psych questions

The Development of Language-Like Communication Without a Language Model

Author(s): Susan Goldin-Meadow and Heidi Feldman

Source: Science , Jul. 22, 1977, New Series, Vol. 197, No. 4301 (Jul. 22, 1977), pp. 401-403

Published by: American Association for the Advancement of Science

Stable URL: https://www.jstor.org/stable/1744359

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science

This content downloaded from �������������152.13.18.10 on Tue, 24 Nov 2020 07:44:53 UTC��������������

All use subject to https://about.jstor.org/terms

enough to act as predators). In the ab- sence of evidence from other sources, it seems most likely that the ocellus func- tions as a component of Batesian mim- icry rather than as a deflective target, but it may also startle a predator.

On the basis of field and aquarium ob- servations, it is apparent that, when threatened, most reef-fish prey species take shelter in the reef and await the

eventual departure of the predator. What then would be the selective advantage to a prey species to pose in a vulnerable lo- cation rather than to flee and hide? The

strategy of the mimic appears to be one of intimidation. Rather than flee into the

refuge of the reef when it encounters a predator, Calloplesiops simulates the abundant and aggresive moray, fright- ening away a predator and thereby reduc- ing the time spent by Calloplesiops in less productive activities.

JOHN E. MCCOSKER

Steinhart Aquarium, California Academy of Sciences, San Francisco 94118

References and Notes

1. W. Wickler, Mimicry in Plants and Animals (McGraw-Hill, New York, 1968); J. Randall and H. Randall, Bull. Mar. Sci. Gulf Caribb. 10, 444 (1960); J. Randall and A. Emery, Zoologica 56, 115 (1971).

2. Aquarium observations are based on six speci- mens, presumably collected in the Philippines, of Calloplesiops that have been in captivity since December 1973. I made field observations

enough to act as predators). In the ab- sence of evidence from other sources, it seems most likely that the ocellus func- tions as a component of Batesian mim- icry rather than as a deflective target, but it may also startle a predator.

On the basis of field and aquarium ob- servations, it is apparent that, when threatened, most reef-fish prey species take shelter in the reef and await the

eventual departure of the predator. What then would be the selective advantage to a prey species to pose in a vulnerable lo- cation rather than to flee and hide? The

strategy of the mimic appears to be one of intimidation. Rather than flee into the

refuge of the reef when it encounters a predator, Calloplesiops simulates the abundant and aggresive moray, fright- ening away a predator and thereby reduc- ing the time spent by Calloplesiops in less productive activities.

JOHN E. MCCOSKER

Steinhart Aquarium, California Academy of Sciences, San Francisco 94118

References and Notes

1. W. Wickler, Mimicry in Plants and Animals (McGraw-Hill, New York, 1968); J. Randall and H. Randall, Bull. Mar. Sci. Gulf Caribb. 10, 444 (1960); J. Randall and A. Emery, Zoologica 56, 115 (1971).

2. Aquarium observations are based on six speci- mens, presumably collected in the Philippines, of Calloplesiops that have been in captivity since December 1973. I made field observations

rather than of the caretakers.

Must a child experience language in order to learn language? Clearly some experience with language is necessary for the child to learn the established lan-

guage of his particular community. The child of English-speaking parents learns English and not Hopi, while the child of Hopi-speaking parents learns Hopi, not English. But what if a child is exposed to no conventional language at all? Surely such a child, lacking a specific model to imitate, could not learn the conventional

language of his culture. But might he elaborate a structured, albeit idiosyn- cratic, language nevertheless?

We have observed a group of children who lack specific linquistic input but who otherwise have normal home envi-

ronments. Our subjects are deaf children

22 JULY 1977

rather than of the caretakers.

Must a child experience language in order to learn language? Clearly some experience with language is necessary for the child to learn the established lan-

guage of his particular community. The child of English-speaking parents learns English and not Hopi, while the child of Hopi-speaking parents learns Hopi, not English. But what if a child is exposed to no conventional language at all? Surely such a child, lacking a specific model to imitate, could not learn the conventional

language of his culture. But might he elaborate a structured, albeit idiosyn- cratic, language nevertheless?

We have observed a group of children who lack specific linquistic input but who otherwise have normal home envi-

ronments. Our subjects are deaf children

22 JULY 1977

along the western coast of Grande Comore Is- land, Indian Ocean, during February and March 1975.

3. Calloplesiops altivelis (Steindachner) was de- scribed as Plesiops altivelis and includes Bar- rosia barrosi Smith in its synonymy.

4. E. Hobson, Fish. Bull. 72, 915 (1974). 5. R. Hiatt and D. Strasburg, Ecol. Monogr. 30, 65

(1960). 6. C. Rettenmeyer, Annu. Rev. Entomol. 15, 43

(1970). 7. The necessity and importance of certain of the

listed characters, particularly items (iii) and (iv), is debatable.

8. Miillerian mimicry is based on the premises that (i) two or more species are unpalatable, (ii) if two or more species are indistinguishable by predators, they will be captured in proportion to their abundance, and items (iv) to (vi) of the Batesian mimicry criteria (6).

9. Because of the scarcity of Calloplesiops, the simple but conclusive experiment of feeding a series of Calloplesiops to various predators was not attempted.

10. J. E. Randall, K. Aida, T. Hibiya, N. Mitsuura, H. Kamiya, and Y. Hashimoto [Publ. Seto Mar. Biol. Lab. 19, 157 (1971)] outlined the procedure for identifying the presence of skin toxins and discovered them to be present throughout the family Grammistidae. Ichthyologists consider the Grammistidae and Plesiopidae to be related families within the suborder Percoidei [see, for example, P. H. Greenwood, D. E. Rosen, S. Weitzman, G. S. Myers, Bull. Am. Mus. Nat. Hist. 131, 341 (1966)].

11. E. Poulton, The Colours of Animals (Interna- tional Scientific Series, LXVIII, London, 1890); H. Cott, Adaptive Coloration in Animals (Meuthen, London, 1940).

12. A. Blest, Behaviour 11, 209 (1957). 13. Six specimens from Steinhart Aquarium and

preserved specimens from the fish collections of the California Academy of Sciences, National Museum of Natural History, and J. L. B. Smith Institute of Ichthyology.

14. A. Blest, Zoologica 49, 161 (1964). 15. Comoran fieldwork supported by a grant from

the C. H. Breeden Foundation. I thank R. Mac- Pherson for insight, P. Ehrlich and G. Barlow for advice, and T. McHugh/Photo Researchers and D. Powell for photographs.

2 November 1976; revised 28 December 1976

along the western coast of Grande Comore Is- land, Indian Ocean, during February and March 1975.

3. Calloplesiops altivelis (Steindachner) was de- scribed as Plesiops altivelis and includes Bar- rosia barrosi Smith in its synonymy.

4. E. Hobson, Fish. Bull. 72, 915 (1974). 5. R. Hiatt and D. Strasburg, Ecol. Monogr. 30, 65

(1960). 6. C. Rettenmeyer, Annu. Rev. Entomol. 15, 43

(1970). 7. The necessity and importance of certain of the

listed characters, particularly items (iii) and (iv), is debatable.

8. Miillerian mimicry is based on the premises that (i) two or more species are unpalatable, (ii) if two or more species are indistinguishable by predators, they will be captured in proportion to their abundance, and items (iv) to (vi) of the Batesian mimicry criteria (6).

9. Because of the scarcity of Calloplesiops, the simple but conclusive experiment of feeding a series of Calloplesiops to various predators was not attempted.

10. J. E. Randall, K. Aida, T. Hibiya, N. Mitsuura, H. Kamiya, and Y. Hashimoto [Publ. Seto Mar. Biol. Lab. 19, 157 (1971)] outlined the procedure for identifying the presence of skin toxins and discovered them to be present throughout the family Grammistidae. Ichthyologists consider the Grammistidae and Plesiopidae to be related families within the suborder Percoidei [see, for example, P. H. Greenwood, D. E. Rosen, S. Weitzman, G. S. Myers, Bull. Am. Mus. Nat. Hist. 131, 341 (1966)].

11. E. Poulton, The Colours of Animals (Interna- tional Scientific Series, LXVIII, London, 1890); H. Cott, Adaptive Coloration in Animals (Meuthen, London, 1940).

12. A. Blest, Behaviour 11, 209 (1957). 13. Six specimens from Steinhart Aquarium and

preserved specimens from the fish collections of the California Academy of Sciences, National Museum of Natural History, and J. L. B. Smith Institute of Ichthyology.

14. A. Blest, Zoologica 49, 161 (1964). 15. Comoran fieldwork supported by a grant from

the C. H. Breeden Foundation. I thank R. Mac- Pherson for insight, P. Ehrlich and G. Barlow for advice, and T. McHugh/Photo Researchers and D. Powell for photographs.

2 November 1976; revised 28 December 1976

of normal intelligence whose hearing losses prevent them from acquiring oral language naturally in the home. These children's hearing parents have decided against exposing them to a manual sign language in order to concentrate on oral education (1). At the point at which we studied these subjects, their oral educa- tion program had not produced signifi- cant learning; they had acquired few, if any, spoken-language items that they could use regularly in their daily activi- ties.

Six deaf children of hearing parents (two girls and four boys), ranging in age from 17 to 49 months at the first inter-

view, were visited in their homes by two experimenters for 1 to 2 hours at inter- vals of approximately 6 to 8 weeks. The

of normal intelligence whose hearing losses prevent them from acquiring oral language naturally in the home. These children's hearing parents have decided against exposing them to a manual sign language in order to concentrate on oral education (1). At the point at which we studied these subjects, their oral educa- tion program had not produced signifi- cant learning; they had acquired few, if any, spoken-language items that they could use regularly in their daily activi- ties.

Six deaf children of hearing parents (two girls and four boys), ranging in age from 17 to 49 months at the first inter-

view, were visited in their homes by two experimenters for 1 to 2 hours at inter- vals of approximately 6 to 8 weeks. The

experimenters provided a standard set of toys for the child to play with during the interview and videotaped the informal interaction of mother, experimenter, child, and toys. Each videotaped session was coded by one of the experimenters or a research assistant. Selected samples were coded by both experimenters in or- der to calculate reliability scores on the coding categories.

The videotaped sessions were used to develop a coding system (2). (i) In- stances of communicative gestures were designated in the stream of motor behav- ior (3). In a randomly selected sample of videotape, 82 percent of the gestures identified by either of two coders were identified and similarly described by both coders. (ii) On the basis of physical criteria, these gestures were broken down into single units analogous to words or signs and into multisign units analogous to phrases (4). Of the gestures identified by both coders, there was 95 percent agreement on sign boundary as- signment and 85 percent agreement on phrase boundary assignment. (iii) By the method of "rich interpretation" (5), ref- erential designates (such as Santa Claus or twist) were assigned to all word signs, and semantic elements, cases, and predi- cates (such as agent or act) (6) were as- signed to the individual signs in all multi- sign phrases. Of the gestures identified by both coders, there was 98 percent agreement on reference assignment and 96 percent agreement on semantic ele- ment assignment.

Using these descriptive categories, we found that each of our deaf subjects de- veloped a structured communication system that incorporates properties found in all child languages (7). They de- veloped a lexicon of signs to refer to ob- jects, people, and actions, and they com- bined signs into phrases that express se- mantic relations in an ordered way.

Lexicon. The children developed two types of signs to refer to objects and ac- tions (8). First, they used deictic signs, typically pointing gestures which, like proforms in English (such as "this" or "there"), effectively allow the child to make reference to any object or person in the present. However, as is the case with proforms, context is necessary to interpret these signs. During the study, David, Donald, Dennis, Chris, Kathy, and Tracy produced, respectively, 4854, 1806, 309, 401, 1218, and 366 deictic signs, representing 52, 62, 49, 41, 52, and

experimenters provided a standard set of toys for the child to play with during the interview and videotaped the informal interaction of mother, experimenter, child, and toys. Each videotaped session was coded by one of the experimenters or a research assistant. Selected samples were coded by both experimenters in or- der to calculate reliability scores on the coding categories.

The videotaped sessions were used to develop a coding system (2). (i) In- stances of communicative gestures were designated in the stream of motor behav- ior (3). In a randomly selected sample of videotape, 82 percent of the gestures identified by either of two coders were identified and similarly described by both coders. (ii) On the basis of physical criteria, these gestures were broken down into single units analogous to words or signs and into multisign units analogous to phrases (4). Of the gestures identified by both coders, there was 95 percent agreement on sign boundary as- signment and 85 percent agreement on phrase boundary assignment. (iii) By the method of "rich interpretation" (5), ref- erential designates (such as Santa Claus or twist) were assigned to all word signs, and semantic elements, cases, and predi- cates (such as agent or act) (6) were as- signed to the individual signs in all multi- sign phrases. Of the gestures identified by both coders, there was 98 percent agreement on reference assignment and 96 percent agreement on semantic ele- ment assignment.

Using these descriptive categories, we found that each of our deaf subjects de- veloped a structured communication system that incorporates properties found in all child languages (7). They de- veloped a lexicon of signs to refer to ob- jects, people, and actions, and they com- bined signs into phrases that express se- mantic relations in an ordered way.

Lexicon. The children developed two types of signs to refer to objects and ac- tions (8). First, they used deictic signs, typically pointing gestures which, like proforms in English (such as "this" or "there"), effectively allow the child to make reference to any object or person in the present. However, as is the case with proforms, context is necessary to interpret these signs. During the study, David, Donald, Dennis, Chris, Kathy, and Tracy produced, respectively, 4854, 1806, 309, 401, 1218, and 366 deictic signs, representing 52, 62, 49, 41, 52, and 52 percent of the signs each child pro- duced.

The children produced a second type of sign, characterizing signs, which are motor-iconic signs that specify actions,

401

52 percent of the signs each child pro- duced.

The children produced a second type of sign, characterizing signs, which are motor-iconic signs that specify actions,

401

The Development of Language-Like Communication Without a

Language Model

Abstract. Deaf children who are unable to acquire oral language naturally and who are not exposed to a standard manual language can spontaneously develop a structured sign system that has many of the properties of natural spoken language. This communication system appears to be largely the invention of the child himself

The Development of Language-Like Communication Without a

Language Model

Abstract. Deaf children who are unable to acquire oral language naturally and who are not exposed to a standard manual language can spontaneously develop a structured sign system that has many of the properties of natural spoken language. This communication system appears to be largely the invention of the child himself

This content downloaded from �������������152.13.18.10 on Tue, 24 Nov 2020 07:44:53 UTC��������������

All use subject to https://about.jstor.org/terms

Table 1. Comparison of number of characterizing signs produced during sessions 1 to 4 by mothers and children. Types refers to number of different characterizing signs; tokens refers to number of occurrences across types.

Types Tokens

Subject In In semantic Subject Child Mother com- Child Mother relation phrases

mon Child Mother

David 56 54 18 107 90 47 9 Dennis 25 23 5 50 58 18 3

objects, and, less frequently, attributes. The form of a characterizing sign is re- lated to its referent by apparent physical similarity. For example, a closed fist bobbed in and out near the mouth re-

ferred to a banana or to the act of eating a banana. Two hands flapped up and down at shoulder height referred to a bird or the act of flying. As a result of this motor-iconicity, the characterizing sign is less dependent on context for in- terpretation than is the deictic sign. Da- vid, Donald, Dennis, Chris, Kathy, and Tracy each produced, respectively, 210, 76, 25, 59, 35, and 95 different types of characterizing signs throughout the study.

Syntax and semantics. In addition to these lexical accomplishments, the chil- dren concatenated their deictic and char-

acterizing signs into multisign phrases that conveyed relations between objects and actions. For example, one child pointed at a shoe and then pointed at a table to request that the shoe (patient) be put on the table (recipient). On another occasion, the child pointed at a jar and then produced a twisting motion in the air to comment on mother's having twist- ed open (act) the jar (patient). Another child opened his hand with his palm fac- ing upward and then followed this "give" sign with a point toward his chest, to request that an object be given (act) to him (recipient). The children tended to produce phrases containing combinations of the patient, recipient, and act semantic elements represented in the examples above: David, Donald, Dennis, Chris, Kathy, and Tracy pro- duced, respectively, 156, 64, 22, 23, 22, and 12 such phrases, representing 63, 76, 80, 79, 66, and 50 percent of the action phrases each child produced. Phrases containing the agent or actor element were produced less frequently than phrases with the other three semantic elements, and phrases with place of ac- tion and instrument elements were rarely produced.

Some of the children tended to pro- duce their signs for the patient, recipient, and act semantic elements in consistent

402

positions of their two-sign phrases. Spe- cifically, as exemplified above, the chil- dren tended to produce phrases with patient-act, patient-recipient, and act-re- cipient orders (Fig. 1) (9). Not all chil- dren showed ordering tendencies for all pairs of the three elements; but if the children showed any ordering tendencies at all, those tendencies were ordered in

50

30

10

I iiii

Dennis

20 Donald oU.

20 z

20 l Kathy

Chris

PR RP PA AP AR RA

Sign order

Fig. 1. Number of two-sign phrases classifed according to the order of each element in the phrase. Abbreviations: P, patient, the object or person acted upon; A, act, the action car- ried out to effect a change of either state or location; and R, recipient, the locus or person toward which someone or something moves. Patient signs tended to precede recipient signs (X2=36, P<.001 for David; by the binomial test, P<.03 for Dennis, P<.02 for Donald). Patients tended to precede acts (X2=5.48, P<.02 for David; X2=7.36, P<.01 for Dennis). Acts tended to precede recipients (X2= 13.00, P<.001 for David; X2= 10.28, P<.001 for Don- ald). Subjects were observed over varying pe- riods of time: David was seen from 2 years 10 months to 3 years 10 months for eight ses- sions; Dennis from 2 years 2 months to 2 years 6 months for four sessions; Donald from 2 years 5 months to 4 years 6V? months for 11 sessions; Kathy from 1 year 5 months to 2 years 8 months for nine sessions; and Chris from 3 years 2 months to 3 years 6 months for three sessions.

the same direction. We can describe the

children's two-sign phrases with the fol- lowing element-ordering rule (10):

Rule A:

(choose any two maintaining order)

Phrase -> (patient) (act) (recipient)

Thus, it appears that some of the chil- dren expressed semantic relations in a systematic way, that is, by following a syntactic rule based on the semantic role of each of the sign units.

The children also produced longer phrases that expressed at least two se- mantic relations. David, Donald, Den- nis, Chris, Kathy, and Tracy each pro- duced, respectively, 240, 12, 4, 8, 11, and 10 multirelation phrases, represent- ing 31, 7, 10, 14, 17, and 12 percent of each child's semantic relation phrases. For example, David pointed at a picture of a shovel, pointed downstairs where a shovel was stored, produced a digging motion in the air with two fists, and final-

ly pointed downstairs a second time. Da- vid had commented in one phrase on two aspects of the shovel, the act usually per- formed on the shovel and the habitual lo- cation of the shovel.

The child inventor. A crucial question is whether the deaf children rather than

their caretakers first elaborated these

signed communications. We observed that the children's mothers did use some

gestures. To determine who invented the system, we transcribed the gestures pro- duced by the mothers of two of our sub- jects during the first four interviews. Our impression was that these mothers did not alter their behavior in front of the

camera and that our samples were repre- sentative of the mothers' communication efforts.

A comparison of the mothers' and the children's signs suggests that indeed it was the children who first produced the system. The children showed that they could invent characterizing signs by creating motor-iconic gestures for new stimulus toys they had not previously en- countered. Although the mothers pro- duced as many different types of charac- terizing signs as did their children, only about 25 percent of these signs were common to both mother and child (Table 1, column 1). There is thus some sugges- tion that the mothers' lexical vocabu- laries differed from their children's and that each of the children could invent characterizing signs on his own.

Furthermore, the children produced multisign phrases that conveyed seman- tic relations earlier than their mothers.

Both children produced a number of these phrases in session 1. David's moth- er produced only three such phrases in

SCIENCE, VOL. 197

This content downloaded from �������������152.13.18.10 on Tue, 24 Nov 2020 07:44:53 UTC��������������

All use subject to https://about.jstor.org/terms

session 1 (compared to David's 27 during session 1), and Dennis' mother did not start production at all until session 2. In addition, the children produced many more multisign phrases conveying se- mantic relations than did their mothers. Over the course of the four interviews, David and Dennis produced 127 and 42 such phrases, respectively, while their mothers produced only 41 and 13, re- spectively. There is thus no evidence that the children learned to concatenate

signs to express semantic relations by imitating their mothers' gestures.

Finally, the children were far more likely than were their mothers to use characterizing signs in their multisign phrases. The mothers produced as many characterizing signs in single-unit phrases as their children but far fewer characterizing signs in multisign phrases (Table 1, columns 2 and 3). Con- sequently, there is no indication that the children learned to integrate their char- acterizing signs into an ordered system by imitating their mothers' productions (11).

We have shown that a child can devel-

op a structured communication system in a manual mode without the benefit of

an explicit, conventional language mod- el. This achievement is cast into bold re-

lief by comparison with the meager lin- guistic achievements of chimpanzees. While chimpanzees seem to learn from manual language training (12), they have never been shown to spontaneously de- velop a language-like communication system without such training-even when that chimp is lovingly raised at a human mother's knee (13). On the other hand, even under difficult circum- stances, the human child reveals a natu- ral inclination to develop a structured communication system.

SUSAN GOLDIN-MEADOW

Department of Education, University of Chicago Chicago, Illinois 60637

HEIDI FELDMAN

School of Medicine, University of California, San Diego, La Jolla 92037

References and Notes

1. Deaf children who are orally trained are in- structed in lipreading and in speech production with no audio feedback. These children have been observed to spontaneously gesture to one another "behind the teacher's back." [L. Fant, Ameslan (National Association of the Deaf, Sil-

session 1 (compared to David's 27 during session 1), and Dennis' mother did not start production at all until session 2. In addition, the children produced many more multisign phrases conveying se- mantic relations than did their mothers. Over the course of the four interviews, David and Dennis produced 127 and 42 such phrases, respectively, while their mothers produced only 41 and 13, re- spectively. There is thus no evidence that the children learned to concatenate

signs to express semantic relations by imitating their mothers' gestures.

Finally, the children were far more likely than were their mothers to use characterizing signs in their multisign phrases. The mothers produced as many characterizing signs in single-unit phrases as their children but far fewer characterizing signs in multisign phrases (Table 1, columns 2 and 3). Con- sequently, there is no indication that the children learned to integrate their char- acterizing signs into an ordered system by imitating their mothers' productions (11).

We have shown that a child can devel-

op a structured communication system in a manual mode without the benefit of

an explicit, conventional language mod- el. This achievement is cast into bold re-

lief by comparison with the meager lin- guistic achievements of chimpanzees. While chimpanzees seem to learn from manual language training (12), they have never been shown to spontaneously de- velop a language-like communication system without such training-even when that chimp is lovingly raised at a human mother's knee (13). On the other hand, even under difficult circum- stances, the human child reveals a natu- ral inclination to develop a structured communication system.

SUSAN GOLDIN-MEADOW

Department of Education, University of Chicago Chicago, Illinois 60637

HEIDI FELDMAN

School of Medicine, University of California, San Diego, La Jolla 92037

References and Notes

1. Deaf children who are orally trained are in- structed in lipreading and in speech production with no audio feedback. These children have been observed to spontaneously gesture to one another "behind the teacher's back." [L. Fant, Ameslan (National Association of the Deaf, Sil- ver Spring, Md., 1972); B. T. Tervoort, Am. Ann. Deaf 106, 436 (1961)].

2. A rationale and justification of our coding meth- ods and a more detailed discussion of results are given by H. Feldman, S. Goldin-Meadow, and L. Gleitman [Action, Gesture, and Symbol, A. Lock, Ed. (Academic Press, New York; in press)].

3. Communicative signs were motor behaviors, di- rected to a person, which served no direct func- tion in the setting. The physical form of the signs

22 JULY 1977

ver Spring, Md., 1972); B. T. Tervoort, Am. Ann. Deaf 106, 436 (1961)].

2. A rationale and justification of our coding meth- ods and a more detailed discussion of results are given by H. Feldman, S. Goldin-Meadow, and L. Gleitman [Action, Gesture, and Symbol, A. Lock, Ed. (Academic Press, New York; in press)].

3. Communicative signs were motor behaviors, di- rected to a person, which served no direct func- tion in the setting. The physical form of the signs

22 JULY 1977

was described by a system similar to the one used to describe American Sign Lanuage. The dimensions used in the descriptions are de- scribed by W. C. Stokoe, Jr. [Stud. Linguist. Occas. Pap. 8 (1960)].

4. A detailed account of the criteria for single signs and an account of the lexical data are given by H. Feldman [thesis, University of Pennsylvania (1975)1; the criteria for sign phrases and for the data on syntactic and semantic relations are de- scribed by S. Goldin-Meadow (Stud. Neurolin- guist, in press).

5. A description of the method of rich inter- pretation is given by L. Bloom [Language De- velopment (MIT Press, Cambridge, Mass., 1970); One Word at a Time (Mouton, The Hague, 1973)].

6. The system we use to describe the deaf child's phrases is an adaptation of the case system pre- sented by C. J. Fillmore [in Universals in Lin- guistic Theory, E. Bach and R. T. Harms, Eds. (Holt, Rinehart & Winston, New York, 1968), pp. 1-88].

7. R. Brown, A First Language (Harvard Univ. Press, Cambridge, Mass., 1973); D. I. Slobin, in Studies of Child Language Development, C. A. Ferguson and D. I. Slobin, Eds. (Holt, Rinehart & Winston, New York (1973), pp. 175-208.

8. The children produced a third type of sign, the marker, which did not refer to things and events but rather served modulation functions. Sign markers were head nods and side-to-side head shakes and were reminiscent of words such as "yes" and "no" in English; for instance, in the sentence "There are no trucks," the "no" mod- ulates, in particular negates, the existence of trucks.

9. The data in Fig. 1 include only two-sign phrases. We exclude phrases containing three elements (such as point at book, "give" sign, point at self, to request that the book be given to the child) and also exclude phrases containing either re- peated elements or simultaneously sign ele- ments (such as point at book, "give," point at book; or point at book signed simultaneously with "give"). In addition, we exclude all phrases containing points at pictures because the children tended to point at pictures before producing other signs. The pictures pointed at were often facsimiles of objects playing the patient role; thus, we would have, perhaps arti- factually, inflated our patient-first orderings if we had included these phrases. As a result, Tracy (observed for two sessions at 4 years 1 month and-4 years 3 months) was not included in

was described by a system similar to the one used to describe American Sign Lanuage. The dimensions used in the descriptions are de- scribed by W. C. Stokoe, Jr. [Stud. Linguist. Occas. Pap. 8 (1960)].

4. A detailed account of the criteria for single signs and an account of the lexical data are given by H. Feldman [thesis, University of Pennsylvania (1975)1; the criteria for sign phrases and for the data on syntactic and semantic relations are de- scribed by S. Goldin-Meadow (Stud. Neurolin- guist, in press).

5. A description of the method of rich inter- pretation is given by L. Bloom [Language De- velopment (MIT Press, Cambridge, Mass., 1970); One Word at a Time (Mouton, The Hague, 1973)].

6. The system we use to describe the deaf child's phrases is an adaptation of the case system pre- sented by C. J. Fillmore [in Universals in Lin- guistic Theory, E. Bach and R. T. Harms, Eds. (Holt, Rinehart & Winston, New York, 1968), pp. 1-88].

7. R. Brown, A First Language (Harvard Univ. Press, Cambridge, Mass., 1973); D. I. Slobin, in Studies of Child Language Development, C. A. Ferguson and D. I. Slobin, Eds. (Holt, Rinehart & Winston, New York (1973), pp. 175-208.

8. The children produced a third type of sign, the marker, which did not refer to things and events but rather served modulation functions. Sign markers were head nods and side-to-side head shakes and were reminiscent of words such as "yes" and "no" in English; for instance, in the sentence "There are no trucks," the "no" mod- ulates, in particular negates, the existence of trucks.

9. The data in Fig. 1 include only two-sign phrases. We exclude phrases containing three elements (such as point at book, "give" sign, point at self, to request that the book be given to the child) and also exclude phrases containing either re- peated elements or simultaneously sign ele- ments (such as point at book, "give," point at book; or point at book signed simultaneously with "give"). In addition, we exclude all phrases containing points at pictures because the children tended to point at pictures before producing other signs. The pictures pointed at were often facsimiles of objects playing the patient role; thus, we would have, perhaps arti- factually, inflated our patient-first orderings if we had included these phrases. As a result, Tracy (observed for two sessions at 4 years 1 month and-4 years 3 months) was not included in

this analysis because she produced very few ac- tion phrases which did not contain points at pic- tures. The data that appear in Fig. 1 represent 64, 83, 92, 70, and 86 percent of all the two-sign, pictureless action phrases produced by David, Dennis, Donald, Kathy, and Chris, respectively.

10. The following conventions are used in describ- ing the order rule: (i)-* indicates that the symbol on the left can be rewritten as the symbol or symbols on the right. The order of the symbols on the right must be maintained in the rewriting process. (ii) ( ) indicates that the symbol in the parentheses is optional, that is, it either can or cannot be chosen in the rewriting process.

11. S. Goldin-Meadow and H. Feldman [Sign Lang. Stud. 8, 225 (1975)].

12. R. A. Gardner and B. T. Gardner, Science 165, 664 (1969); B. T. Gardner and R. A. Gardner, Behav. Non-Hum. Primates 4, 117 (1971); A. J. Premack and D. Premack, Sci. Am. 227, 92 (Oc- tober 1972). Gardner and Gardner report that Washoe has invented signs for certain objects; although striking, this accomplishment does not address the issue of whether or not Washoe would invent such signs if she had not been ex- posed to a standard manual language model.

13. C. Hayes, The Ape in Our House (Harper, New York, 1951); W. N. Kellogg, Science 162, 423 (1968). Although the Kellogg chimpanzee Gua occasionally did gesture (such as protruding lips toward a cup to mean "drink"), her gestures ap- peared to be far less explicit than our deaf chil- dren's signs (such as tilting a C-shaped palm to- ward the mouth several times without the cup in the hand, which was David's signs for "drink"); moreover, Gua did not combine signs into phrases as did our deaf children.

14. We thank D. Burke, J. Huttenlocher, K. Kaye, J. McClelland, and B. Meadow for reading ear- lier versions of this paper; E. Newport for help- ful suggestions; L. Tefo and B. Gray for help in coding videotapes; our subjects and their fa- milies for continued cooperation throughout the study; and L. Gleitman for contributions to both our thoughts and language. Supported by a Spencer Foundation grant to S.G.-M. and H.F. while they were students at the University of Pennsylvania, an NSF graduate fellowship to H.F., an American Association of University Women predoctoral fellowship to S.G.-M., NIH training grant HD 00337 under the direction of J. Aronfreed, and NIH research grant HD 52744 to R. Gelman.

24 May 1976; revised 11 February 1977

this analysis because she produced very few ac- tion phrases which did not contain points at pic- tures. The data that appear in Fig. 1 represent 64, 83, 92, 70, and 86 percent of all the two-sign, pictureless action phrases produced by David, Dennis, Donald, Kathy, and Chris, respectively.

10. The following conventions are used in describ- ing the order rule: (i)-* indicates that the symbol on the left can be rewritten as the symbol or symbols on the right. The order of the symbols on the right must be maintained in the rewriting process. (ii) ( ) indicates that the symbol in the parentheses is optional, that is, it either can or cannot be chosen in the rewriting process.

11. S. Goldin-Meadow and H. Feldman [Sign Lang. Stud. 8, 225 (1975)].

12. R. A. Gardner and B. T. Gardner, Science 165, 664 (1969); B. T. Gardner and R. A. Gardner, Behav. Non-Hum. Primates 4, 117 (1971); A. J. Premack and D. Premack, Sci. Am. 227, 92 (Oc- tober 1972). Gardner and Gardner report that Washoe has invented signs for certain objects; although striking, this accomplishment does not address the issue of whether or not Washoe would invent such signs if she had not been ex- posed to a standard manual language model.

13. C. Hayes, The Ape in Our House (Harper, New York, 1951); W. N. Kellogg, Science 162, 423 (1968). Although the Kellogg chimpanzee Gua occasionally did gesture (such as protruding lips toward a cup to mean "drink"), her gestures ap- peared to be far less explicit than our deaf chil- dren's signs (such as tilting a C-shaped palm to- ward the mouth several times without the cup in the hand, which was David's signs for "drink"); moreover, Gua did not combine signs into phrases as did our deaf children.

14. We thank D. Burke, J. Huttenlocher, K. Kaye, J. McClelland, and B. Meadow for reading ear- lier versions of this paper; E. Newport for help- ful suggestions; L. Tefo and B. Gray for help in coding videotapes; our subjects and their fa- milies for continued cooperation throughout the study; and L. Gleitman for contributions to both our thoughts and language. Supported by a Spencer Foundation grant to S.G.-M. and H.F. while they were students at the University of Pennsylvania, an NSF graduate fellowship to H.F., an American Association of University Women predoctoral fellowship to S.G.-M., NIH training grant HD 00337 under the direction of J. Aronfreed, and NIH research grant HD 52744 to R. Gelman.

24 May 1976; revised 11 February 1977

Relative Fecundity and Parental Effort

in Communally Nesting Anis, Crotophaga sulcirostris

Abstract. The contribution of eggs to the communal clutch by females of the group and the genetic contribution by males of the group are significantly skewed. The amount of parental care performed by each bird is correlated with relative egg own-

Relative Fecundity and Parental Effort

in Communally Nesting Anis, Crotophaga sulcirostris

Abstract. The contribution of eggs to the communal clutch by females of the group and the genetic contribution by males of the group are significantly skewed. The amount of parental care performed by each bird is correlated with relative egg own- ership for both sexes.

True communal nesting, in which sev- eral females regularly deposit their eggs into a single nest, is now known to occur in a number of avian species such as rheas, tinamous, anis, ostriches, magpie geese, and pukekos (1). While the coop- erative nature of this breeding system has been emphasized, the degree of skew in the clutch sizes of communal females

has not been reported for any of these species. If the number of eggs the group can incubate or raise successfully is lim- ited, females should attempt to ensure that the largest possible fraction of the communal clutch is theirs.

A phenomenon commonly observed in some of these species is the presence of eggs strewn about in the vicinity of the nest. Several explanations of this appar-

ership for both sexes.

True communal nesting, in which sev- eral females regularly deposit their eggs into a single nest, is now known to occur in a number of avian species such as rheas, tinamous, anis, ostriches, magpie geese, and pukekos (1). While the coop- erative nature of this breeding system has been emphasized, the degree of skew in the clutch sizes of communal females

has not been reported for any of these species. If the number of eggs the group can incubate or raise successfully is lim- ited, females should attempt to ensure that the largest possible fraction of the communal clutch is theirs.

A phenomenon commonly observed in some of these species is the presence of eggs strewn about in the vicinity of the nest. Several explanations of this appar-

ent wastage have been offered, usually in terms of negligence, poor breeding syn- chrony, improperly built or unfinished nests, the onset of male incubation, or predators (1). As part of a broader study of communal nesting in groove-billed anis (Crotophaga sulcirostris), I exam- ined this question of egg loss and its im- plications. I report here that (i) egg loss- es are a direct result of competition among females, (ii) egg losses create a skew in the egg contribution of each fe- male to the communal clutch, and (iii) the amount of parental care is correlated with relative egg contribution for both males and females (2).

Nesting groups of groove-billed anis consist of from one to four monogamous pairs. Such breeding units are stable

403

ent wastage have been offered, usually in terms of negligence, poor breeding syn- chrony, improperly built or unfinished nests, the onset of male incubation, or predators (1). As part of a broader study of communal nesting in groove-billed anis (Crotophaga sulcirostris), I exam- ined this question of egg loss and its im- plications. I report here that (i) egg loss- es are a direct result of competition among females, (ii) egg losses create a skew in the egg contribution of each fe- male to the communal clutch, and (iii) the amount of parental care is correlated with relative egg contribution for both males and females (2).

Nesting groups of groove-billed anis consist of from one to four monogamous pairs. Such breeding units are stable

403

This content downloaded from �������������152.13.18.10 on Tue, 24 Nov 2020 07:44:53 UTC��������������

All use subject to https://about.jstor.org/terms