and emotions. 


Vocalizations are mixed with 
sound-play such as gurgling 
or bubble-blowing; appears 

to wish to imitate sounds, 
but the imitations are never 
quite successful; beginning 
to differentiate between 
words heard by making differ- 
ential adjustment. 


Table 4. (cont.) 


At the can- 
pletion of: 


Motor Development 


Vocalization and Language 


12 months 


18 months 


24 months 


Walks when held by one 
hand; walks on feet and 
hands--knees in air; 
mouthing of objects al- 
most stopped; seats self 
on floor. 


Grasp, prehension and re- 
lease fully developed; gait 
stiff, propulsive and pre- 
cipitated; sits on child's 
chair with only fair aim; 
creeps downstairs backward; 
has difficulty building 
tower of three cubes. 


Runs, but falls in sudden 
turns; can quickly alter- 
nate between sitting and 
stance; walks stairs up 
or down, one foot forward 


only. 


Identical sound sequences are 
replicated with higher rela- 
tive frequency of occurrence 
and words (mamma or dadda) are 
emerging; definite signs of 
understanding some words and 
simple commands (show me your 


eyes ). 


Has a definite repertoire of 
words--more than three, but 
less than fifty; still much 
babbling but now of several 
syllables with intricate into- 
nation pattern; no attempt at 
communicating information and 
no frustration for not being 
understood; words may include 
items such as thank you or 
come here, but there is little 
ability to join any of the 
lexical items into spontaneous 
two-item phrases; understand- 


ing is progressing rapidly. 


Vocabulary of more than 50 
items (some children seem to 
be able to name everything in 
environment); begins sponta- 
neously to join vocabulary 
items into two-word phrases; 
all phrases appear to be own 
creations; definite increase 
in communicative behavior and 
interest in language. 


Table 4. 


(cont. ) 


At the con- 
pletion of: 


Motor Development 


Vocalization and Language 


30 months Jumps into air with both Fastest increase in vocabulary 
feet; stands on one foot with many new acditions every 
for about two seconds; day; no babbling at all; utter- 
takes few steps on tip- ances have communicative in- 
tee; jumps from chair; tent; frustrated if not under- 
good hand and finger co- stood by adults; utterances 
ordination; can move consist of at least two words, 
digits independently; many have three or even five 
manipulation of objects words; sentences and phrases 
much improved; builds have characteristic child 
tower of six cubes. grammar, that is, they are 

rarely verbatim repetitions of 
an adult utterance; intelli- 
gibility is not very good yet, 
though there is great varia- 
tion among children; seems to 
understand everything that is 
said to him. 

3 years Tiptoes three yards; runs Vocabulary of some 1000 words ; 
smoothly with accelera- about 80% of utterances are 
tion and deceleration; intelligible even to strangers; 
negotiates sharp and fast grammatical complexity of ut- 
curves without difficulty; terences is roughly that of 
walks stairs by alternat- colloquial adult language, al- 
ing feet; jumps 12 inches; though mistakes still occur. 
can operate tricycle. 

k years Jumps over rope; hops on Language is well-established; 


right foot; catches ball 
in arms; walks line. 


deviations from the adult norm 
tend to be mere in style than 


in grammar. 


Note that chronological age is really not crucial to the comparisons 
required by Teble 4. What is important is to determine whether motor and 
speech milestones interlock in the manner presented above. A chart should 
be prepared for each vhild, indicating his age as closely as possible, 
evidence of his level of motor development, and a brief description of the 
main features of his vocal and verbal behavior. 


Lenneberg has done preliminary cross-cultural study of the questions 
raised here, and the fieldworker will be aided by his observations: 


How universal in human society is the onset of speech? 
Do cultural attitudes toward child-rearing influence the 
emergence of speech, or are there languages that are either 
so complicated that no one can master oral communication until 
puberty, or so primitive that the entire system is learned by 
every child before he begins to toddle? . . . Strangely enough, 
the onset of speech has rarely been & subject of a detailed 


the vocalizations or communicative behavior among the children 
of "primitive" and "western" man. I have investigated this 
problem further with the cooperation of several students in an- 
thropology (Samuel Putnam, Mary Ann Whelan, and Eleanor Crocker ) 
who made direct field observations among the Dani of Dutch New 
Guinea, the Zuni of the American Southwest, and the Bororo of 
Central Brazil. : 


In these investigations, children were given tests of 
sensory-motor development such as coordination for reaching, 
nature of the grasp, and the ability to walk, stand on one leg, 
or throw a ball. Tape recordings were made of the vocalizations 
of the babies before they appeared to possess the common lan- 
guage, as well as of their utterances throughout their physical 
development to age three. In addition, information was obtained 
from native informants about the linguistic competence of the 
various children studied, their fluency, types of mistakes in 
articulation, syntax, and choice of words, and information was 
gathered on the parents’ attitude toward their children's speech 
development. In some instances, the chronological age of the 
child studied was not known, and therefore neither the motor nor 
the language achievements could be compared directly with the 
age of emergence in American children. But the chronological 
age was not as important as the question of whether develop- 
mental progress, gauged upon the emergence of definite motor 
skills, marked also the beginning and major milestones of the 
child's speech development, and whether the concordance be- 
tween speech and motor development observed in western children 
were also found in children of these cultures. As far as we 
can judge from the analysis of the material, the answer is 
clearly yes. The first words appear at about the time that 
walking is accomplished, and by the time a child is able to 
jump down from a chair (or its equivalent), tiptoe, or walk 
backward three yards, he is reported by the informants to be 
communicating fluently, even though certain inaccuracies and 
childlike usages seem always to persist for a longer period. 
Anthropologists have pointed out that the label "primitive 

" 4s misleading when applied to any natural living 
language. The developmental studies here support the view that 
no natural language is inherently more complicated or simpler 
to learn by a growing child than any other language. There 
seems to be no reletion between progress in language acquisi- 
= na culturally determined aspects of language (1967, pp- 

-9). 


It is our hope that users of this Field Manual will gather fur- 


ther evidence in regard to the provocative thesis developed by 
Lenneberg. 


1.2. LINGUISTIC DEVELOPMENT: PHONOLOGY 


| 1.2.1. Definitions 


It is common usage in modern American research to refer to three 
stages of development of infant sounds. Cooing refers to a particular 
} type of sound which is frequent in the firs year. It is distin- 
, guished from crying, from fussing, and from organic sounds like belch- 
ing. For definitions of the other categories, see Lenneberg, Rebelsky, 
and Nichols (1965). Cooing sounds are primarily vocalic. If consonan- 
tal sounds are heard at this stage, they most often are back consonants 
or continuants, such as nasals. Cooing has 4 slow onset and therefore 
( gives the impression of leisurely, pleasurable sound. Experimental 
evidence suggests that by three months of age cooing can be increased 
in frequency by social reinforcement. At any event it steadily in- 
creases, while crying decreases, in observational studies. 


Babbling tends to replace cooing, sometimes as early as three 
months, but more usuelly after six months. Babbling refers to vocal- 
izing which gives the impression of syllabic repetition. There are 
repetitions of brief sounds, rather than long gradually changing vowels, 
and there is closure of the vocal apparatus, with much more precise use 
of the articulators, such as tongue and lips. While the babbling may 

sound to a casual listener a lot like adult syllables, ©.g- { dadada] 
= or [bababa], an observer may be surprised to see different articulation 
- from adults, e.g. a tongue protruding between the lips. 


Speech begins when there is a consistent meaning for the child's 
utterances other than a broad affective category, though the phonetic 
variation for a morpheme may be wide at first. The report of the 
child's caretakers is probably the best clue that the child has 
speech. There is sometimes a4 relatively silent period between 
babbling and the onset of speech, and the child with speech may have 
simpler and less frequent utterances than he did when babbling. 


Once speech appears, there are sequences which cannot be glossed, 
which one can call gibberish or nonsense. It is not known whether 
nonsense employs a wider repertoire of sounds than meaningful speech, 
and therefore is more like babbling in form. 


1.2.2. Imitation 


Imitation data permit the writing of precise input-output rules 
and therefore may allow rapid collection of much more complete infor- 
mation on @ child's system than would be obtained from spontaneous 
speech alone. The rule structure my be slightly different than the 
system for perception or production alone, of course. 


Children too young to instruct to imitate frequently do so spon- 
taneously, or can be trained to do so. Training consists of rewarding 


Section 1.2 was written by Susan M. Ervin-Tripp. 


any spontaneous instances which seem close to imitation, and gradually 
"shaping" better imitation until the child responds imitatively with 
fair frequency. As in other procedures, one may note certain times of 
day or special circumstances when the child seems more ready to imi- 
tate. Wait until the child is freely imitating live voices before 
setting up equipment for imitation of recorded stimuli. One must thus 
be prepared to bring in the imitation procedure, or to end it, accord~- 
ing to the child's oscillating whims. 


|For children C-N, the tape recorder itself will be a distraction 
and will orient the child away from the voice and toward the machinery. 
Give the child experience in listening to a preparation tape first. 
This includes music, odd sounds, and voices speaking nonsense. When 
the noveltywears off, the child may be ready to attend to tapes de- 
signed to test his imitations. 


As a practical matter, we have prepared a set of stimulus tapes. 
The use of the same tapes in all field sites will allow comparisons 
of results. We can include a wider range of material than would be 
possible if we were confined to oral stimuli provided by the field- 
workers and their,assistants. The stimulus tapes can be purchased 
from the authors. These tapes consist of a set of three-inch reels 
with contents as follows: 


Tape contents Child informants 

A-B C-D-E F-J K-N 
) Preparation x x x 
| Vowels and prosody x x 
) Single consonants x x x 
) Disyllabic patterns x x 

Clusters x x x 

Trisyllabic patterns x x 


The preparation tape is used to accustom the child to listening 
to the recorder rather than handling it. The prosody section con- 
tains intonation patterns and other suprasegmental variations. The 
vowel section contrasts high and low, front and back, rounded and un- 
rounded vowels. The consonant and cluster tapes provide examples of 
highly frequent initial consonants or clusters, and the multisyllabic 
patterns employ simple components but combined in complex ways to 
test assimilation. 


Procedure 


Continue training and preparation until the child seems ready to 
respond. The stimulus tapes are run on a second tape recorder, if 
necessary on a small cartridge machine. The usual machine is employed 
for recording the child's responses. In order to conserve response 
tape, start and stop the response machine only when the child is 
producing by the criterion for his age, as indicated below. 


. Available from Dr. Dan I. Slobin, Department of Psychology, 
University of California, Berkeley 94720. 


ad 


response. For C-N, there must be evidence that the vocalization is 4 
result of orientation to the stimulus. First, keep distracting people 
and objects away in any testing session. Use the preparation tape till 


talk about. ‘The child's caretaker can tell you if the child is talking 
about something else, and not paying attention to the sounds. If re- 
sponses are present bearing & rough similarity to the input to con- 
vince you the child is listening, then the responses are acceptable. 


Sampling 


stimilus pattern. If the child responds several times to a pattern, 
so much the better. These repetitions will allow some assessment of 
the stability of his substitution patterns. On the chart, the chil- 
dren are given between two and four input tapes in all, so sam chil- 
dren must be tested more than others. Since no transcription is 
necessary, the field time consists only of the recording. During the 
recording, keep some written notes to help distinguish pairs that 
sound alike on tape like p-k. 


1.2.3. Babb 


According to several theories, babbling should gradually become 
aifferentiated in different linguistic milieux. This prediction has 
never been tested, though some current work in spectrographic analysis 
of babbling tapes from different milieux is being carried out. In 
order to test this hypothesis, one needs samples of speech addressed 
to the infants, as well as samples of their pabbling. The speech 
They hear establishes a baseline for judging the differences in input 
to the children. Such data will be collected in the studies of baby 
talk (1.5.3). Since it will be used for acoustic analysis, care must 
be taken to avoid echoes and background noise. A Lavalier microphone 
on the neck of the caretaker, on the excuse it is used for listening 
to the child, might give best results. 


Babbling samples should be collected in standard situations: 
half just after feeding, and half while handling objects. To maxi- 
mize the variety of sounds collected, try to accumulate a total of 
about thirty minutes of pabbling in three minute samples. Probably 


stimulation by others. Avoid background noises where possible, even 
by bringing the infant into a different setting, unless vocalizing 
seems to be reduced then. There will be three such tapes of children 
A and B at bi-monthly intervals. If the recording of the babbling 
occurs in the midst of other taping, & stopwatch might be used, with 
a "time in action” attachment, to keep & running record of how many 
pabbling minutes have been collected. Note that these tapes require 
no processing in the field, since they will be used for laboratory 
analysis at home. ode 


14 


1.2.4. Speech Samples 


The speech samples collected for the st of vocabulary in chil- 
dren C and D (1.3.1) and syntax in E and F (1.4.1) should be of such 
quality as to allow later phonological analysis. For this to be the 
case, two criteria are necessary, good quality acoustically and suf- 


ficient range of sampling. 


It is of particular importance for these tapes to be of good 
acoustic quality since it will be more important to hear phonological 
details than in the tapes of the oldest children. For purposes of 
the phonological analysis, the mturalness of the recording situation 
is less important for a young child than the lack of noise in the 
background, and the lack of metallic or other polished surfaces nearby 
for the reflection of the high pitched voice of the child. If pos- 
sible, keep the microphone near the child but out of his hands. 

Speak softly yourself so that the recording level can be kept high. 


Ideally, in a phonological study one would keep an ongoing 
analysis to make sure the data are sufficient at a particular point 
in development. This is not possible in this study, so that instead 
one should aim for maximum variety in the tape. Try to tape in situa- 
tions where the child is most fluent, and when he talks about enough 
different things to get a varied phonological sample. Ideally one 
would also get some repetitions, separated enough so that some of the 
range of free variation for particular morphemes could be sampled. 


Transcribing is facilitated in several ways. Try to keep a 
record, if you can, of the child's speech at the early stages in 
writing. Some consonant sounds are hard to discriminate from tapes, 
and such @ written notation could aid. Also, it will help you notice 
when the child spoke inaudibly so that if the item is important you 
can elicit it again. It would help greatly if the child's caretaker 
made a running translation of the child's utterances immediately after 
they occur. This is not too difficult at the early stages. You can 
hear Jater on the tape which are expansions and which accurate 
repetitions. Also the caretaker's version will reveal the overt norm 
for the phonemes, as the child is given them as a model. Notes on 
the ongoing activity are necessary, since one always underestimates 
the dependence on situational environment of young children's speech. 
If the child's caretaker is not available to help transcribe the 
tapes, at least the comments on the tape will be a partial substitute. 


We have found several mechanical aids to transcription that you 
may use once back from the field, for ambiguous passages. One is 
slowing down the tape to half speed. Another is creating a tape 
loop for ambiguous passages. Systematic testing of alternative inter- 
pretations is always necessary, since there is a strong tendency to 
assimilate what is heard to the norms of the hearer. 


(Additional advice on transcription is given in Appendix 8.3 


15 


1.3. LINGUISTIC DEVELOPMENT: VOCABULARY 


A large-scale study of vocabulary development is beyond the scope 
of the field project envisioned here. Limited study of vocabulary, 
however, can contribute to several questions of psycholinguistic 
theory. In the phonology section, above, reference has been made to 
vocabulary study in the process of revealing a child's phonemi 
system and his substitution rules. | 


1.3 ebe Initial Vocabulary 


A cross-cultural comparison of the initial vocabularies of chil- 
dren may be revealing in regard to questions of language functions 
and contexts of language acquisition. Accordingly, we have designated 


children C and D as subjects for the study of initial vocabulary. 


These showld be children early in the one-word stage, having a small 
vocabulary (4-40 words, as estimated by mother) at the beginning of 
the study. C is a boy and D a girl, and each should be visited at 
least twice, as indicated in Table 3. The researcher's task will be 
to attempt to elicit as much of the child's vocabulary as possible. 
(See suggestions in Section 1.2.4, above.) Initially, some time will 
have to be spent with the child and its mother--listening carefully, 
asking the mother what the child is saying and what words it knows, 
and so on. This task is scheduled for the sixth research month, as 
it will pose the researcher with especially difficult problems of 
speech perception and elicitation. 


We can only provide general research advice in regard to elicit- 
ing initial vocabularies. The task would be greatly facilitated if 
the researcher could find a child able to respond to requests for 
names (e.g. @ child who will name an object pointed to by an adult, 
respond to a question such as "What's this?", and so on). It may be 
easy to train some children to perform in such fashion. The child 
should then explore his familiar world with the researcher, with en- 
couragement to name objects. The aim of this exploration will be to 
reveal (1) the extent and range of vocabulary, (2) the semantic do- 
mains named by the child, (3) the "innominate" zones, (4) the nature 
of generalizations of meaning, and (5) the extent of homonymity. 


At the same time, the researcher should be alert to the means in 
which adults name objects for children (cf. Brown, 1958; and 3.4.3.3). 
If the mother is cooperative, it would be informative to present her 
with objects and ask her to name them for the child. Note if this is 
something which is habitually done, the sort of name offered by the 
mother (e.g. diminutive, class name, proper name, etc.), the mother's 
concern for securing the attention of the child and ensuring the re- 
tention of the word, and so on. Note also if requests for names are 
made by the child, and, if so, how they are responded to by adults. 


Section 1.3 was written by Dan I. Slobin and Susan M. Ervin-Tripp. 


a athe on 


16 


The following methodological suggestions pertain to the study of 
first words both from the point of view of semantics and phonology: 


1) Try to get to know the child so he is not silently inspecting 
you. Bracelets, keys, clock are good things to interest a child. On 
the other hand, toys which involve 2 lot of walking around or make too 
much noise themselves may interfere with the recording. 


you may be able to jdentify some vocabulary they have not noticed, 

just by noting regularities of occurrence. Whenever you hear 4 repeti- 
tion of sound correlated with circumstance, make 4 phonetic transcrip- 
tion of what you hear. Also note exactly what the child was looking 
at, pointing to, or handling. You may be able to get the word also by 
repeating it and getting an imitation, but make & separate notation 
when you have done this. 


3) Test the semantic range of the child's naming. What else 
will elicit the same word? Some children will react to questions , 
such as "What's this?" if done in the right way. Try having an older 
sibling or the mother do this, or see if they have advice about how to 
get the child to respond. How you test for semantic range will depend 
on your own ingenuity. Pictures, "play-doh" and other objects all may 
work. 


4) Set up a range of pictures and/or objects and see what the 
child will comprehend when you say, there's the X?" or, “Bring me the 
x." Ask the mother what the child is likely to respond to when ad- 
dressed in this way. Comprehension may be ahead of naming. Note that 
you can test your hunches about the phonemic system now by varying the 
word and seeing if it is still understood. 


5) In analyzing the phonemic system, note the canonical form of 
words. What are the free variants in each position? Do certain 
vowels occur with certain consonants only? What features are constant? 


6) To determine substitution rules, get the child to imitate you 
and see if you are able to predict his substitutions systematically, 
including the canonical form for words. For example, are polysyllabic 
words always imitated as CVCV; monosyllabic as cvc? If you repeat the 
imitation, does the child alter the word he says? If so, how? This 
gives a clue as to the unstable features of his system. 


1.3.2. Vocabulary Related to Grammatical Categories 


When working with older children (E-N), the researcher should 
attend to the acquisition of those lexical categories which have 
tical significance in the language. It has been found, for 
example, that lexical items referring to certain semantic categories 
appear in child speech at the same time as those categories become 
morphologically marked. A good example of this phenomenon is the 
finding of Gvozdev (1949) that, ina Russian-speaking child, the 
first use of the word mnogo [much, many} appeared at the same time 


17 


as the singular-plural distinction in noun markings, and that the 
words se [right away} and skoro [soon] entered at the same 
‘time as the future tense. The researcher would do well, therefore, 
to watch for the appearance of words referring to such categories as 
time, number, conditionality, causality, and the like, if the lan- 
guage also marks such categories by gramm”. al] means. 


q q » 


antago 


1.4. LINGUISTIC DEVELOPMENT: GRAMMAR 


1.4.1. Collection of Basic Corpora 


The most extensive recorded speech corpora to be collected in 
the study are those produced by children E and F. Great care should 
be exercised in selecting these children, as they will provide the 
basic data for as full a description of grammatical development in 
the given language as possible. The research aim is to make these 
data roughly comparable to those gathered by Roger Brown, Ursula 
Bellugi, and Colin Fraser in the United States (Bellugi, 1964, 1965, 
1967; Brown & Bellugi, 1964; Klima & Bellugi, 1966), and by David 
McNeil in Japan (McNeill, 1966a; McNeill & McNeill, 1966). ‘The rich 
grammatical analyses resulting from those studies are based on record- 
ing of spontaneous speech, in home settings, chiefly between mother 
end child. As far as possible, we wish to replicate that situation 
in our field studies. 


Children E and F are to be picked to correspond to the first and 
fourth stages in the study of Brown et al. Brown's longitudinal study 
covered several years in the life of each child; since our fieldworkers 
will probably be in the field for only one year, the best we can do is 
to select children at different stages of development, and hope to put 
together some sort of continuous picture--however hazardous--by juxta- 
posing the records of several children. Because of the very large 
amount of time required to transcribe spontaneous speech data, we 
recommend that the sample be limited to two children. Brown's defini- 
tion of stages is based on mean utterance length and upper bound 
(longest utterance )--in morphemes. His sample includes three chil- 
dren, for whom the first and fourth stages--the stages we wish to 
pick for children E and F, respectively--are defined as follows: 


Number of Morphemes 


Mean Utterance Upper Bound 
Length 
First Stage 1.75 5.0 
Fourth Stage 3.50 11.0 


Brown's sample at each stage consists of about 700 utter- 
ances, on the basis of which he is able to write systematic and 
thorough grammars for each of three children at five stages of de- 
‘velopment. According to Ursula Bellugi (pers. comm., 1966): 


The grammars include information on the appearance of 
grammatical markers and constructions like the following: 
lexical items, count nouns, mass nouns, proper nouns, 


Section 1.4 was written by Dan I. Slobin. 


19 


personal pronouns, ‘>cersonal pronouns, articles, adjectives, 
quantifiers, process and status verbs, separable verbs, loca- 
tive and limiting adverbs, prepositions, negative words, in- 
terrogative words, imperative words, rejoinders, inter jections, 
nounphrases, noun markers, plural and possessive, case marking 
in pronouns, number agreement, transitivity, prepositional 
phrases, verb modifications, progressive, past, indicative, 
complements, interrogative constructions, auxiliary elements, 
phase structure rules and transformations, etc. 


It is our aim to collect samples of at least 700 utterances from 

| children E and F, following the schedule presented in Table 3. (These 
data will be intensively analyzed return from the field, but 
should be transcribed in the field.) To be maximally comparable to 

| Brown's children, these children should (1) have a high rate of 
speech, (2) be largely intelligible, and (3) have a high rate of verbal 

interchange with mother or mother substitute. Brown and his co- 

; workers found it necessary to visit a large number of homes in order 

| to select three children fitting these criteria. Although we cannot, 
with a sample of two, either control for or study sex differences, it 

is at least advisable that children E and F be of the same sex, as we 

: wish to treat their records as at least vaguely continuous. (It is 

: therefore desirable, of course, that they and their home settings be 


similar in other respects as well). 


In addition to seeking children who meet Brown's three main cri- 


teria, the investigator must also find children roughly corresponding 
- to Brown's stages of linguistic development. Mean utterance length 
and upper bound (in morphemes ) should be determined for eligible 
, children on the basis of a sample of 100 utterances (either noted by 
hand or tape-recorded, as circumstances allow). The following are 
| Brown's rules (pers. comm.) for calculating mean utterance length. 
They will, of course, have to be modified on the basis of the lan- 


guage under investigation. 
| 1. Omit unclear and partial utterances. 


-> | 2. Include all exact utterance repetitions. Stuttering is 
: marked as repeated efforts at a single word; count the 
- word once in the most complete form produced. In the 
few cases where a word is produced for emphasis or the 
[ like (e.g., no, no, no) count each occurrence. 


3. Do not count such fillers as mm or oh, but do count no, 
| yeah, and hi. 


h, All compound words (two or more free morphemes ), proper 
- names, and ritualized reduplications count as single words. 
Examples: tel , Rackety-boom, choo-choo, quack-quack, 
a nigtan’ a pocketbook, see-saw. Justification is that no 
evidence that the constituent morphemes function as such for 


these children. 


| Semen 


20 


5. Count as one morpheme all irregular pasts of the verb. 
Examples: got, did, went, saw, etc. Justification is 
that no evidence that the child relates these to present 
forms. 


6. Count as one morpheme all diminutives (doggie, moumie,etc. ) 
because these children at least do not seem to use the 
suffix productively. Diminutives are the standard forms 
used by the child. 


7. Count as separate morphemes all auxiliaries. For example: 
is, have, will, can, must, would. Also all catenatives: 
gonna., hafta, wanna. These latter counted as single mor- 
phemes rather than as going to or want to because evidence 
is that they function so for the children. Count as sep- 
arate morphemes all inflections. For example: possessive 
{s}, plural {s}, 3rd person singular {s}, regular past {D}, 
progressive {in}. ? 


The investigator cannot, of course, hit the definitions of Stage 1 
and Stege 4 exactly on the mark. Brown himself has the following 
spread in his data, where the figures are based on about 700 utterances 
per child. 


Stage Child Mean Utterance Upper Bound 


Length 
1 Adam 2.06 4.0 
1 Eve 1.68 4.25 
1 Sarah 1.73 4.0 
4 Adam 3255 12.0 
h Eve 3.70 12.0 
h Sarah 3.62 10.0 


Note that no mention has been made here of chronological age. 
The goal of the investigator should be to find articulate and coopera~ 
tive children at these stages of development, regardless of age. 
(Child E, of course, will be younger than F.) E should be found by 
the third month in the field; F by the fourth. E, the younger, should 
be visited more frequently than F, as the rate of development will 
probably be more rapid at earlier stages. As shown in Table 3, E is 
to be visited six times, at intervals of about a month-and-a-halt ; 
while F is to be visited three times, at intervals of three months 
each. For each visit, recordings should be made on two consecutive 
days. Recordings should be mace at the same time of day, for each 
child, throughout the series. The aim should be to collect at least 
700 utterances at each two-day visit. We recommend two consecutive 
sessions of 1.5 hours for each visit to E, and two consecutive ses- 
sions of 2 hours for each visit to F. 


These visits should consist of free interaction between the child 
and mother (or mother-substitute), with minimal intervention by the 


21 


investigator and his assistant. The assistant should make on-the-spot 
notes about utterances which may be difficult to transcribe later, 
while the investigator should make sufficient notation of actions and 
situations to make it possible to retrieve the settings and meanings 
of utterances when making the transcription. The tapes should be 
transcribed as soon as possible, with the aid of the native assistant, 
and, if necessary, of the mother. (Appendix 2 contains advice on 
transcription problems and format.) It would be wise for the investi- 
gator and his assistant to get to know the child first, and perhaps 

to ask the mother to interpret the child's speech before each taping 
session t©gins. Time of day and activities should be selected with 
ease of recording in mind--e.g., circumscribed movement, alertness 

of child and mother, possibility of normal interaction, etc. 


The investigator should also administer to children E and F, at 
each visit, the standard imitation, comprehension, and eliciting pro- 
cedures described in Sections 1.4.2, 1.4.3, and 1.4.4 below. Work 
with children E and F should also be approached in the light of 
theoretical and practical issues raised in 3.2 and in Appendix l. 
Frequent reference should be made to Table 2, in order to ensure that 
all necessary data are gathered on these two central children (motor 
development, interaction networks, parental speech and attitudes, etc.). 
It may also be useful to occasionally give the mother definite tasks 
to communicate to the child in order to elicit speech from her (e.g. 
teach words, give order:, teach a game, play a game, read a story, etc.). 


22 


1.4.2. Elicited Sentence Imitation 


The standardized procedures described in this section, and the 
next two sections (1.4.3 and 1.4.4), will be administered to children 
G-N at two-month intervals, as shown in Table 3, and also to children 
E and F. At the beginning of testing, children G-N, as shown in 
Table 2, range in age from 230 to 432, with one boy and one girl 
picked at age intervals of eight months. That is, testing will begin 
with two children each of ages 230, 238, 334, and 4;2. The children 
should be articulate and cooperative. The imitation, comprehension, 
and eliciting procedures can probably all be administered in a single 
session. The major task in regard to imitation will be to construct 
and administer a standard set of about 50 sentences which the children 
will be asked to imitate at each session, as described below. 


Construction of Model Sentences 


The model sentences should be carefully designed to reflect the 
range of sentence types typical to child speech. They should vary in 
regard to grammatical form, intonation and stress patterns, and length 
(ranging from two to about eight morphemes). The major sentence types 
and syntactic processes of the language should be included: affirma- 
tive, negative, imperative, yes/no question, "wh" questions, negative 


questions, sentence conjunctions, embeddings, etc. The sentences should 


also sample major grammatical categories, such as tense, mood, aspect, 
etc. 


Some sentences should successively increase in length, in order 
to determine the nature of omissions in imitation (e.g. He sees the 
ball; He can see the ball; He sees the red ball; He can see the 
red ball; He can see the big red ball ...). It is especially valu- 
able to ask the child to imitate sentences which seem to exceed his 
productive capacity, allowing for revealing alterations imposed on the 
model sentence by the child's own system of grammatical rules. For 
example, Bellugi found the following informative imitation in a child 
whose speech was in a stage of multiple negation (pers. comm.) 


Adult: "Adam, say what I say: ‘The boy can't have any supper 
and the girl can't either.'” 


Adam: "The boy can't have no supper and the girl can't have 
none neither." 


The model sentences should also allow the child to impose normal- 
izations in his repetitions. For example, Slobin and Welsh (1967) 


found that a two-year-old girl tended to ignore repeated words in 
model sentences, as in: 


Adult: "Mark fell fell off the horse." 


Child: "Mark fell offa horse.” 


23 


Adult: “I can can can eat." 
Child: "I can eat." 


Similarly, contracted forms, at a given stage, may be repeated as un- 
contracted (e.g. he'll as he will), or uncontracted as contracted 
(e.g. will not as won't). Such findings should be tested cross- 


linguistically. 


Another phenomenon worthy of cross-linguistic validation is the 
finding of Slobin and Welsh that conjoined sentences are frequently 
inverted in inmediate imitation. For example: 


Adult: "Mommy ate the candy and Mommy ate the ice cream." 

Child: "Mommy eat the ice cream and Mommy eat a candy." 

In constructing the model sentences, the researcher should be 
careful to avoid phonological ambiguity (e.g. given a model sentence, 


"That truck is big,” it would be difficult to tell if the child imi- 
tated the first word as that or the). 


In designing model sentences, and evaluating the results of the 
imitation tests, the findings of previous investigators should be 
borne in mind. Attempts should be made to produce data relevant to 
the findings and speculations of American psycholinguists, as exem- 
plified in the passages quoted below: 


With increasing age, children produce more imitations 
that are morphemically identical with the original. With 
increasing age, the imitative utterances produced include 
larger number of morphemes--approaching the numbers in the 
model sentences. The morphemes produced are invariably in 
r) their original order. Omissions do not appear to be random 
| or idiosyncratic. On the contrary, it looks as if, across 
children and across sentences, there is a consistent ten- 
dency to retain one kind of morpheme and drop another kind. 
The two sorts of morphemes contrast on several correlated 
dimensions. The morphemes most likely to be retained are: 
morphemes that occur in final position in the sentence; 

[' morphemes that are reference-making forms; morphemes that 
belong mainly to the large and expandable noun, verb, and 
adjective parts-of-speech; morphemes that are relatively 
unpredictable from the context; and morphemes that receive 
the heavier stresses in ordinary English pronunciation. 
The morphemes least likely to be retained are: morphemes 
that occur in intermediate positions in the sentence; 
morphemes that are not reference-making forms; morphemes 
that belong to such small-sized grammatical categories as 
the articles, modal auxiliaries, and inflections; morphemes 
that are relatively predictable from context and so carry 
little information; and morphemes that receive the weaker 
stresses in ordinary English pronunciation (Brown & Fraser, 


1964, pp. 75-6). 


ter ga 
en 


x & * 


2h 


This very preliminary analysis [of 1000 elicited imitations 
from one child} has convinced us that sentence recognition 
and imitation are filtered through the individual's productive 
linguistic system. More specifically, we believe that we can 
tentatively offer the following generalizations: 


Echo [the child's pseudonym] can spontaneously utter 
sentences which she cannot imitate. On the other hand, she 
can give recoded imitations of model sentences which exceed 
her productive capacities. 


Emphasis can lead her to repeat words she would normally 
omit from imitation, but she generally ignores repeated words 
in imitating model sentences. 


If she comprehends a sentence, she need not repeat it in 
the order given. Reordering can also take place as a result 
of imposing SVO [subject-verb-object] constructions upon model 
sentences. 


The process of sentence recognition includes retrieval 
of both form and content. Syntactic structures take up 
space in memory, and frequently content will. be sacrificed 
to the retention of form in immediate, rote imitation. On 
the other hand, if content has been retrieved and stored, it 
may be encoded in the child's own syntax in repetition. 


A fine-grained analysis of repeated imitations of system- 
atically varied model sentences can reveal aspects of the 
child's theory of syntax, including transformational rules and 
the syntactic and semantic markers borne by lexical items 
(Slobin & Welsh, 1967, p. 17). 


* %* & 


It is clear that A and B structures were different for our 
subjects: the A structures were relatively easy for them to 
repeat, the B structures were quite difficult. .. . We dif- 
ferentiate between A and B surface structures in terms of a 
property I will call compression. Compression refers to the 
way semantic information occurs in a sentence. When sentences 
have low compression, semantic information is distributed 
fairly evenly throughout the sentence; when sentences have 
high compression, semantic information is bunched together, 
or compressed, at the nounphrase or verbphrase level. in 
terms of tree structure, highly compressed sentences have 
NP or VP nodes dominating several information-carrying 
elements. In sentences with low compression, NP or VP nodes 
dominate relatively few informetion-carrying elements. 


The A structures, which the children found easy to re- 
peat, have relatively low compression; B structures, which 
were more difficult, have relatively high compression (Smith, 


25 


1966, pp. 11, 13-14). [Smith's model sentences can be found 
in Appendix 1.1.5.] 


%* % * 


One may state that within the bounds of a two- to nine- 
wore, sentence, the length of the sentence is not critical in 
determining the success of repetition even for children as 
young as 3 years. The differences in the ability of children 
to repeat the various sentences seems to be dependent on the 
particular rules used to generate these sentences rather than 
sentence length. . . . The most frequent types of deviation 
from complete repetition found were modification of trans- 
formations and correction of restricted forms [i.e. forms 
restricted to child grammar]. The modifications always took 
the form of using a set of rules which come earlier in the 
grammar than those used in the sentence. .. . When given the 
memory eid of immediate recall of . . . sentences, @ significant 
number of the children reproduced transformations which they do 
not use in their own sentences (Menyuk, 1963b, pp. 436-7). 
[Menyuk's model sentences can be found in Appendices 1.1.2 and 
1.1.4.) 

* * * 

Before attempting to design model sentences in the field lan- 
guage, the investigator should carefully examine Appendix 1.1, which 
contains lists of model sentences from a number of American studies 
of elicited sentence imitation. 


Administration and Pranscription of Imitation Test 


The sentences should be presented by a native speaker, in normal 


Bi. speaking intonation, and both the model sentence and the child's imi- 
tation should be recorded. It should be fairly simple to train chil- 
r dren to imitate. For American children, the instruction, "Say what I 


say," is generally sufficient, although brief preliminary training is 
necessary. Menyuk used the following instructions: 


The E [Experimenter) said, "I'm going to say some 
= sentences for you. I want you to say just what I say. If 
I say ‘The sun is shining’ I want you to say A 
| If the correct response was obtained the E proceeded. If 
i: not, the E stated, "No, you say just what I say. If I say 
'The sun is shining’ you say -" No more 
than these two examples of what was required was needed by 


any of the Ss [children aged 2310 - 631] (1963b, p. 431). 


If a child fails to repeat a sentence, he should be given a sec- 
ond chance, with a second presentation of the model sentence. If he 


te seems to fail to respond because of distraction, the examiner should 
return to the sentence later in the series. Special note should be 

4 made of sentences a child is reluctant to imitate, or refuses to 

ee imitate. 


: 


eee ee ee 
—_ "Tr _— 


26 


Before beginning each recorded session, the investigator should 
verbally note on the tape the name of the child, the date, the time of 
day, and the experimental setting (location, other people present, 
etc.). This information should also appear on the transcribed data 
sheets. The tapes should be transcribed soon after the session, with 
the help of the native examiner. The transcriptions need contain only 
the children's utterances, as the model sentences will be standardized. 
The model sentences should be numbered, and the transcriptions should 
indicate the number of the model sentence corresponding to the child's 
utterance. 


In analyzing the data of the imitation test one should note: 
(1) omissions and changes introduced by children of different ages, 
(2) number of children of a given age imitating a given sentence 
correctly, (3) changes with age in a given child's imitation of the 
same sentence, (4) degree of preservation of pitch, stress, and order 
of elements. 


The data should be examined before the next session with a given 
child, to allow the experimenter to follow up any suggestive leads 
with the administration of additional model sentences for repetition 
between standard sessions. 


27 


1.4.3. Speech Comprehension 


When working with children E-N, the investigator will want to dis- 
cover the range of linguistic structures which they can comprehend, as 
well as imitate or produce. Ideally, all three aspects of linguistic 
performance--imitation, production, and comprehension--should be ex- 
amined for each linguistic structure studied. Developmental psycho- 
linguistics, however, has had little success in devising adequate 
measures of the comprehension of linguistic structures, except for 
those grammatical distinctions most readily picturable (singular- 
plural, subject-object of action, etc.). 


An important, unsystematic cue to a child's linguistic compre- 
hension comes from an examination of his answers to questions. This 
means of discovering comprehension will be most valuable for children 
E and F, for whom the investigator will have many hours of recorded 
mother-child dialog. Comprehension can also be tested in experimental 
children G-N through carefully constructed questions, especially if 
applied to stories which have just been told, to pictures, or to other 
recent or ongoing events. 


More systematic techniques can be developed with the use of 
standardized pictures. One technique, developed by Fraser, Bellugi, 
and Brown (1963), involves pairs of pictures illustrating a minimal 
grammeticel contrast, such as singular-plural, and the instruction 
to point to the picture corresponding to a given utterance (e.g. 
“The kitten plays" vs. "The kittens play,” presented in conjunction 
with appropriate pictures). ‘The pictures used by Fraser et al. are 
described in Appendix 1.2.1. 


Our pilot studies have raised certain problems, however, in re- 
gard to the use of pairs of pictures. It is often difficult to get 
children to point clearly to one of the pictures, and, even if the 
child does point, it is difficult to determine if he is guessing. 
Osser (pers.comm.) has developed a modification of this format, using 
three pictures: "To test for comprehension of a specific structure 
the child is presented with a set of three pictures, two of them cor- 
responding to the pair of test sentences, the third picture being a 
‘neutral’ picture with the same visual elements as the other two 
pictures, but with a different arrangement." For example, to test 
for comprehension of subject-object relations in the passive, Osser 
uses the following pictures: 


@ girl pulling a boy in a wagon 
@ boy pulling a girl in 4 wagon 
a boy and a girl pulling an empty wagon 


The child is then presented with either the sentence, "The boy is 
pulled by the girl,” or, "The girl is pulled by the boy, " and must 
indicate the appropriate one of the three pictures. 


28 


This technique, as the picture-pair technique, requires that all 
of the pictures in a given presentation contain the same elements 
(e.g. boy, girl, wagon). The experimenter should first offer the 
child both sentences, and then ask for the picture ccrresponding to one 
of them. For example, “Here are some pictures. One of them is called, 
‘The boy is pulled by the girl'; the other [or another] one is called, 
‘The girl is pulled by the boy.’ Now show me: 'The girl is pulled by 
the boy.*" 


In order to avoid the ambiguity of an uncertain pointing response 
in a young child, it might be best to mount each picture separately, 
and ask the child to hand the correct picture to the experimenter. 
(The pictures, of course, should be sturdily mounted and covered with 
cellophane or the like to protect them from sticky fingers and other 
hazards.) Fraser et al. offer the following useful methodological ad- 
vice for research of this kind: 


The pictures belonging to a contrasting pair were mounted 
side by side in such a way that there was no consistent 
relationship on any problem between the order in which a 
verbal contrast was given and the positions (left or right) 
of the pictures. 


We wanted to be certain that censistently correct re- 
sponses should be possible only for Ss [subjects] having 
control of the grammatical contrasts presented. A S ought 
not to be able to infer correct responses from the positions 
of the pictures, the order of the problems, the order in 
which he was asked to respond to the parts of the contrast, 
or from any interrelationship among these variables. .. - 
To make sure of these things we used counterbalanced and 
randomized orders. . « e« 


Before performing any of the three tasks, the S was 
shown a colored picture-book and encouraged to talk about 
the pictures. This helped the child to overcome any re-~- 
luctance to talk and provided a small sample of the child's 
spontaneous speech. 


. . « After performing both halves of an item, S, whether 
or not he had answered correctly, was encouraged by E 
[experimer.ter] with some such remark as "That's the way." 
Nonverbal rewards were not used (1963, pp. 128-9). 


The pictures, of course, will have to be designed to illustrate 
relevant grammatical devices of the language in question. In Russian, 
for example, a single pair or triad of pictures depicting subject- 
verb-object relations would not be sufficient; rather, 2 number of 
such picture sets would have to be drawn in order to present animate 
and inanimate objects in the masculine gender, and also objects in 
the feminine and neuter genders, because the object in Russian 
sentences receives a variety of inflectional markers as 2 function 
of animation and gender. 


29 


The experiment of Fraser et al. demonstrates that, in spite of the 
problems involved with the technique of pointing to one of a pair of 
pictures, such a comprehension test can yield unexpectedly rich data. 
For example, they found that American three-year-olds treated passive 
sentences on the basis of standard English word order, treating the 
first noun as subject and the last as object, thus revealing an im- 
portant aspect of the English-speaking child's processing of sentences. 
They conclude: 


In active-voice sentences, subject and object appear in 
that order, whereas in passive-voice sentences the order is 
object and subject. Suppose the 3-year-old processes each 
passive-voice sentence as though it were in the active voice. 
"The girl is pushed by the boy” is not computed as: Object-- 
Verb an the passive--Subject, but rather as: Subject--Verb in 
the active with odd appurtenances--Object. The odd appurtenances 
are is, -ed, and by which S$ may take to be signs of some uncommon 
tense like "will have pushed." Processing the sentence in this 
way would enable S to maintain the generality of the usual rule 
of English word order in which the subject precedes the object. 


With the Subject/Object contrast in the passive voice we 
turned up a revealing pattern of evidence by accident. If an 
investigator wanted to use the .. . procedures to test hypotheses 
about particular aspects of grammatical operation, he could 
easily design problems that would be revealing by intention 
(1963, p. 133). 


A variety of instructions to manipulate objects can also be re- 
vealing of a child's capacities to comprehend speech. In construct- 
ing such situations, however, the investigator must be careful to 
ensure that the only cue to correct performance is linguistic. 
Parental reports of children's comprehension are almost invariably 
based on understanding of entire utterances in situations with a good 
deal of non-linguistic support. For example, if a child performs 
correctly when told to “put the glass on the table” and "put the milk 
in the glass," there is no evidence of his comprehension of the dif- 
ference between the prepositions on and in. He may have responded in 
like fashion to the two sentences with the prepositions switched, or 
omitted. A controlled situation such as the following, however, can 
serve to test comprehension of the two prepositions alone: Present 
the child with two cups, one inverted and one open, and instruct him 
to put an object "in" or "on" the cup. Or present the same instruc- 
tions with only one cup present, inverted or right-side-up. The im- 
portant control in situations such as these is that there is no cue 
to the child other than the preposition. (Additional examples of 
object-manipulation teciniques for testing comprehension can be found 
in Appendix 1.2.2.) 


The problem facing all such techniques, however, is that only a 
limited number of grammatical distinctions can be realized in the form 


30 


of pictures or object manipulations. Comprehension of subtle matters 
such as combinations of tense, mood, and aspect, or embedding, and s0 
forth, will have to be examined by the construction of ingenious story 
comprehension and question-and-answer techniques which probably go 
beyond the scope of the present project. 


Appendix 1.2 should be carefully studied to provide ideas for 
the construction of feasible comprehension tests in the field. 


31 


1.4.4. Elicitation of Speech 


In addition to recording spontaneous speech, the investigator 
will want to attempt to elicit particular types of constructions. 
Appendix 1.3 contains many examples of eliciting procedures which have 
been used with English-speaking children; additional advice on elicita- 
tion can be found in Section 3.4.3.5. Eliciting procedures can help to 
increase the richness of data yielded by recorded sessions with chil- 
dren E-N. The American literature contains a number of useful sug- 
gestions for eliciting specific grammatical constructions from little 
children. The following quotations from research studies should be of 
value to the investigator: 


- « » attempts to elicit negatives were made by putting 
clothes on the wrong doll or puzzle pieces in the wrong 
place. Doll clothing was used for eliciting possessives. 

To elicit an interrogative sentence, two dolls were “fed, " 
and the child was told to “ask Joe what he wants to eat. 
Plurals were produced for nonsense objects made of play-doh 
[or wood]. If one of the dolls was left behind, "because 

he is sick," the child was asked to talk on the toy telephone 
to find out how "Joe" was, another device for eliciting 
questions. Telephone conversations with the child tested his 
reliance on verbal rather than gestural cues... . 


On the pronoun test, the child was questioned about 
pictures, the questions being designed to elicit sentences 
containing possessives and nominative pronouns varying in 
number and gender. In the test of discourse agreement, 
questions tested the class of responses, verb restrictions, 
comprehension of subject-object distinctions ("Who is he 
feeding?" vs. "Who is feeding him?") and "why" questions. 
There were two matched forms alternated each month, using the 
same pictures (Miller & Ervin, 1964, pp. 12-13). 


* + & 


To test for yes-no inversion, the child would be instructed 
to “ask the doll if he wants some food"; or for the morpho- 
phonemic putterns of irregular verbs we would say "Ask the 
doll what ne saw yesterday" and the child's response would 
be either "What did you saw yesterday?” or "What did you see 
yesterday?". Having the child imitate sentences or phrases 
provides another check, since it has been found that children 
seldom imitate grammatical patterns that are not a part of 
their own language system. Still another technique is to 
present the child with a choice such as: "Is it a new red 
dress, or a red new dress?" The child almost always picks 
the last of two alternatives if he lacks the appropriate 
grammatical rules (Miller, 1964, p. 7). 


* * 


We hoped to get as much speech as possible in as little 
time as possible and to have examples of the full variety of 
sentence types the child could produce. There were those who 
warned that the child would be shy and speechless in our 
presence; this was not the case. Mothers told their children 
that visitors were comiag and, in general, we were eagerly 
welcomed, shown a parade of toys and games, and talked to 
rather steadily. It became clear that the child expected a 
guest to put in some time as a playmate, and so the recording . 
was @ two-man job with one of us taking data and the other 
prepared to play cowboy, horsie, blocks, coloring, trains, and 
the mule in "Kick the mile." [If this is a cross-cultural 
= the investigator will have more than his hancs 
fullij} ... 

Much of the time the child was occupied with his normal 
routine of play, talking with his mother, washing, and eating. 
So long as these activities involved a reasonable amount of 
speech, we took no active part beyond delivering signals of 
attention and approval. When the operant level was very low, 
we sometimes tried to raise it. In the first days we did the 
sort of verbal prompting that is anyone's first notion of a 
technique for eliciting speech from children. You ask what 
something is called and this brings out vocabulary items --which 
are not useful for a study of grammar. Or you ask a "yes-no" 
question such as: "Is that your horsie?" to which the answer 
is either "yes" or "no." We eventually learned that it is 
easier to "inspire" speech by doing something interesting than 
to elicit it with questions. If the adult "playmate" starts 
a game that is simple, repetitious, and destructive, the child 
will usually join him and start talking. A universal favorite 
is to build (painstakingly) an unsteady tower of blocks and 
register chagrin when the child sends it crashing dow. A 
simple game involving implicit rules--such as the green blocks 
belong to me and the red ones to you--creates a situation in 
which negative sentences can be elicited. If the adult play- 
mate breaks the established rule and moves one of the chi1d's 
blocks, he is usually told that he is not to do that (Brown & 
Fraser, 1964, pp. 51-2). 


In addition to these general suggestions, Appendix i.3 con- 
tains explicit tests, designed by Ursula Bellugi, for the elicitation 
of negation (App. 1.3.1), interrogation (App. 1.3.2), reflexivization 
(App. 1.3.3), tag questions (App. 1.3.4), and transformations (App. 
1.3.5). Many of these techniques have clear cross-linguistic applica- 
bility. 


Another valuable eliciting technique is to introduce new 
words and see how they are dealt with by the child. The introduction 
of artificial woris into a child's vocabulary, of course, is but a 
special case of what happens every day. When new words are used in 4 
form other than that in which they were introduced, the child's lin- 
guistic competence is often clearly revealed. For example, when a 


33 


child spontaneously says “two mouses" or "I breaked the glass" it is 
clear that his speech is based on an underlying system of rules, that 
he has unconsciously made inductive generalizations which go beyond 

the specific utterances he has heard. Overregularizations and analogic 
forms have long been noted in child speech. When carefully studied 
they can reveal much of the child's developing grasp of the morpho- 
logical devices of his language. 


It is difficult, however, for the investigator to sit back and 
wait for the child to make such errors in speaking. Jean Berko (1958) 
has made it possible to gather such data quickly and easily in in- 
genious test in which little children are presented with nonsense 
words and asked to supply plural forms, change verb tense, and so on. 
A similar technique has been developed in the USSR by Bogoyavlenskiy 
(1957). Kernan and Blount have adapted this technique for use with 
Spanish-speaking children in Mexico (1966), and Ervin-Tripp and Miller 
have elaborated the technique for use with very young children (Miller 
& Ervin, 1964). All of the materials of these investigators are pre- 
sented in Appendix 1.3.6, and the general technique could easily be 
adapted to special problems of interest in a given field language. 
Berko describes her technique as follows: 


In order to test for the child's use of morphological 
rules of different types and under varying phonological con- 
ditions, a number of nousense words were made up, following 
the rules for possible sound combinations in English. Pictures 
to represent the nonsense words were then drawn on cards. 
There were 27 picture cards, and the pictures, which were 
highly colored, depicted objects, cartoon-like animals, and 
men performing various actions. . . . several actual words 
were also included. A text, muitting the desired form, was 
typed on each card. An example of the card test for the 
reguiar plural allomorph in /-z/ cun be seen [below]. 


» « - each child was .. . told that now he was going to 
look at some pictures. The experimenter would point to the 
picture and read the text. The child would supply the missing 
word, and the item he employed was noted phonemically. 


TiiTS IS A WU. 
: ) p) 
NOW THERE IS ANOTHER ONE. 


THERE ARE WO OF THEM. | 
| STHERE ARE TWO pee 2 


34 


As a warm-up, the experimenter should begin by presenting familiar 
objects and using real words. This will acquaint the child with the 
procedure, and will also provide necessary data on the child's use of 
the morphological features in question when speaking normally. The 
experience of Kernan and Blount in administering a Spanish form of the 
Berko test in Mexico is informative: 


In contrast to Berko, we experienced considerable dif- 
ficulty in eliciting answers from subjects. For this reason, 
it was decided to precede each series of questions which 
tested for a particular grammatical category with an example 
question which used a real noun or verb. For example, before 
he was tested on the formation of plurals, each child was 
shown a picture of first one cat, and then two, and was told, 
"Este es un gato. Ahora hay otro. Hay dos de ellos. Hay 


dos “(mis is a cat. Now there is another one. There 
are two of them. There are two .-) This is a departure 


from Berko's procedure which was necessitated by the children's 
difficulty in understanding the task at hand. This may in part 
be explained by a difference in familiarity to testing situations 
of this kind. American middle-class children have probably been 
exposed to formal testing situations far more frequently than 
have Mexican children of the lower socio-economic class. We 
feel, however, that the use of examples did not affect the re- 
liability of the test (1966, p. 3). 


Adults should also be tested, in order to determine the end state 
towards which the children are progressing. If possible, a wider range 
of children should also be tested--perhaps all of the children in the 
sample (A-X). Kernan and Blount tested children between the ages of 
five and twelve, and found continuing morphological development beyond 
the age range of four to seven tested by Berko in the United States. 


"Nonsense objects’ can be used as well as pictures in such testing. 
Miller and Ervin-Tripp constructed a collection of pairs of novel 
wooden objects in longitudinal testing of extension of English plural 
morphemes. An important aspect of their testing was the repeated ad- 
ministration of the same items, a feature also included in our general 
research design. 


The texts were at first coliected weekly . . . because 
of rapid change. [Collection began at about age two.) As 
the rate of change decreased and the fluency of the children 
increased, the frequency of text collection was gradually 
reduced, until texts were collected at two-month intervals, 
in two or three sets of closely spaced interviews totalling 
four or five hours. . . « 


The plural test consisted of 17 items. In each case a 
toy object or picture was shown, its name elicited, and then 
two were shown. "Here are two what?" ‘Thus the test did not 


35 


| give the children the option of using a syntactical plural 
signal rather than a morphological one. The items included 
certain pairs in which nonsense items (wooden constructions ) 

were given names which had the same final consonant as a 

familiar word (boy-kigh, block-bik, bed-pud, horse-tass, orange- 
bunge)- Irregulars were foot, man, and house. The singular 
of the regular and nonsense words was offered by the investi- 
gator if it was not offered by the child. In the case of foot 
and man, however, the singular was not offered by the investi- 

- gator. This procedure was followed in order to determine which 
form, e.g-., foot or feet, was used by the child as the singular. 
Testing was stopped on items after they had been contrasted for 
several months (Miller & Ervin, 1964, p. 12). 


Longitudinal testing of this sort in the field, using real words 
and analogous nonsense words, would provide a cross-linguistic check 
on Miller and Ervin-Tripp's finding that 


nearly always the contrast with familiar forms preceded the 
contrast with nonsense forms. . . . The average gap for boy vs. 
kigh was 2.4 months; for ball vs. kigh was the same; for block 
vs. bik the gap was 1.1 months; for cup vs. bik 1.5 months ... . 
Thus the child who calls one cup by a different term than two 

| cups might do the same with the nonsense toy bik in a month and 

b. a half (Miller & Ervin, 196%, p. 33). 


Procedures of this sort can easily be adapted for use in the field. 
The selection of grammatical features to test will depend on the lan- 
guage, aS will the construction of the nonsense words. If 2 native in- 
, formant is available prior to departing for the field, some parts of the 
/ ; test can be designed in advance. However, note that a test such as that 
4 used by Berko (given in full in Appendix 1.3.6) cannot be translated 
a wholesale into the field language. The distinctions chosen must be based 
( on the particular language. For example, Berko tested for singular and 
es plural. If the field lenguage, however, also has a dual, the two wugs 
picture would not be sufficient: an additional, multi-wug picture would 
aT be needed. Or if the language has genders which determine the form of 
{| noun plurals, special nonsense pictures and words will be required, and 
the introductory sentences will have to indicate the gender of the ob- 
ject. (E.g., "Das ist ein Wug. Jetzt gibt es noch ein Wug. Es geben 
zwei von ihnen. Es geben zwei os 


Another set of eliciting procedures requires the child to repeat 
fairly lengthy material offered by the experimenter. The length of 
Li the material makes the task exceed inmediate imitation span, thus re- 
vealing the child's productive system through the recoding which 
takes place. One such procedure is to tell the child a short story, 
perhaps accompanied by pictures, followed by the child's retelling of 


P the story. A similar technique is to devise a simple game, unfamiliar 
; to the child, and explain the game to him. He is then asked to explain 
[) the game to som2one else. It would be of interest to vary the age of 


36 


the target person, comparing explanations offered by the subject to a 
younger child, a friend of the same age, and an older child. (When 
the mother is present, a useful variant of this technique is to ex- 
plain the game to the mother and have her convey it to the child. 

This may be an important tool for the elicitation of mothers’ speech. ) 


Another useful eliciting technique is to ask a child to tell you 
a story. (Note if special narrative style is used--intonation, lexi- 
con, formlae, etc.) 


i 


37 


1.5. COMMUNICATIVE DEVELOPMENT | 


1.5.1. The Child's Lifespace 


A number of. sociolinguistic studies point to the decisive influence 
cultural factors exert on language development. Verbal. ability as 
measured in a number of ways has been found to correlate with socio- 
econamic status and ethnic background (e.g. Bernstein, 196la,b, 1962, 
1964; Anastasi & de Jesas, 1953). It is generally assumed that these 
observed differences result from differences in the life experiences 
of children from differcat sociocultural packgrounds. Little, however, 
has been done to relate these findings to the actual cultural practices 
which may engender these differences. 


Any account of the child's acquisition of communicative competence 
would be incomplete without some general investigation of the child's 
lifespace. Such an account should include a description of the child's 
physicel environmen’, and of the individuals who make up his network of 
interaction. For each of the 24 children we would like to have a de- 
scription of: the child's range of movement; the amount and type of 
linguistic input he is exposed to; where and with whom he spends his 
time; and the range of human artifacts, fauna, and flora with which he 


comes in contact. The goal, then, is to establish typical daily routines 


for children A-X. 


This information should be gleaned both from observation and from 
interview of the children and their parents. The investigator should 
begin as soon as possible to keep a daily diary of all of the settings 
in which he observes the children. The nature of the activity and the 
age, sex, and status of the individuals participating should be speci- 
fied. Interviews of the parents or caretakers of the children will 
provide information about the children's cycles of activities when not 
under direct observation. We hope in this way to be able to establish 
a yearly cycle of activity for children of various ages and to deter- 
mine the culturally defined categories of age-grading. For example, 
are weeks differentiated in such @ way that a child characteristically 
engages in different activities on some days as opposed to others? 
What role, if any, do seasonal changes play in circumscribing the 
activities of children? Into what time units is the annual cycle di- 
vided, and what special events punctuate the cycle? 


At some point during the field stay the investigator should de- 
yote some major part of a day (5-6 hours) to following each of the 
children and describing in detail the activities of the child for that 


Section 1.5.1 was written by Claudia Mitchell and Dan I. Slobin. 


38 


day. ‘These data, along with the evidence gathered from interviews of 
both the child and his parents, should give some idea of what a com- 
posite day in the life of a child of a given age and culture is like. 


If the society has formal education, part of the detailed observa- 
tion showld include classroom behavior of the school-age children under 
study. In so doing, the investigator should bear in mind the points 
raised in Section 3.4.4 (Formal Education), especially questions bearing 
on the role of speech, reading, and writing in school. 


The following is a check-list of information which should be 
gathered in regard to the lifespace of each of the children. 


I. Characteristics of Child's Household 


A. Physical Setting (provide map and pictures). 
1. Is it made up of one building or several? (How many? ) 
2, Are there room divisions? (How many?) 
3. What are the main functional areas of the dwelling 
space? 
h, Proximal relations of community 
a. Is the household a nuclear or extended family 
dwelling? 
b. Is it isolated or part of a compact village? 
c. Is the community rural or part of an urban center? 
d. Are the members of the community interacting 
individuals? 
e. What is the 6ize of the commnity? (Obtain 
census data if available.) 


B. Human Component 
1. Who are the individuals who make up the household? 
(Specify age, sex, relationship, status, occupation. ) 
>. What is the child's relationship to each member of 
the household? 
3. Do all members of the household have equal access to 
the entire dwelling area? 
4h, Are any members of the household, either by design or 
choice, generally or always confined to particular areas? 
5. Are there ever individuals present in the household who 
do not live there? If so: 
a. Who are they? 
b. How often do they visit? 
c. How much time do they spend there? 
a. What are their activities when they are present? 
6. How much time does the child spend in the household? 
7. Who else is present when the child is in the household? 
8. How much time do other members of the household spend there? 


ee See 


39 


II. Caretaking 


A. Which members of the household have something to do with the 
child? What? (Be specifice) 

B. Who are the child's main caretakers? (Do older children be- 
come parent substitutes for subsequent children?) 

C. What is the child's range of movement? (Is locomotion re- 
striction attempted? Specify.) 


III. Linguistic Input 


A. Who talks to whom? 

B. What topics are talked about? 

Cc. Are children ever instructed to be silent? On the basis of 
topic? People present? Etc. 

D. What settings are children freely admitted to? 

E. Are children excluded from some settings? 

F. Is language mastery ever a basis for the child's inclusion 
or exclusion from setting or conversation? 

G. Are any topics avoided in the presence of children? 

H. What are the settings in which the child is a passive on- 
looker? An active participant? 

I. Who are the individuals with whom the child characteristically 
interacts? (Be specific.) 
1. What topics are discussed? 
2. What is the duration of the interaction? 
3. Is the child the main focus of attention? 

J. Are metalinguistic comments ever directed to the child? 
By whom? 

x * * 

Besides focusing on individual children, questions such as those 
posed above can be turned around and applied to groups of children and 
settings in which they are found. The investigator should consider the 
following questions and suggestions in attempting to describe inter- 
action networks of children of different ages: 


1) Where do you find chi}“ren of each age? In what settings are 
pre-school children found? No children? Etc. 


2) How large are the groupings in each setting? 


3) vo children spend their time in age-homogeneous groups? 
(Does this depend on the age of the child?) 


4) Who interacts with the children, and what are they doing 
{| while children are present? What is the language input 
1 | to children in these settings? 


* 5) Study a specific setting for several days at the same time 

| of day. Describe interactions in detail: who initiated 
interaction and how? Who received the interaction? What 
were the consequences? Notice if the same patterns are 
repeated but do not appear elsewhere. 


y 


6) Ask informants to tell you who young children talk to, and 


7) 


check this information against direct observation. 


Once you have determined the nature of a given interaction 
network, try to get someone to violate the network--both as 


a test of your analysis, and in order to discover how viola- 
tions are responded to. 


41 


1.5.2. Dialog between Children 


In this section and the next (Parentei Baby Talk), we wish to ob- 
tain recorded samples of the speech styles used in addressing children 
of different ages. Here we are concerned with situations in which the 
interlocuter is an older child, whereas in the case of baby ‘alk it is 
the child's mother. In this section we are also concerned with ob- 
taining recorded samples of natural dialog between children of various 


ages. 


The most convenient time for the collection of these data is dur- 
ing the all-day behavioral sample of each child described in the pre- 
ceding section. On the day that a given child is followed about by 
the investigator, opportunities for the recording of natural conver- 
sation between children should be exploited. No doubt, many additional 
opportunities will arise in the course of the field stay. 


In addition to the recorded samples, which will be necessary for 
an adequate analysis of speech style directed to children of various 
ages, written notes on the form and content of interchild dialog will 
be of value. What do children of different ages discuss, argue about, 
play at? In taking these notes, the investigator should bear in mind 
the many theoretical issues and research suggestions raised in Section 
3.4.1, The Relation of Linguistic to Social Categories, and in Sec- 
tion 3.5, Usage and Styles. 


Section 1.5.2 was written by Dan I. Slobin. 


1.5.3. Parental Baby Talk 


Baby talk seems to be a feature of most, if not all, natural lan- 
guages. Here the term refers to that special form of language that is 
used primarily for talking to young children and is not the normal 
adult usage of the language. Baby talk is a conventionalized aspect 
of language which is learned and culturally transmitted as is the rest 
of language. If baby talk is found in 4 child's own speech, it has 
generally been acquired from an older member of the speech community, 
and is seldom invented by the child himself. 


As indicated in column XII of Table 2, we wish to collect recorded 
samples of mother's speech to children A-F, Q, R, W, and X. The older 
children are included as a control for the identification of special 
features of maternal speech directed to very young children. (Although 
only mothers are mentioned here, the heading of parental baby talk in- 
dicates that it would be of value to record the speech of fathers to 
children as well.) It would be best to have a standard setting, such as 
asking the mother to give certein instructions to the child, describe an 
event to him, tell him a story, and so on. If possible, some sections 
of the tape should contain stretches of the mother 's speech alone, allow- 
ing for later "blind" analysis of the tape, with no cue from the child's 
voice as to the age of the child being addressed. 


There are three broad areas to be investigated aid described ccn- 
cerning the use of baby talk by any linguistic group: the linguistic 
features of baby talk; the social context in which baby talk occurs; 
and the belief system of the group concerning baby talk and its use. 


Linguistic Features 


Before any analysis cam be done of the linguistic forms, the 
forms themselves must of course be elicited. Informants, when asked 
for terms that differ from the usual adult terms and that are used 
primarily witu children, may deny that any such baby terms exist. It 
may therefore be necessary to seek the use of baby talk in ongoing lin- 
guistic interaction. Furthermore, it will be necessary to be thoroughly 
conversart with the normal adult language so that baby words may be 
recognized as such. Once a few baby words have been discovered it 
should be possible to elicit others in an interview with an adult by 
presenting him with the terms you have recorded and contrasting them 
with normal adult usage. This should provide a clear example to the 
informant of the kind of data you are attempting to obtain. Another 
way to elicit baby terms is to ask the adult informant how children 
say certain words that are in the lexical domains in which other lan- 
guages often have baby terms. This my elicit some baby words from 


Section 1.5.3 was written by Keith Kernan. 


4s 


those adults who assume that the baby words in their language are 
actually the way all babies say things. 


Phono. . Baby words should be transcribed phonetically. A gloss 
of the term s be given and the normal adult form should also be re~- 
corded phonetically. The adult forms and baby-word forms should then be 
contrasted to determine what type of differences exist between them. 
The following phonological features are common to baby talk as con- 
trasted to normal adult usage in a number of languages: simplifica- 
tion of consonant clusters; replacement of velars (e.g. "o") by 
apicals (e.g. "d"); replacement of an "y" by another consonant; some 
kind of interchange among affricates, velars, ani stops; distant nasal 
assimilation; loss of unstressed syllables; swall selection of vovels; 
a morpheme canonical form of CVC or cvc(c)v. 


Grammar. The grammar of baby talk may differ from that of adult 
language in both form and function. On the one hand, baby talk may make 
use of the same grammatical categories and convey the same type of in- 
formation that the adult language does, but in a different manner; plural- 
ity for example, may be expressed by means of reduplication whereas it is 
expressed in the adult language by means of affixation. Baby talk, on 
the other hand, may convey information absent in the adult language and 
make use of grammatical categories that the adult language does not em- 
ploy; for example, a diminutive form that is totally lacking in normal 
usage may be present in the baby talk. More commonly, baby talk ignores 
grammatical distinctions made in the adult language and fails to convey 
grammatical information (such as tense) that is conveyed by adult Jan- 


guage. 


Gremmatical features common to baby talk in a number of languages 
include: a diminutive form, reduplication, absence of inflectional 
affixes, the presence of a special baby talk affix. 


Lexicon. The words of baby talk seem to be concentrated more in 
some domains than in others. This is in part the result of the small 
child's lack of contact with parts of his cultural environment and in 
part the result of the emphasis given by edults to the particular do- 
mains of the culture they deem important and thererore present to the 
infant early in his life. Care should be taken to determine whether the 
baby words have the same range of reference as the adult words they 
resembie. 


Among the most conmon topics for which lexical items occur in the 
baby talk of a number of languages are: kin names, nicknames, body 
parts, body functions, basic qualities (e.g. good, bad, little), names 
of animals, nursery games, toys, imperatives for common actions. 


Intonational and Paralinguistic Features. Paralinguistic and in- 
tonational features may mark baby talk as such. Because these features 


might not be presented by an adult informant in an interview situation, 
they should be recorded from ongoing linguistic behavior in which baby 
talk is being used. Some common intonational features that accompany 
the use of baby talk in a suzber of languages are: higher over-all 
pitch, preference for certain contour patterns, and more or less volume. 
Paralinguistic features might include facial expressions, physical 
closeness to the baby, and the like. 


Social] Context 


Social context includes both the participants and the setting or 
situation and each should be described fully for every particular speech 
event involving baby talk. 


The participants in the speech event include the sender, the re- 
ceiver, and the audience. It is not enough simply to identify by name 
the individuals who are filling each of these positions. Each partici- 
pant in a speech situation has a particular role to play vis-2-vis the 
other participants, and an attempt should be made to determine what 
this role is for each speech situation in which baby talk is used. 

For this, it will be necessary to have a good deal of background in- 
formation, which may be gathered either before or after the speech 
event itself is recorded. Such information should include as many a6 
possible of the positions in society, or socially defined statuses 
that the perticipant occupies. For exemple, it is not enough to say 
that Mr. Smith, a forty-two-year-old man spoke baby talk to his son. 
Mr. Smith's relevant statuses as defined by society must be noted. He 
may be classified, for example, as an elder who 4s an important person 
in his community and is expected to act with dignity at all times. It 
is important to record, therefore, not only the participants * roles 
vis-a-vis each other, but also the statuses each holds in the wider 
cultural context. 


The setting not only involves such information as time and place 
but should also include the native definition of the situation. That 
is to say, is it merely an informal speech event, or is it something 
that is defined as a particular kind of social situation by the native 
speakers themselves? What kinds of meaningful situations exist in any 


Baby talk has been defined above rather narrowly, and in any cul- 
ture terms that are primarily used for speaking to young children may 
also be used in other situations, such as with animals, supernatural 
beings, between lovers, and the like. It may be, also, that only a 
small and specific part of baby talk is extended into other areas. All 
such extended uses should be thoroughly investigated. 


5 


Children may imitate the adult use of baby talk when speaking to 
other small children. This imitative behavior will often provide in- 
portant clues to the child's impressions of and competence with adult 
speech. (See Section 1.5.2, above.) 


Belief System 


It is important to investigate not only what people do with baby 
talk but also what they think about it and what they think they are 
doing with it. It is important to ask what the purpose of baby talk 
is considered to be: Is it to facilitate teaching the child to speak? 
Is it easier for the child to hear? Is it an imitation of child 
speech? Is it an expression of affection? Does baby talk inhibit 
| the child's acquisition of normal speech? At what age should children 
stop using baby talk? What kinds of sanctions should be applied to 
force the child to stop using baby talk? Are adults embarrassed about 
using baby talk? and so on. 


. Geen | a 


a 


Some of these questions my be asked directly of adult infor- 
mants (see Section 1.6, below), but others, such as whether adults are 
embarrassed about using baby telk, may have to be discovered by ob- 
servation of situations in which baby talk is being used. Data ob- 

= tained from observing the use of baby talk may then be contrasted 
with what adults have said about its use. 


In addition, recordings of adults using baby talk my be played 
back to them and information gathered on their reactions; for example, 
surprise, embarrassment, and so on. Hypothetical situations may be 
given or tape recordings of baby talk played back to informants, who 
may then be questioned about the situation, the participants, and the 
setting and asked to give their interpretations and their attitudes 
concerning it. The reaction of informants to accounts of baby-talk 

= practices in other cultures may be obtained to discover their atti- 
| tudes concerning its use. 


Additional information on baby talk can be found in Casagrande 
] [1948] and Ferguson [1956, 1964].) 


® 
| 
\% 


Pig eh ge hn 


1.5.4. Communicative Routines 


Because of their brevity and surface predictability, some kinds 
of communicative routines are easy to identify. Routines mean utter- 
ances which are predictable or contain predictable segments. They in- 
volve very smll classes of alternatives. They are predictable from 
the social situation, and include two main types, prescribed features 
known to participants, and normal features of which participants are 
not usually aware. Since the former are easier to identify, we will 
treat them here, and reserve the latter for Section 3.5.3, The Analy- 
Sis of Natural Converse tion. 


1) Situation boundary markers. Within the daily cycle, there 
are routines marking changes of locale, activity, or participants. 
They are similar to rites of passage in the life cycle, making clear 
to everyone that a new phase has commenced. Examples are greetings, 
farewells, hosts' welcomes, pre- and post-prandial comments like 
"bon appétit"™ or grace, remarks proper to entering or leaving parti- 
cular locales. 


2) Accident markers. Unpredictable events like sneezes, 
belching, tripping, or dropping something may be followed by routines. 
The actor or some particular other may be expected to perform the 
routine. 


3) Apologies are comments pertaining to violations of social 
rules. They may occur prior to violation, in the form of asking 
permission, or afterwards. In the latter case they might overlap 
with accident markers. One can conceive that both might occur for the 
same event--e.g. an apology by a sneezer, and “salud” by his companion. 


4) Thanks refers to verbal compensation when goods or services 
are offered or received from another. 


5) Requests refers to complex functional category, all those 
utterances which have the effect of eliciting goods and services from 
others. In English, not all requests contain routines like "please," 
"would you mind..." “could you..." Requests may be more dif- 
ferentiated according to the social category of the addressee than 
other forms, so they provide a nice indicator of social learning. 


6) Gesture-word games. While purely verbal games are discussed 
in Section 3.5.2, we may include in routines those classes of utter- 
ances which occur only in the situation of a stereotyped gesture or 
performance. For instance, in European cultures, babies receive such 
gesture-word games from adults. Some, like "cou-cou" or “peek-a-boo,” 


Section 1.5.4 was written by Susan M. Ervin-Tripp. 


ee 


47 


start as passive audience games for the child, but later children will 
act and speak, or act on command of the word. "Bye-bye" and waving, 
"pat-a-cake" and clapping have the same relation. While the words may 
for the child function as names of the activity, there is a difference 
from other activity names in that one rarely occurs without the other, 
at least in the adult performance. 


Many of these categories can require the inclusion of address 
terms. Thus in French one does not use greetings or farewells with- 
out some address term. For this reason, routines may provide a con- 


text in which appropriate terms of address are acquired and can be 
studied. 


Some Problems to Study 


Since the categories above have different properties, we would 
not expect them to be mastered at the same age. Some considerations 
to take into account in doing a study of these forms are these: 


1) Forms requiring merely that the child respond to a routine 
initiated by someone else can begin as early as any comprehension of 
language. So waving in response to “pye-bye" could be found quite 
early. 


2) Reciprocal routines, where two parties say the same thing, 
should appear early since they can be cued by mimicry. For this 
reason situation boundary markers might appear earlier than apologies 


or thanks, provided they are reciprocal. So when an adult says 
"hello" the child can too. 


3) One might expect some forms of error to be common. For ex- 
ample, non-reciprocal "thank you" might be interpreted by the child 
as merely a label for exchanging things. If this is the case, we might 
find him using it when he gives objects, not just when he receives 
them. Also, the form might decline in frequency when the frequency of 
labeling declines, in the third year. 


4) If selection of the routine is based on complex social stimli 
the routine should be learned late. Thus, we might expect apologies to 
appear later than greetings. 


> 


5) If the categories of events requiring the routine are complex, 
we might expect selective usage. For example, in Japanese apologies 
are differentiated in a complex fashion. Some classes of infraction 
may not elicit apologies from children at first, and there may be over- 
generalization of the routines. There are interesting semantic problems 
in studying these overgeneralizations. 


6) Short isolated phrases like "please" should be easier to learn 
as routines than syntactically bound utterance segments like "canI.. ." 


and “may I..." It would be valuable to locate instances where 
there are functionally equivalent alternatives with different lin- 
guistic properties, to see which is learned earlier, and which is 
used more often in explicit instruction. 


7) Where there is evidence of social differentiation in fre- 
quency of routines with different addressees, it would be useful to 
check instructional practices to see if they support the difference. 
For instance, if a child says "please" less often to his age-mates 
than seniors, is this because he is more often reprimanded for omis- 
sion when addressing seniors? 


8) A variety of instructional methods are likely to be ob- 
servable. Routines being fixed, like lexical units, they can be 
taught either by mimicry ("say 'please'"), correction of deviant or 
omitted routines, or simply by letting the child mimic whatever 
occurs naturally, without any special tuition. 


Methods 


We suggest four routes to obtaining data on routines, including 
all the children in the sample. 


1) Notice instructional behavior during other observational 
sessions. Even with babies in the babbling stage, there may be in- 
struction in simple gesture games, waving, and so on. 


2) Include questions about speech etiquette in interviews. 


3) Note the appearance of repeated forms at fixed points in so- 
cial situations. The analyst's problems in identifying such forms in 
adult usage, or in the usage of the older children, may suggest some 
hypotheses to test in the children's acquisition. 


4) Identify situational slots, and look for regularities in 
verbal or gestural behavior in these situations. It might be desirable 
to find adults or older children as points to begin such a study. How- 
ever, just as in the case of vocabulary or grammatico-semantic cate- 
gories like negation the match of semantic category and verbal repre- 
sentation may not be the same as for adults at first, one should look 
for any regularities of behavior in these situations, whether or not 
they match the adult routine. For instance, whereas adults in a family 
may say "goodnight" before retiring, children may kiss or wave. This 
difference may or may not be a result of deliberate instruction of chil- 
dren. Waving might arise from generalization from farewells. On the 
other hand, caretakers may invent boundary markers to help control chil- 
dren's activities. 


Sampling 


Once an analysis is made of the routines employed in the adult cul- 
ture, one should try to collect evidence from each of the children A-X 


a 


| 
| 
| 


ee 


kg 


for the appropriate situation. Some, like accidental events, may not be 
possible to observe easily in the children explicitly selected for the 
sample. However, if one is alert to the categories, in the course of 
the year some other children in the same age bracket may substitute. 
Where the situation elicits the action, as in thanks end the situation 


| poundary markers, one can simply semple those situations on purpose. 


a 


1.6. LINGUISTIC BELIEF SYSTEMS 


Anthropologists have traditionally made a distinction between ! 
: what people do, what they say they do, and what they think they should 
do. Most of this manual is devoted to discovering what people do when 
learning their language as children and what they do as adults in the 
presence of children learning languages. In this section we address 
ourselves to the task of discovering what adults report that they and 
their children do in the language-learning situation, and what they 
say they should do to assure rapid and complete language learning on 
the part of their children. 


This information is best obtained, of course, in an interview 
situation. Before beginning to conduct interviews, the investigator 
should carefully read the sections on the use of interpreters and 
interviewing of informants in the Field de for a st of social- 
ization (Whiting et al., 1966, pp. is6-ith = volume which each in- 
vestigator should have with him in the field. 


For comparative purposes, we suggest @ specific list of interview 
questions below. This should be considered a minimal list, and all 
researchers should ask at least these questions. It is by no means a 
complete list, however, and additional questions will undoubtedly oc- 
cur to the investigator while in the field. Neither is it necessary 
for the questions to be put to the informants in the exact form pre- 
sented here. This list is designed to serve more as & check list for 
the investigator than as an interview schedule to be strictly followed. 
(See Section 2.3 for additional cautions on the use of native infor- 
mants, assistants, and interpreters. ) 


We suggest that this section of the study be carried out at the 
beginning of the stay for a number of reasons: It is one of the 


language and speech. And,perhaps more important, it will introduce 
the investigator to the community and give its members some idea of what 
he is doing there. 


A beginning step in the investigation of a people's belief system 
concerning language, perhaps, should be a familiarization on the part 
of the investigator with the metalanguage used in the culture for dis- 
cussing speech behavior. Concepts equivalent to such English words 
as "word," "crying," "cooing," "babbling," etc. should be learned and 


Section 1.6 was written by Keith Kernan. — 


| 


[| ee ee ee eee 


: 


pe 


clearly understood. This might be best done with the help of a bi- 
lingual assistant who is clearly aware of the investigator's aims and 
interests. in 


Beliefs regarding language learning should not be collected from 
& single segment of the populetion. such as parents of young children; 
rather, some sort of selection proc:cure should be used which will 
assure representation in the sample of all culturally relevant groups. 
Answers should be collected, for exemple, from men as well as women, 
from adults as well as children, from parents as well as couples with- 
out children, from parents with c.der children as well as parents with 
younger children, etc. (As a miiimal sample, the mothers of children 
A-F, M, N, Q, R, YW, and X should be interviewed, as indicated in 
Table 2.) In addition to such obvious and easily identifiable cate- 
gories of individuals as those just mentioned, there may be additional 
groups which are more difficult to identify and which are formed by 
criteria which may be unique to the particular culture being investi- 
gated. Social class, however defined in the minds of the people of 
the culture, might be an important determine of attitudes concerning 
beliefs about language acquisition, for example. 


Since group membership may be an important factor in beliefs con- 
cerning language acquisition, it will be necessary to record informa- 
tion concerning each informant questioned. Some of this information 
will undoubtedly be determined by the types of relevant groups which 
exist in the culture. Whether an individual is a member of an or- 
ganized religion, for example, may be important in some cultures and 
not in others. For this reason, it would seem wise to record as mech 
biographical data as possible for each informant at the beginning of 
the project at least. Checking answers to questions against biograph- 
ical data while in the field my lead to the discovery of interesting 
covariations which will enable the formation of preliminary hypothe- 
ses and point to the uselessness of some biographical data which had 
been collected in the past and the necessity for obtaining other more 
relevant facts. 


In addition to culturally specific facts such as caste membership 
which might be suspected by the investigator to have possible relevance 
in determining beliefs about language acquisition, a minimal amount 
of biographical information would include the following: 


1. Age 
2. Sex 
3- Occupation 
4. Socioeconomic status as defined by the criteria of the 
culture itself 
5. Monolingual or multilingual 
(non-native languages spoken) 
6. Number of children 
7. Sex of children 
8. Ages of children 


OR 6 Tn A Rao 
Seaotpiienas een oneal 


52 


9. Amount and type of education 
10. Place of residence 
11. Number of people in household 
12. Relationship of otner individuals in household 

to the informant 

13. Number of siblings 
14. Age of siblings 
F 15. Sex of siblings 
“7 16. Age of parents 

: 17. Education of parents 
18. Occupation of parents 
19. Socioeconomic status of parents 
20. Place of residence of parents 
21. Languages spoken by parents 


Suggested Minimal Interview Concerning Beliefs about Language Acquisi- 
tion and Use: 


A. Pre-Verbal Vocal Behavior 


1. At what age does an infant begin to make sounds? 

2. Can you describe these sounds? - 

3. Does he make these sounds at any particular time of day? 

h. Does he make these sounds in any particular situations? 

5. Does he make different sounds in different situations? 

6. Why does he make these sounds? 

7. Does the child mean anything by these sounds? 

8. Is he trying to speak when he makes these sounds? 

9. Do you understand what he means or wants when he makes 

these sounds? , 

10. Do all babies make the same sounds before they are able 
to speak? 

11. Can anyone at all understand what the baby wants or 
means when he makes these sounds? 

12. Do babies understand adult speech before they learn to 
talk? 

13. Can babies think before they learn to speak? 

14. Do babies imitate adult sounds? 

15. At what age does this imitation begin? 

16. What is the easiest sound for a baby to imitate? 

17. Is this sound learned first? 

18. What is the most difficult sound for a baby to imitate? 

19. Is this sound learned last? 

20. Would a baby learn how to speak if no one ever talked 
to him? 

21. Do you speak in a special way with babies (e.g., louder, 
softer, faster, etc.)? 

22. Does anyone speak in a special way with babies? 

23. Who doesn't speak in a special way with babies? 

2k, Do you use special words with babies? 

25. What are these special words? 7 


23 


26. Are these words easier for the baby to understand? 

27. Does the us2 of these words make it easier for the 
baby to learn to speak? 

28. Does the use of these words retard the baby’s learn- 
ing adult speech? 

: 29. At what age is the child when you stop talking to him 

| like this? 


B. The Acquisition of Speech a 


| : lL. How does a child learn to speak? Must he be taught? 
2. Do children learn to speak by hearing others and 
imitating them? 
3. What is the first word a child usually learns to speak? 
Ls 4. At about what age does he learn this word? 
5. Does he use it correctly? 
( | 6. Does it mean the same thing to him as it does to adults? 
é 7. If not, what did it mean to him and how did he use it 
incorrectly? 
r 8. At what age do children begin to put words together? 
: | } ; 9. Do they put them together correctly? 
10. What kinds of mistakes do children make in the sounds 
* of words? 
i ll. What kinds of mistakes ‘do children make in putting 
= words together to form sentences? 
12. Do all children mke mistakes in sounds and ‘putting 
| words together? 
i | 13. Do all children make the same mistakes in sounds and 
putting words together? 
4 14. Do boys or girls make more mistakes when learning to 
speak? 
, 15. Do boys or girls learn to speak earlier? 
q 16. Do any children learn to speak earlier than others? 
! 17. Why do some children learn to speak earlier than others? 
18. Is it beneficial to the child to learn to speak early? 
19. Is there anything that can be done to make a child learn 
to speak earlier than usual? 
From whom does the child learn most of his language? 
21. Is it better that the child learn his language from 
some people than from others? 
j | - 22. Do children learn many words from other children? 
23. What kinds of words do children learn from other 
children? 
24. Do children ever invent new words? 
a 25. Are the words that children invent ever used by other 
: people? 
26. Do children ever ask other people the names of things? 
27. At what age do children begin to ask the names of 
things? 
28. Whom do children usually ask the names of things? 
29. What kinds of things do children usually ask to be named? 
30. Do children ever ask the meaning of words? 


nN 
o 
oe 


54 


C. 


31. Whom do children usually ask the meaning of words? 

32. How do you think a child should be answered when he asks 
the meaning of a word? Can you give me an exemple? 

33. What kinds of mistekes do children make in speaking? 
Examples? 

34, Should a child be corrected when he misprewounces @ word, 
makes a mistake in putting words together, cr calls some- 
thing by the wrong name? 

35. How should he be corrected? 

36. Should anyone who hears a child make a mistake correct 
the child? 

37. From whom do children learn "not nice" words? 

38. What should be done when a child uses a "no? nice” word? 

39. Do children learn to speak better or more correctly by 
going to school? 

ho. Do you think there are many times when children should 
not talk? 

41. Why do you think it is wrong for children to speak at 
these times? 

4O. How should children be taught not to talk at these times? 

43. Should children speak differently to adults than to 
other children? If so, in what way and how is the child 
taught this? 

hh. How do you feel when your child makes a mistake in his 
speech in the presence of adults? Are you embarrassed? 

45, Do you think it is important for a child to learn to 
speak well? 

46. If so, why is it important for a child to learn to speak 
well? 

47, Is it more important for some children to be able to 
speak well than for others? 

48. Is it more important for boys or for girls to be able 
to speak well? 

49. Do children have any word games or special languages? 
Can you describe them? 

50. At what age do children begin to speak like adults and 
not make mistakes? 

51. Do all children learn to speak like adults at about the 
same time? 


Speech Behavior and Ability 


1. Are some people better speakers than others? 

2. In what way are they better speakers? 

3. How did they become better speakers? 

h. Are there any advantages to being a good speaker? 

5. Who do you know that's a good speaker? 

6. Are some people poorer speakers than most people? 

7. In what way are they poorer speakers? 

8. Are there any disadvantages to being a poor speaker? 
9. Who do you know that's a poor speaker? 
10. Do you know anyone who makes mistakes in his speech or 

sounds strange when he speaks (i.e., pathologies )? 


29 


11. What causes this pathology? 

12. Do you have a term for this strange kind of speech or 
for people who speak like this? 

13. Do you know of any other kinds of mistakes in speech or 
strange sounding speech (i.e., other speech pathologies )? 
What do you call them? 

14. What causes these speech pathologies? 

15. Can anything be done to correct these speech pathologies? 

16. Are some people more likely to have any of these speech 
pathologies than other people? 

17. Do you know anyone who speaks more than one language? 

18. What ere the languages they speak? 

19. Is it advantegeous or disadvantageous to be able to 
speak more than one language? In what way? 

20. Do you think it's a good idea for children to learn to 
speak more than one language? Why? 

21. What additional languages do you think children should 
learn to speak? Why? 

22. At what age should children learn to speak another 
language? 

23. Does learning a second lenguage hinder or improve 
ability with the first language? } 

2h, Is it easier for some people to learn to speak another 
language than it is for other people? Why? 

25. Are some languages easier for people, who already speak 
your language, to learn? Why? 


D. Language in General 


1. How do you think your language began? 

2. Why are languages different? 

3. Are some lauguages easier to speak than others? Why? 

4h, Are some languages better sounding than others? 

5. Can one say things in one language that couldn't be 
said in another language? 

6. What do you think people would be like if there were 


no languages? 
7. Do you think it would be a good idea if everyone spoke 


the same language? Why? 

8. What language should that be? 

9. Do animals have any kind of a language? Which animals? 

10. Can animals understand human language? 

11. Can humans understand animal languages? 

12. Is your language changing? 

13. In what way is your language changing? 

14. Do you think it is good or bad that your language is 
changing? 


15. Where do new words come from? 


1.7. NOTES AND QUERIES FOR OBSERVATION 


What follows is a scattered selection of issues » questions, and 
reminders which the investigator should try to keep at the top of his 
mind. This section should be scanned every few days, and the entire 
Manual should be scanned every few weeks. The investigator should 
faithfully keep a daily journal, in which he includes not only prob- 
lems raised in the Manual, but comments about his own mood, the 
weather, the social climate, and so on--such variables may have an im- 
portent effect upon the scientific observations. (See discussion of 


daily log in the Whiting et al. Field guide [1966, p. 166].) 
Recording 


Do not be seduced by the ease of tape recording vast amounts of 
data. Remember: it will take at least as long to listen to a tape as 


it did to record it, and much longer to analyze or transcribe it! 


Clean and lubricate the tape heads about once a week, and de- 
magnetize the tape heads (if required by the type of machine) about 
once a month. 


Observe Children 


What devices do children have for initiating information ex- 


changes--e.g. tugging, turning face, using openers like "do you know 
what?" Whom does the child address? 


Do children know proverbs, riddles, poems, chants, songs, prayers, 
etc.? How is musical development related to language development? Do 
children play musical instruments? How well can children of different 
ages sing back a melody or beat back a rhythm? Do the children dance? 
is there musical training? Are people expected to be creative musi- 
cally? To appreciate music? Are there formal musical performances? 
Do children go to them? 


Watch children play. What do they argue about? Who wins and 
why? Do they argue about the rules of the game? Is winning ever re- 
lated to verbal skills? Try to observe children playing at adult roles 
as @ means of determining their notions of adult speech styles and role 
behaviors. 


What chores are done by children of various ages and sexes? What 
are children expected to be able to do? To understand? 


Watch for cases of children using baby talk. 


Section 1.7 was written by Dan I. Slobin. 


Pay special attention to late-occurring errors, as these reveal 
points of special difficulty in language acquisition. 


How do children categorize other people in terms of variables such 
as age, status, sex, relationship, etc.? 


Observe Adults Relating to Children 


Where are little children carried? If on someone's back, does the 
child face forward, and can vocal interaction occur? 


What is the physical distance between speakers and listeners? Is 
this adjusted to the age and height of the listener? Where do adults 


_gtand in addressing children of different ages? Do they attempt to 


establish eye contact? 


What devices do adults use for getting a child's attention? How 
is a child called when he is out of sight? When he is within sight but 
out of reach? When he is within reach? 


How is a child sent to get something? 


How does an adult stop a child from doing something? How is a 
child treated if he interrupts a conversation between adults? How is 
a child scolded? Threatened? (What is the lexicon for verbal punish- 
ment?) Is praise or blame verbally administered by adults to children? 


What language modeling behavior is structured specifically for 
children? (E.g. fairy tales containing stylistic emphasis and repeti- 
tion, rhymes, chants, instructional materials, etc.). Do adults read 
to children? Tell them stories? 


How is reading taught, and who teaches it? What are the atti- 
tudes of children and parents towards learning to read and towards 


reading? 


Pay special attention to adults' responses to children's speech. 
At what age does speech become an object of explicit correction? Up 
until what age do adults correct a child's speech? What aspects of 
speech are corrected at various ages? How do children respond to 
speech corrections? 


What are children of different ages expected to be able to give 
verbal accounts of? (E.g. "what did you do in school today?" “What 
did you see at the circus?") Do children take part in dinner-table 
cohversation? Is there any special time or place for dialog between 
adults and children? 


How are children with deviant language or language deficiency 
treated? How are the deaf treated? How are the blind treated? 


Language 


What is the native terminology for language, speech, speech acts, 
etce.? Is there a term for stuttering? (Does stuttering occur?) Are 
other speech disturbances labeled? 


Be alert to the use of special styles of speaking at all times. 


If there are children's books, what is the style and content of 
such books? 


Pay special attention to address forms and third person reference 
(pronouns, names, kin terms, etc.). 


Are there word games, puns, riddles, limericks, etc.? 


Social Structure 


Observe sex-linked interaction in the community in order to de- 
termine the role played by your own sex in interacting with various 


categories of people. 


What role does birth order play in the society? Are special 
privileges and/or burdens dependent upon birth arder? 


What settings are children admitted to or excluded from on the 
basis of language mastery? 


What ritual and religious events occur in the community? Who 
participates? What is done? How are children initiated into religious 


rites? 
Linguistic Exercises 


Do a close phonetic transcription of a text, pushing your phonetic 
skills to the limit to force attention to details. (This should be 
done about four months after arrival in the field.) 


Ask natives to teach you proper speech etiquette for interacting 
with various kinds of people. 


Try to get children to teach you the language, and to help you 
understand the speech of other children. 


Present native informants with sentences uttered by children and 
ask them to say each sentence so it sounds right (or some such in- 
struction). This should give you som information about the systematic 


regularities in the child's language which are deviant from the adult 
language, and some general notions of areas of difficulty in the lan- 


guage acquisition process. 


Go through a published grammar of the language with a native in- 
formant and have him comment on it in the light of actual usage. 


Iearn songs, poems, chants, etc. 


Alert natives to the variables you are interest in, and see what 
sorts of data they bring you. 


Dell Hymes' Questions 


(The following questions were posed by Dell Hymes in his paper 
"Linguistic aspects of studying personality cross-culturally" | 196 j 
and ae among the original stimuli for the construction of this field 
manual. 


Societies may differ not only in the age at which children 
typically acquire speech, but also in the context of its acquisi- 
tion. For any society, one would want answers to such questions 
as: When is the child considered capable of understanding speech? 
(Among the Tlingit, "when the infant is but a few months old the 
mother talks to him, tells him moral tales, 'trains him'"). Is 
acquisition of speech accompanied by pressure, or treated as 
something that comes in due course? Are there special word 
games or speech patterns for teaching children? If there is 
pressure, at what stage of psychosexual development is the 
pressure applied? When are other socialization pressures applied, 
before or after the acquisition of speech? Various writings make 
clear that pressure, deprivation, and overprotection may variously 
induce speech defects or the preservation of infantile speech 
habits. = e * 


Other questions: to what extent is a child rewarded by 
verbal praise, in cotrast to material rewards such as candy, 
or physical affection? To what extent is the child punished 
by verbal reprimand, as opposed to deprivation, or physical 
pain? What is the conception of proper speech behavior on the 
part of the child, relative to particular persons and behavior 
settings? Are a child's questions about words and meanings wel- 
comed or rebuffed? Overall, is a child allowed much or little 
oral gratification through speech? Is a child encouraged, dis- 
couraged, or ignored in efforts to find satisfaction in speech 
play? What is the proportion of speech activity to communica- 
tion by other means, such as gestures, on the part of the child? 


Are there special settings for verbal instruction of chil- 
dren? If there are, how frequent and with what personnel, and 
about what topics? Is the instruction conventionalized in con- 
tent as are proverbs and myths, or only in theme? Is the tone 
of instruction categorical (as among many American Indian groups) 
or not (as in West Africa)? Is sex involved? ... 


PART TwO 


METHODOLOGICAL FACTORS 


2.1. CONTRASTIVE ANALYSIS 


On arriving at a field site, there are a few heuristic devices in- 
vestigators can use to help them identify the most important speech 
varieties, and on the basis of these varieties, the aspects of the so- 
cial structure which they must describe. 


Members of different speech groups or individuals in aifferent 
speech situations may change their speech in a variety of ways: (1) 
topics or functions of interaction are likely to be changed; (2) lin- 
guistic codes for representing the same functions may be changed; (3) 
linguistic features organized as a code may differ; and (4) vocal 
signals may be altered. Sources for further detailed reading are given 
below in this section. Dell Hymes' Language in culture and society 
(1964) contains pertinent material in Parts V and VII, and Shuy's Social 
dialects and language learning (1964) includes relevant articles by 


Bailey, Labov, and Stewart. 


Contrastive analysis is important for this project because it is 
necessary to identify linguistic contrasts in order to know whether 
and when children begin to produce stylistic shifts. They may of 
course use style shifting which is not similar to adult shifting; in 
such a case an analysis of child tapes similar to the analysis of the 
adult tapes described below will be necessary. In addition to & con- 
trastive analysis of the speech varieties mastered by children at 
various ages, it is important to understand which varieties are intel- 

ible to them, and how the speech varieties known to them are evalu- 
ated. Both comprehension and evaluative norms must develop with age, 
and possibly there are some sharp changes with formal training. These 
points are discussed later in this section. 


2.1.1. Methods 


After a few months in the field, the investigator will be able to 
observe on his own whether there are gross differences in speech be- 
tween certain speakers, addressees, situations. To begin systematic 
contrastive studies, there might be two starting points. One might be 


Section 2.1 was written by Susan M. Ervin-Tripp. 
61 


to start with some general informant work. Later informant interviews 
of course would be necessary to refine this information. The purpose 
of the earlier work is the identification of same gross contrasts which 
can help inform observations. For example, ask: "Can you tell if your 
wife is talking to the baby or to your older son, even if you can't see 
them?" "Can you tell if she is talking to & man or to a woman?" These 
questions are based on the assumption that age and sex of addressee are 
likely to be factors influencing speech in most societies. (See the 
sample questionnaire in Section 1.6 for further instances.) 


The data obtained from questionnaires of this sort have value (1) 
to help identify some social comparisons one must be sure to include 
in observations; and (2) to give indications of the stereotypes about 


social variations in speech. 


We know that such reports by informants give information about 
only a small fraction of actual variation in speech. However, they 
will suggest some starting points for observation. Formal observations 
should include following the same speaker through a variety of situa- 
tions in which there are different topics, different settings, and dif- 
ferent addressees, as suggested in Section 1.5.1. One should be sure to 
include the comparisons suggested by the interview comments as so marked 


that even native speakers notice them. 


Thus informant work and observation are both places to begin study- 
ing speech variation. In the next phase of the study, informants can 
act as listeners. Two tape recorders are linked by a patch cord to 
allow recording on a new tape of selections from the field recording. 
Then the natural recordings are edited to remove topical clues, proper 
names, and references which can reveal the speech situation. The in- 
formant should listen to the tape in order to try to identify the set- 
ting and addressee. He may also be able to specify when he could make 
the identification and may even notice what the marker was. 


Tapes successfully recognized by adults can be played to children 
of appropriate characteristics. In this way one can find when children 
learn to recognize differences in speech in different settings or to 
aifferent addressees, even if they do not themselves switch styles. 
(See Section 3.5.1, The Acquisition of Usage Patterns. ) 


The next task is to identify the linguistic differences between 
these tapes. In part this may be done efter the study is completed, 
with the aid of a linguist. However, there are some advantages to 
working in the field on the task of identifying distinctive linguistic 
markers. For example, removing from a tape all but a single linguistic 
difference can be used as a test for which markers are most important. 


2.1.2. Style vs. Speaker 


Contrastive analysis is possible for the speech of two different 
speakers or for the speech of the same speaker in two different situa- 
tions. There are some situations which allow for style switching (or 


for language switching) and therefore these situations are not appropri- 
ate for definitional purposes. If one knows the culture well enough, 
one may be able to predict the circumstances under which even this 
switching will occur. 


Both types of contrast will prove relevant in this study. For 
example, we want to know whether men and women have different speech, 
since if they do we will want to be able to find out how children's 
speech relates to the two norms and how the differentiation is rein- 
forced. But we also want to know if men speak differently when ad- 
dressing men and when addressing women, and whether these differences 
show up in children's speech. The first is a contrast between speakers, 
the second between styles. 


While the same types of contrasts can exist in both of these cases, 
there is a difference between them. The different styles co-exist in a 
single speaker, and there are therefore some pressures towards merger 
of systems. For example, at the phonetic, functional, and semantic 
level, it is difficult for a single speaker to maintain two sets of 
categories that are nearly identical but not quite. The two tend to 
merge. In the extreme case, as Gumperz (1967) has found, the differ- 
ences may be extremely minor, for bilinguals who have spoken with other 
bilinguals over the centuries. 


Thus contrasts may be relatively few for style differences. The 
two systems tend to have a common core, with merely a few variables 
which mark style switching. ‘The contrast between different speakers 
will tend to be reduced in a similar way, if the speakers must inter- 
act with each other often. For instance, men's and women's speech must 
be similar in many ways since each must understand the other. They 
must have a common core grammer since the rules for production and com- 
prehension exist in the same persons and there are no"monolinguals." 

In such cases there may merely be different phonetic realization rules 
or other low-level differences. 


Where some groups occupy desirable statuses and there is emulation, 
style shifts may be related to speaker differences. Thus Labov (1966) 
has suggested three levels of relationship depending on conscious aware- 
ness of group differences and whether style shifts are apparent. In- 
dicators are contrasted features which are unconscious, and define group 
membership but are not used i style shifting. Markers do influence 
style shifting so that in the more conscious and careful style the 
normative version is used. Stereotypes are conscious and subject to 
overt comment and stigma. 


We shall confine our discussion to the last three types of contrast, 
which we must assume can be revealed with topic and function held con- 
stant. 


2.1.3. Functional Realization 


In American society, 
formulated depending on the re 
the request is considered part 


extra. Direct imperatives tend to be 
ample. A simple way to obtain such data in another society is to make 


some simple categories like ‘bringing objects," “supplying simple in- 
formation." One begins to see fairly quickly 


ference in the formula used, according to the speaker or according to 
the setting or addressee. Requests are the easiest function to identify, 


since they are defined as any speech resulting in action on behalf of 
the speaker by the other party. Greetings and other routines would also 
be easy to study. (See Section 1.5.4, Communicative Routines. ) 


2.1.4. Linguistic Contrasts 


The purpose of 4 linguistic contrastive analysis is to define as 
variables those things which are not alike and to characterize all else 
as alike. It is obviously impossible to do a complete linguistic an- 
alysis, but any category described for one system should be described 
for the other. It is important to try to locate the contrasts that are 
sufficiently frequent in texts to permit their practical use &s cri- 
teria for identifying shifting in children. Each instance should be 
analyzed for its systematic implications, not given merely as an isolated 
example, and in each case the equivalent rule in the other system should 
be analyzed. Enough phonetic detail should be given to define phonetic 
realization rules, where there are likely to be many style variables. 


2.1.4.1. Phonological Contrasts 


The easiest clue to a difference in the phonological systems is 
comparison of realization rules for the same morphemes. Clever intro- 


duction of topics while eliciting comparison tapes can help in locating 
the same morphemes in contrasted texts. Otherwise, one may have to make 


judgments that the morphemes are sufficiently similer in phonemic struc- 
ture to permit one to reach conclusions without having the identical 


forms. 


1) Two systems may differ in the phonetic features they use. One 
system may have a set of palatalized consonants but the other not, so 
that in the simpler system the other's 7 talized forms" are homonymous 
with the non-palatalized. In "Diglossia" (1959), Ferguson describes 
eases in high and low styles which differ in that the low system is sin- 
pler, and includes 4 subset of the phonemes in the high system. 


2) One system may have @ special set of allophonic changes which 
the other does not have. These may introduce sounds or features not 
present in the other. 


a AUD a ei bed Bo 


ee ei A A ES A A ORT A A A 


3) A particular item may be present phonemically in one system 
and not in the other, as in the contrast between English dialects with 
and without post-vocalic /r/. 


4) A part of the system may coalesce in one system but not in the 
other. For example in New York City Labov (1966) noted that the vowel 
systems employed different contrasted categories which in a different 
style were coalesced. In a similar way, the distinction between alveolar 
stops and fricatives tended to disappear in spontaneous speech. In 
these cases the variables must be measured since there is not likely to 
be sharp shifting but increases and decreases in the likelihood of making 
the distinctions that exist in only one of the systems. Casual speech, 
in which the fewest distinctions were made, was more systematic and con- 
sistent than styles approaching "standard English." 


5) Phonetic realization of shared phonemes is usually different 
in different styles. In baby talk in English there is palatalization 
of many consonants, in comparison to the normal style. (These may be 
features added to specific items describable segmentally or my be 
long components added to a whole sequence, but tape analysis is neces- 
sary to find out which is the case.) 


6) Distribution according to syllable location may be different; 
e.g. Labov describes an English dialect in which /s/ + stop is not per- 
mitted in syllable final position. Also, though /6/ and /£/ and their 
voiced counterparts are distinctive in other positions, they are both 
realized as labiodentals in final position. 


7) Morpheme formation canons may differ so that when morphemes 
are adapted to the other system their structure mast be changed. For 
example, English words integrated into Japanese bilinguals’ speech may 
have added vowels to form CV syllables. 


8) Phonemic realization of the same morpheme may be different, 
e.g. English "ing" may be realized variously as /in/, /in/, and /iyn/ 
in different styles, the second two being formal. 


2.1.4.2. Lexical and Morphological Contrasts 


Morphological contrasts are obvious if high-frequency inflectional 
morphemes are different in the two systems. Stewart (in Shuy, 1964) 
describes such contrasts between two systems of English inflection. 

For example, "I see it (today)" vs. "I see it (yesterdey)." This 
system may differ either at the morphological or at the semantic level 
from standard English which contrasts present and preterite. However, 
the presence in the same system of "I don't see it (today)" and "I ain't 
see it (yesterday)" indicates that there is an underlying contrast in 
categories which simply is neutralized in the affirmative. 


When there are lexical differences it is necessary to be assured 
that the base meaning for the two forms is similer. Examples in English 
are "dyad" and "couple," "policeman" and "fuzz," "children" and "kids." 


eee ey ye — 


Note that there may be some connotative differences due to the partici- 
pation of these forms in systems which differ in connotations, e.g. in 
attitudes toward police. Whenever in another culture you find alterna- 
tive words with "the same meanings" look for stylistic or speaker differ-~- 
ences restricting selection. 


2.1.4.3. Syntactic Contrasts 


One strategy in identifying syntactic contrasts is to get an infor- 
mant to translate from one system into the other, and then see if this 
translation can be reduced to a word-for-word translation which a third 
informant will accept (Gumperz, 1967). If there is difficulty in trans- 
lation from one system to the other, which may be the case with style 
shifting, then you might be able to rewrite the text making morphologi- 
cal substitutions and, using an informant, see if it is possible to make 
an acceptable text from this word-for-word translation you have made. 
The measure of syntactic difference between two systems is the number 
of order changes and other syntactic rules you need to produce an ac- 
ceptable translation from the word-for-word text. 


An example of a syntactic difference of sufficiently high fre- 
quency to be useful is the contrast between English dialects requiring 
subject-verb agreement in number and not having this agreement. An- 
other difference, though less observable because rarer, is between 
systems which have a past and future perfect and those which realize 
this semantic difference by the use of adverbials and clauses. 


Two styles or two dialects of languages may differ in semantic 
categories, in semological features cross-cutting categories, and so 
on. It is important to distinguish between three easily confused forms 
of contrast. If we compare speech to adults and to children we might 
find that there were different topics selected from the range available 
to the speaker. If a given topic is selected which is not usual for 
the speech of an adult to a child, the adult may have to make semantic 
distinctions which are not available in the system of speech to children. 
Either he tries to teach the distinction or must avoid this feature of 
the topic. Finally, there may be a distinction present which is 
characterized by different lexical selections. He might use the term 
"acquisition" to an adult in making the same semantic distinction he 
makes in talking about "getting" or "buying" to a child. 


There may be differences in the semantic systems which underlie 
grammatical or morphological rules in two systems. For instance, in 
Stewart's analysis of "basilect" (in Shuy, 1964) there is a contrast 
between a simple perfect and a completive perfect, as realized in we 
seen it" and "I been seen it." 


2.1.5. Vocal Signals 


Vocal signals may be noted for two reasons. One is that they may 
constitute "channel cues" that there is a shift of style taking place 
(Labov, 1966). If there is a sudden increase in speed, fluency, a 


67 


change in pitch, there may be an indication that a shift has occurred. 
Usually some topical and linguistic variables will also change about 
the same time. 


Another reason for noting vocal signals is that they may then- 
selves constitute systems which change when style shifts. For example, 
syllable dynamics and voice quality may change in two dialects. Pause 
markers differ--e.g. unh in English, este in Mexican Spanish, palatal- 
ized nasal in Russian. Falsetto is used for particular purposes in some 
types of Negro speech. Special combinations of prosodic rules may have 
particular semantic significance in one system and not in the other-- 
e.g. 2@ sudden downward shift in pitch at a stressed adjective in a sen- 
tence in standard English implies irony. 


The most exhaustive list of vocal signals is given in Pittenger, 
Hockett, and Danehy (1960). We include these simply as reminders of 
items to listen for. 


Quality changes which may differ from the norm in being above or 
below it, utilizing a wider or narrower range of variation than is 
usual, include loudness, pitch, and rate, including drawling and clip- 
ping. For example, speech to babies my have a wider range of pitch 
variation than to older childreN,as well as generally higher pitch. 


Voice qualities which have been noted include rasp or metallic 
quality, glissando or slurring of pitch changes, breaking of glottal 
tone, as in incipient crying, breathiness, nasality, resonance, ex- 
treme vocal openness, slurring. Momentary effects which could be used 
for stylistic purposes are inhalation or exhalation, hissing, clicking, 


sighs, silence, yawming, crying, laughing, whispering, and so on. 


One of the great difficulties in previous use of these vocal sig- 
nals in research is that since many of them do not have the discrete 
signaling function of the ordinary speech channel, there has been no 
clear guide to when a "complete" analysis has occurred. The emphasis 
on contrast puts some inherent limits on what it is practical to in- 
clude in the present study. We suggest that investigators who note 
some overall differences between two styles in any of these features 
try to locate the description in one of the published lists. A tape 
is available for the Pittenger, Hockett, and Danehy text permitting 
calibration against their categories. 


2.1.6. Intelligibility 


In @ commnity where children are exposed to a varied set of lin- 
guistic norms, it is important to find how early children can under- 
stand varieties they hear. For example, children on the Navaho reserva- 
tion go to the curing sings from infancy on; by school age they have 
control over a surprising repertoire of chants. On the other hand, if 
asked to interpret the text of a chant, they might fail, because of its 
special lexicon. : 


2.1.6.1. Glossing a Text 


- A traditional method for assessing intelligibility is to ask an 
informant to provide a translation or explanation of a sample of the 
speech variety. One could probably make gross differentiations, but 
it might be difficult to evaluate the adequacy of two accounts of this 
kind. 


2.1.6.2. Repetition 


Repeating or imitating a tape is a useful test if there are codes 
with syntactical differences and the speakers in each group do not use 
each other's speech forms. For example, Labov has asked Harlem adoles- 
cents to repeat "He asked if I could go to the game today," and con- 
sistently received as an accurate imitation "He asked could I go to 
the game today." In this case the informant translated from one diae- 
lect to the other, keeping the meaning. Note that asking informants in 
this case to provide a gloss would seem strange. 


2.1.6.3. "Cloze" 


On a tape, one can excise every nth item, and ask informants to 
guess what was said. (This is called the "Cloze" technique. Refer- 
ences to its use can be found in Ervin-Tripp & Slobin, 1966, pp. M-5.) 
This seems to an informant like trying to listen through noise. If the 
original and reconstructed texts are compared, the similarities can be 
counted objectively. For example, Peisach (1965) used the method for 
first and fifth grade children varying in race and socioeconomic status. 
For the first graders she omitted the last word in each sentence. She 
found that by the fifth grade the social classes differed only when IQ 
was not controlled, when their guessing of the teacher's speech was 
tested. 


There are several precautions in the use of Cloze. Elaborated 
speech may be very hard to reconstruct, even for someone close to the 
speaker. Since different values of n or a different location of 
omission can have a pronounced effect, it is better to compare infor- 
mants listening to the same tape than to compare tapes or styles. For 
instance, once a tape is prepared, one could compare children of dif- 
ferent ages. Or one could compare children of different age or other 
status. 


All of these methods could also be used across languages in mlti- 
lingual commnities. For studies of multilingual variety, as well as 
dialectal variety, one should remember that the prestige of the spoken 
variety may markedly influence the intelligibility to the listener and 
their willingness to repeat it. 


2.1.7. Evaluation 


Both Lambert and Labov have used subjective evaluation tests to 
determine unconscious attitudes towards speech varieties. In a series 


eet 


| 2 ~~» 


69 


of studies of a variety of bilingual commmities, Lambert and his col- 
leagues (Anisfeld & Lambert, 1964; Lambert, 1967; Lambert et al., 1966) 
have found that the same voice was evaluated differently depending on 
the phonological system it used. Thus Canadian judges rated a man with 
an English Canadian speech guise as more intelligent and competent than 
the seme man in a French Canadian guise. 


Labov (1966) has used a similar method in studying the evaluation 
of speech variables within a monolingual community. He constructed 
passages containing the forms that had the phonological’ indicators in 
which he was interested. Then he took carefully selected sentences 
from a test group which repeated these passages. He selected the sen- 
tences to provide examples of given values of each speech variable. 

Then listeners were asked to listen to each sentence and rate the 
speaker. The ratings were rather more practical than the traits judged 
by Lambert's informants. For instance the rater was asked to be an 
employer and judge the highest job the speaker could hold. He found 
that in New York City the use of a dental fricative in thing raised a 
speaker's judged job suitability but lowered his toughness or masculin- 
ity. | 


When are these values acquired? Lembert found that they showed 
up by 12 in the inter-language situation. Labov, dealing with the com- 
plexity of urban dialects, found changes ‘still taking place up until 
20. Presumably formal training institutions could have a marked im- 
mediate effect on attitudes towards speech varieties. Both Labov and 
Lambert have found that values about speech variation can be shared 
even between speakers who themselves speak quite differently. 


If there is any form of speech stratification in the community 
studied, it would be of value to prepare a stimulus tape for evaluation 
studies with children. If possible, talented speakers who can "shift" 
are the best for making such recordings. In a literate community, avoid 
the use of reading for getting comparable texts. If necessary, get 
repetitions of oral material for control over content; reading intro- 
duces a whole new set of artificialities into speech. 


———————ll—E>—EE 


ee te ef 


2.2. RECORDING APPARATUS AND TECHNIQUES 


2.2.1. Apparatus 


Tape recorders come in different sizes, price ranges, qualities, 
and types. If the field situation is in a city it is then possible 
and even advantageous to use a heavier machine operating on house cur- 
rent in order to record stereophonically, particularly if more than 
one person is being recorded at a time. On the other hand, outside 
of a city it will be necessary to use a battery-operated tape recorder 
that is as small and light as possible without sacrificing recording 
fidelity unduly. This will rule out stero for a couple of years at 
least. A number of good machines are currently on the market, such 
as those put out by Uher, Tandberg, Nagra, Sony, and Ampex. In order 
to simplify matters the Uher 4000 Report L and the Uher 4000 Report S 
are here recommended for field situations in which house current is 
not available during recording. Notwithstanding the fact that some 
claim it to be less rugged than might be desirable, the Uher 4000 com- 
bines such features as four speeds, battery or wall operation, good 
fidelity, quality workmanship, and reasonable cost so as to leave it 
with no real competitor. Although the Nagra surpasses it in fidelity, 
the Nagra has only one speed and a very high price. The battery- 
operated Sony lacks the fidelity and the versatility of four speeds 
as well as reliability of its electronic components, although when 
price is considered, the fidelity is very high indeed, and suitable for 


many purposes. 


It is advantageous, almost imperative, to bring two tape recorders 
into the field, preferably of the same make and model. In the first 
place if one breaks down the other can be used, and if both break down, 
@ judicious combination of parts from each may yield one serviceable 
machine. “In the second place, patching (recording from one machine to 
another ) necessitates two tape recorders. Patching is used for editing 
tapes, re-recording at a different speed, or re-recording on a tape of 
a different thickness. In addition two recorders can be used for pro- 
ducing a stimulus on one recorder and recording both the stimulus and 
the listener's response on the other. 


Transport 


The means by which the tape is transported across the heads from 
one reel to another is important for fidelity of reproduction. The 
transport mechanism consists of one or more electric motors and a band 
or rubber idler and capstan assembly to transfer the rotation from the 


Section 2.2 was written by Brian Stross. (Additional advice on tape 
recording can be found in Samarin, 1967, pp. 88-102.) 


motor to the tape reels. It is important that the transport pulls the 
tape across the heads evenly and smoothly at an invariable speed. Var- 
jations in speed that produce unwanted distortions are called wow (slow 


petter. 


It is possible to purchase accessories for the testing of speed 
accuracy, but these are usually not feasible for the field. One solu- 
tion is to blow a 440-A pitchpipe before and after every recording. In 
playing back the recording, the pitch on the tape should match the 
pitch of any 440-A pitchpipe. This procedure should also allow one to 
tell when the batteries are beginning to run down. The pitch of a re~- 
cording made when the batteries were known to be in good condition 
should at any time match the pitchpipe. A lower pitch means that the 
transport is going slower ané the batteries may not be up to charge. 

A higher pitch means that the tape speed is faster than normal in 
which case it is likely that some tape has wound itself around the 
capstan (probably because the plastic tips that hold the tape reels in 
place were not given the half turn necessary to hold the tape in posi- 
tion while the recorder is moved about). When this happens, the tape 
wound around the capstan should be cut out or pulled off the capstan. 


Heads 


The heads are among the most critical features on a tape recorder. 
They are tiny electromagnets that put material on the tape, pick up 
recorded material to play it back, and erase tape. Most recorders un- 
der consideration for fieldwork will have either quarter track or half- 
track heads. The quarter-track machines double the playing and record- 
ing time that would be given by a half track machine for the same 
length of tape. On the other hand some fidelity is sacrificed by add- 
ing tracks in that the signal to noise ratio is smaller on the quarter 
track machines. The Uher is a half track. 


Generally the quality of a recorder with three heads, one each 
for recording, playing, and erasing, is better than one with only two 
in which recording and playing are both done with a single head. The 
compensation of the two-head machine is that it is easier to align the 
head. 


Volume Level Indicator 

Input must be adjusted to the right level with the recording volume 
knob in order to record the most efficiently. If the level is too lo, : 
the signal will be weak and tape hiss plus other noise will drown out | 
cme of the signal. If the level is too high then the signal will be ' 
distorted, a problem which is just as serious. There is one point at 
which the signal to noise ratio is optimal. This point is indicated on 
the VU meter, a volume level indicator, and is as close as possible to | 
the red zone (the red zone being that area on the meter in which the 4 


numbers have plus rather than minus values) without being in it. The 

meter needle, which is moving with the input signal, must be kept as — 

: close as possible to this optimm point at all times. However the 

: needle should never enter the red zone. Sometimes, when there are wide 
variations in the volume of the input, and when it is possible to some- 

what predict these fluctuations, it is a good idea to adjust the volume 
level by hand during recording to compensate for the variation while 


maintaining a strong enough signal. 
| Digital Counter 


This is a device very similar to the odometer on a car. It con- 
sists of three rows of numbers that count the amount of tape that has 
gone by the heads. In order to rapidly locate any given point on the 
tape, the counter must have been set to 000 at the beginning of the. 
tape, and the number of the desired location jotted down, perhaps with 
a description of what is to be found at this point. Then the beginning 
of the tape should again correspond to a setting of 000 and the tape 
rapidly wound until the right number shows on the face of the counter. 
Another way of finding a certain spot quickly is to place a tiny strip 
of paper between the layers of tape at this spot as the tape is being 
rewound. 


Tape Speeds 


One tape recorder will offer a variety of speeds. The speeds are 

in inches per second. The higher the speed the better the fidelity, 
but also the more tape is needed to record for the same length of time. 
Varying situations and varying needs call for different speeds. When 

| working in the researcher's native language where the interest is only 
in content, 15/16 has the tremendous advantage that the tape will run 
eight times as long as it would at the high fidelity speed of 7-1/2. On 

the other hand the fidelity at this speed is wholly inadequate for 
phonological work which in most cases would require 7-1/2 or faster. 
Editing is another consideration. It is easier to edit a tape recorded 
at 7-1/2 than one recorded at slower speeds. It is far better to sacri- 

| fice the tape in field recording under almost every conceivable condi- 
tion except that of tape shortage. It will be difficult enough to tran- 
scribe the tape anyway, and every bit of fidelity will help make the 

| job easier. In the interest of economy the Uher machine at 3-3/4 ips. 
can still be expected to yield reasonable good quality for voice and 
will probably be the speed most frequently used. 


Recorder Inputs 


. The tape recorder will have at least two inputs; one for high level 
sources and the other for microphones (low level sources). When patch= 

1 ing, use the high level input as described in the manual that comes with 
L the recorder. The microphone input can be high impedance (5,000 - 50,000 

} ohms) or low impedance (50 - 250 ohms). The microphone impedance should 
) match the impedance of the input. A low impedance microphone and input 


Th 


will allow a longer microphone cable to be used, but due to the lower 
voltage output requires 4 transformer to step it up. The recorder may 
have a built-in transformer, or @ socket for plugging in an accessory 
one. The high impedance microphone and input socket obviates the need 
for a transformer, but necessitates a shorter lead cable. Above all, 
when buying a microphone, make sure that the impedance matches or is 
made to match that of the recorder input, in terms of high or low. 


Recorder Outputs 


There should be a separate output jack on the tape recorder built 
for high impedance or for low impedance headphones. This can be used 
for monitoring during recording as well as for playback if the recorder 
has been wired accordingly, as has the Uher. The Uher output jack is 
low impedance. 


Tape 


Tape consists of a thin jron oxide coating on a plastic base. | 
Acetate tape has only economy to recommend it. Mylar is more expensive, 
but is superior in almost every way including durability and humidity 
resistance. The five-inch reel will accommodate up to 1800 feet of 
the thinnest mylar (.5 mil) which works well with the Uher. Some ma- 
chines are less delicate with the tape than is the Uher, and for these 
1 or 1.5 mil tape is recommended. The more feet to the reel, the 
longer the recording time without switching reels, but also the more 
delicate the tape and the more likely is the problem of print-through 
during storage (print-through being a pre-echo of louder passages that 
can be heard during softer passages during playback). 


Tape also comes as either regular or as high output. This refers 
to the signal output potentiality without distortion during playback. 
High output tape can play louder without distortion. This advantage 
should be weighed against a higher cost, in deciding which to use. 
There are several good brands of tape and a higher price usually cor- 
responds to better and more consistent quality brands. Don't buy 
acetate, and don't buy tape that is cheap or on sale. This writer 
recommends Scotch 290 or BASF 1800° either of which should cost about 
$4.25 per 5-inch reel when gotten at a discount. 


It would be a good idea to bring several extra empty take-up reels 
into the field. They provide a good way of siphoning off recordings for 
filing, for making available special sections of tape, and make it pos- 
sible to always go into a recording session with a completely empty 
tape. Twenty or so of these reels would not be too many. 


Microphones 


The components of a tape recorder can be thought of as a chain. 
The finished product is only as good as its weakest link. The weakest 
link of most tape recorders is the microphone that comes with it. Even 


79 


the Uher recorder comes equipped with an inadequately performing mike. 
Microphone prices range from about $10 for a cheap crystal mike to about 
$200 for a professional quality capacitor mike. The best bet for field 
recording is a dynamic microphone costing from $50 to $100 (e.g. an RCA 
BK12-A or an Electro-Voice 654A) of a lavalier type able to be hung 
around the neck. ‘The ribbon microphone is a very responsive mike to 
all of the hearable frequencies, but is not quite as durable as the 
standard dynamic type. (The usual crystal mike is notoriously sen- 
sitive to heat and humidity changes. Silica gel packing for the crystal 
may help in this regard. The problem can be better avoided by using a 


dynamic mike. ) 


Dynamic microphones come with varying degrees of directionality. 
An omnidirectional mike is one that picks up sounds equally from any 
direction. A directional microphone usually called ecardioid because 
of its heart-shaped pickup pattern, will not pick up sounds from the 
rear, and is relatively insensitive on the sides. Directionality in a 
microphone allows the mike to be placed at a greater distance from the 
subject while keeping background noise at a minimum. Any situation of 
high background noise, such as outdoors, suggests the use of a direc- 
tional mike. An omnidirectional mike can be used when background sound 
is wanted, in a quiet situation, or if the mike is placed closer to the 
subject. Stereo recording, for the best effect, necessitates cardioid 
mikes. ‘The Shure 545L Unidyne III is an ideal cardioid mike for the 
field. If only an omnidirectional mike is available, some amount of 
directionality can be gained from using a rolled up magazine and aiming 
at the sound source. Care must be taken not to scrape the mike with 
the magazine or unwanted popping and other noise will be recorded. 


Some directional mikes are now being sold that are so highly di- 
rectional that whispers can be picked up a block away. These are 
highly specialized, demanding close investigation before purchase. 


If money allows, a good solution to the microphone problem is to 
purchase more than one microphone including one omnidirectional mike 
and one cardioid mike. 


When recording is done outdoors, the microphone should have a wind- 
sereen attached. A handkerchief can serve as a makeshift windscreen if 
necessary. 


When recording is done indoors and a lavalier mike is not used, a 
stand is advisable. Placing a microphone directly on a table allows 
the pickup of all kinds of unwanted vibrations and other sounds that can- 
not be heard until the recording is played back. This is especially 
true if the tape recorder is on the table also. 


Mixer 


When recording with more than one microphone a simple Y connection 
is not really adequate unless the mikes are all the same make and model. 
In order to blend the inputs of the microphones to achieve any desired 


76 


balance, a mixer is necessary. On stereo machines, of course, @ mixer 
is only needed for more than two microphones. 


Headphones 


Because much time will be spent wearing headphones for listening 
to and transcribing tape recordings, headphones must be comfortable 
and must provide as good fidelity reproduction as possible. This can 
be done for between $25 and $50 (the Koss SP-3X for $25 is an excellent 
low impedance headphone that can be adapted for the Uher input with no 
trouble and the SP-*SM by Koss is a good versatile Stereo/Mono headset). 
Headphones are used for listening to recordings rather than the speakers 
built into the recorder, because headphones can be padded against out- 
side noise, because they do not disturb other people, and because of 
the incomparable difference in fidelity. Headphones should be as light- 
weight as possible, rugged, and nicely cushioned, in addition to the 
high fidelity requirements. 


Foot Controls 


The Uher and many other tape recorders can be optionally supplied 
with accessory foot controls for starting and stopping. This apparatus 
is necessary for translation and transcription of tape recordings. 


Batteri
…[truncated]