comprehension skills of language-competent and ... - CiteSeerX

Language & Communication, Vol. II, No. 4, pp. 301-317, 1997 0 1997 Elsevier Science Ltd All rights reserved. Printed in Great Britain

0271-5309/97 $17.00+0.00

PII: SO271-5309(97)00012-8

COMPREHENSION SKILLS OF LANGUAGE-COMPETENT AND NONLANGUAGE-COMPETENT APES S. L. WILLIAMS, K. E. BRAKKE and E. S. SAVAGE-RUMBAUGH Initial language studies of humans and of apes revealed that humans, unlike apes, already have a solid understanding of the nature of linguistic reference when they begin to produce language (Huttenlocher, 1974). Overlooking the sophisticated comprehension that undergirds a child’s understanding of language, the first studies with apes began simply by teaching rote symbol-object associations wherein an ape looked at an object and responded by producing a sign or selecting a symbol. Production learned in this way, however, did not generalize across communicative function; nor did production generalize to receptive competence (Savage-Rumbaugh, 1981). More recent work has stressed the comprehension of language as a communicative instrument prior to the onset of production and has shown that apes, like children, can learn to differentiate speech sounds, attach communicative intent to those sounds and even link specific sounds with appropriate environmental events and referents (Savage-Rumbaugh, 1986; Sevcik and SavageRumbaugh, 1994). In 1993, Savage-Rumbaugh and her colleagues demonstrated the ability of a young child and a bonobo chimpanzee to comprehend simple sentences and simple syntactic structures in an experimental setting (Savage-Rumbaugh et al., 1993). The level of comprehension required in this setting was quite advanced, since experimental controls strictly precluded both child and bonobo from using any contextual assists in understanding the sentences spoken to them. In a normal language-acquisition environment, context and speech are often paired and so entrenched in the culture that the learner acquires the meaning of words and syntactical structures by exposure to speech uttered in a context that may convey the speaker’s intent. Examples of such contexts include turn-taking games and interindividual routines promoting joint attention, such as diaper-changing, playing, eating, preparing food, etc. (Brunei-, 1975; Bruner, 1983; Bakeman and Adamson, 1984). Since context can determine and/or clarify a speaker’s intent, it is important to ascertain the extent to which this may or may not be occurring in language comprehension studies with both children and apes (e.g. Benedict, 1979; Savage-Rumbaugh, 1991; Bates, 1993). It is all too easy to assume that language comprehension is occurring when indeed it is actually comprehension of overall context - and what should be done in that context - that is really happening. This does not mean that comprehension of context and what to do in context is in itself a trivial phenomenon. For example, a caretaker may ask a nonlanguage-using ape to ‘Make the chimp doll tickle your foot.’ The caretaker may say this in the context of just having demonstrated this action upon herself. In order to respond appropriately, even without understanding the sentence, the ape has to recognize that it is being asked to act Correspondence relating to this paper should be addressed to S. L. Williams, Language Research Center, 3401 Panthersville Road, Decator, GA 30034, U.S.A.

301

S. L. WILLIAMS et al.

on the object and with the object as though the object was an animate being. The ape is also expected to know that it must imitate the actions of the caretaker - actions that during the demonstration are directed toward the caretaker herself. The ape needs both to imitate what it sees and also to understand the intent of the speaker. A dog cannot achieve such an understanding, but apes often can (Hayes, 1951), as can young children (Bates et al., 1979), making it difficult to determine whether correct responses with such contexts are based upon language comprehension per se or upon extralinguistic cues. Of course, it is often impractical, with apes and children alike, to demand that language comprehension be context-free before one recognizes that it is occurring. To ensure the very process of language acquisition, the parent or caregiver must have some understanding of a young child’s or ape’s level of comprehension even as they start to comprehend. The beginning of language comprehension in children can be observed between 8-10 months of age (Bates, 1993). Throughout this time, an adult may not always be aware of the degree to which he or she is monitoring a child’s development of comprehension, but nonetheless the caregiver has a relatively good understanding of what the child can comprehend; and she or he will direct communications to the child accordingly (e.g. Bates, 1993). Given that context does influence this process, we looked at the comprehension of spoken English sentences in a pair of bonobos (Pan paniscus). To date, three bonobos and one common chimpanzee (Pan troglodytes) have demonstrated a capacity to understand speech and graphic symbols when exposed, by the age of 6 months, to caretakers who communicate vocally and lexically with them (Savage-Rumbaugh, 1984; Savage-Rumbaugh et al., 1992, 1993; Sevcik and Savage-Rumbaugh, 1994). Three additional common chimpanzees, all exposed to language after they were 2 years old, have also acquired competence with lexical symbols, but have failed to understand speech in spite of many years of exposure to speech and symbols (Rumbaugh et al., 1982, for review; Savage-Rumbaugh, 1986). Additionally, these common chimpanzees did not acquire lexical symbols spontaneously, but required explicit training, with repeated practice on each symbol (Rumbaugh, 1977; Savage-Rumbaugh, 1986). The learning environment plays an especially important role in the development of language comprehension and production in humans (Curtiss, 1977; Lock, 1991). As the previously cited studies with apes indicate, age of exposure to the linguistic environment during rearing can be an important factor. A number of studies indicate the existence of a critical developmental period that interacts with environmental factors in shaping an individual’s later behavioral/cognitive repertoire (Lenneberg, 1967; Miller, 198 1; Werker and Desjardins, 1995). In humans, these environmental factors largely arise from people and their culture, both of which ensure that the following occur during early infancy: speech is child-directed, repetitive acts are routinized, verbal utterances are used to indicate change in behavior, focused attention is achieved and shared with the caregiver, and that infants are encouraged to engage in communication and to respond appropriately to utterances addressed to them (Snow, 1977; Lock, 1980; Brunei-, 1983; Bakeman and Adamson, 1984). Given sufficient exposure to spoken language at an early age, one ape, Kanzi (Pan paniscus), has been shown to comprehend novel sentences on the first presentation (Savage-Rumbaugh et al., 1993). Unlike dogs, or other animals who have been trained to respond in a specific manner to spoken commands, Kanzi displays a comprehension of words that exists apart from any particular response to an occurrence of the word. Thus, when he hears a word such as ‘ball’, Kanzi can not only retrieve a ball, he can also select a

.

COMPREHENSION SKILLS OF LANGUAGE

303

picture of a ball, select a previously learned arbitrary symbol for ball, or, most importantly, respond to simple sentences in which the word ‘ball’ is combined with other words (Savage-Rumbaugh, 1987). Apes can thus acquire these linguistic skills spontaneously, as do children, if they are reared from infancy in a linguistically rich environment (SavageRumbaugh et al., 1985, 1986; for details on the early rearing experiences of Kanzi see Savage-Rumbaugh et al., 1986). Another bonobo, Panbanisha, has also demonstrated receptive competency to verbal sentences. As one indication of her level of comprehension, 483 unique sentences were scored from daily notes when she was 3 years old. Results showed that she responded appropriately to 93Oh of these sentences (Brakke and Savage-Rumbaugh, 1995), many of which were context-independent. Panbanisha was thus able to attend to caregivers when asked, to extract information from verbal requests, and to carry out appropriate responses that could only be determined by the syntactical structure of the sentence. Neither Kanzi nor Panbanisha were ‘trained’ to use lexigrams or to understand speech in the traditional sense of the word (i.e. via consistent associative relationships, limited by context). Rather, they were exposed to language in a manner similar to human children in that intentional, functional, and referential use of words in varying contexts constituted the multitude of linguistic experiences (Savage-Rumbaugh, 1987). Just as with children, early communications were contextually bound and then became more independent as time went on. These early experiences required that apes and their caregivers share a familiar context, thus simplifying the form and interpretation of each communication (Bruner, 1983). Both Panbanisha and Kanzi came to understand spoken English words and specific word order in context-supported and context-independent situations. Their ability to comprehend the relational meaning in sentences with three elements reveals their sensitivity to word order and their understanding of elements of speech. Both demonstrate the importance of early exposure to a linguistic environment in which language can be learned via observation in a species other than man. The current report takes these findings a step further by looking at early rearing variables that are potentially important to the development of sophisticated language comprehension skills in apes. Data were collected on two sibling bonobos, Panbanisha and Tamuli, differing in age by one year. Panbanisha was exposed to an enriched linguistic environment at 7 weeks of age, and Tamuli was exposed after 3.5 years of age. Responses to context-appropriate sentences were recorded for both during normal, daily activities. The findings provide a comparison of the effects of differential age-determined exposure to an enriched linguistic environment on the development of linguistic comprehension. Method

Subjects Panbanisha, a female bonobo, was born in November 1985. Seven weeks later she was introduced to human caregivers and given her first exposure to spoken English and a visual-graphic lexigram keyboard system’ (Brakke and Savage-Rumbaugh, 1995). Panbanisha formed a strong bond with her caregivers, thus providing an environment conducive to communicative interchange. Panbanisha was raised in a complex social environment that included trips in the woods, social exchanges with other apes and humans, and routine daily activities such as meal preparation and cleanup. Human companions encouraged the apes to attend to lexical communications, but did not require compliance or comprehension. The apes were


allowed to play and interact socially with other apes or humans as they desired (for a more detailed description of the rearing environment see Savage-Rumbaugh et al., 1990). Every intention, travel desire, need and want of ape and human was negotiated to ensure that unambiguous decisions were agreed upon by all. As an initial baseline measure of comprehension, Panbanisha had previously been given a single-word vocabulary test when she was approximately 3.5 years old (see Brakke and Savage-Rumbaugh, 1995); she demonstrated receptive competency with 179 of 2 17 words presented.* Tamuli, a female bonobo born in May 1987, served as the mother-reared control subject. While Panbanisha experienced language input from human caretakers during most of the day, Tamuli remained with her mother, Matata. Tamuli was provided only intermittent access to Kanzi and Panbanisha, as well as to human caregivers. During this time, caretakers did not regularly participate in keyboard-based linguistic interactions specifically addressed to Tamuli, although they did talk to her and play with her. When Tamuli was 3.5 years old she was permitted to accompany Panbanisha or Kanzi on daily excursions through the forest or engage in other activities with the same caretakers who raised Kanzi and Panbanisha. The caretakers spoke to her as they pointed to the keyboard and described their activities. She enjoyed these activities a great deal and eagerly sought to accompany her siblings. Unlike her siblings who had been exposed to language as infants, Tamuli never exhibited any functional use of the lexigrams during social interactions. Nor did she appear to understand the use of lexigrams by others even though her daily exposure to spoken English and lexigrams after the age of 3.5 was similar to Panbanisha’s and Kanzi’s experiences at an earlier age. She did, however, clearly understand the value of the keyboard as a communicative device and often attempted to use it to convey messages. Yet, there was no reliable correspondence between the symbols that she selected and her behavior. These activities continued for 7 months prior to the study reported in this paper. Near the end of that time, after Tamuli was at ease in the linguistic environment and welladjusted to caretakers, she was exposed to a setting in which she had to sit still and listen to caretakers deliver sentences. Several practice trials were presented for the first few days to acclimatize Tamuli to the new situation. Practice sentences were not included in test sessions. Data were collected during the fall of 1991, when Panbanisha was 5.5-6 years old and Tamuli was 4-4.5 years old. Procedure This study was designed to provide a measure of the apes’ competencies relative to those of each other and children in everyday situations; it was not designed to systematically eliminate all contextual support from the assessment situation. Tests for purposes of scientific validation of linguistic understanding per se were conducted at an older age with Kanzi, as his comprehension skills increased to the point where he would respond reliably even in such artificial situations. However, initial comprehension skills cannot be assessed reliably in such settings either in children or apes because noncontextually based utterances seem ‘odd’ or ‘out of place’ and the reason for uttering them is unclear to the child or ape (Bridges, 1979; Bates, 1993). Consequently our techniques for ‘early assessment’ of word-order comprehension were similar to those used by Bates and others for children in


30.5

their ‘in home’ assessment instrument (Chapman and Kahn, 1978; Bates et al., 1988; Dale et al., 1989; Dale, 1991). One hundred and forty-five sentences were presented to both bonobos under identical conditions. During the presentation of spoken sentences, caretakers made efforts to avoid the use of gestures, body postures or glances that could serve as aids to speech comprehension; however, strict controls to prevent inadvertent cueing are impractical in nontest settings. In these test sessions, Panbanisha and Tamuli were asked to identify objects, pictures, food items, etc., or to carry out a behavior in a situation in which many alternatives were possible. For example, if the bonobo was asked to ‘put the tomato on the blanket’, other items (ball, bubbles, umbrella, potty, tomato, blanket, hammer, backpack, rubberband, vacuum cleaner, videotape, clay, plastic bag) were also available for consideration. Sentences were contextually appropriate, for example, the ape might be asked to ‘wash the apple’ while the caretaker was participating in food preparation. During these tests, the bonobos were given free access to food and allowed to play between trials. Trials were never initiated unless they were willing to participate; therefore, it was important to maintain a cooperative attitude in both bonobos in order for data collection to progress at a reasonable pace. Because the caretakers also participated in the apes’ rearing, the bonobos were accustomed to cooperating with them on a daily basis. The caretakers were also adept at determining the bonobos’ moods and at setting occasions that would facilitate a cooperative attitude. In spite of this, the bonobos sometimes balked and occasionally refused to participate. When this occurred, they were engaged in another activity and the sentence was represented later. Presentation of the sentences permitted us to obtain a measure of language comprehension that reflected each ape’s overall ability to respond to a variety of linguistic communications in many familiar and unfamiliar situations. Speakers stated each English sentence in a clear and concise manner. The sentences consisted of two or more words and were delivered when the bonobo was attending to the speaker. If there was no response after repeating the sentence several times, caretakers were told to use the lexigram keyboard concurrently with the next presentation of the sentence. If this failed, the trial was considered over and, if conditions were favorable, caretakers could then demonstrate the sentence for the bonobos in order to assess whether they could then carry out the request. Sentences were categorized according to sentence type with the following elements: action (A), object (0) recipient (R), and location (L). If a particular sentence type was not presented it was eliminated from the analysis so that zeros represented true cases in which responses to the utterances were wrong. The trials were videotaped and recorded verbatim and later entered into a computer database. Bonobos’ responses to sentences were scored as being correct, partially correct (i.e. response to at least one but not all elements of the sentence was correct), totally incorrect, or no response, based on behavioral responses immediately following each statement. If the response was not entirely correct, it was given a code to identify the type of error(s) observed. If, upon observing the video tape, the caretaker began a sentence when the ape was already holding or touching one of the items mentioned in the sentence, that sentence was not scored as correct, even if the ape responded appropriately. This happened rarely. If, after the trial was over, the caretaker decided to demonstrate the correct response, a code was also assigned to delineate the following categories: imitation, partial imitation,

306


no change in behavior (i.e. does nothing), or molding (caretaker puts correct object in hands and/or then manipulates hands/body into the correct behavior. Tamuli was 4.5 years old during data collection. At this age, she gave no pretest evidence of comprehending any single words, though sometimes, in context, she appeared to understand a few requests. Kanzi and Panbanisha however could identify a minimum of 150 single words and could respond to very complex sentences in context, even when there appeared to be no obvious cues. By comparing Tamuli and Panbanisha in a systematic manner we hoped to determine whether or not the context itself was permitting Panbanisha (and by inference Kanzi as well) to appear to comprehend language. Since Tamuli failed single word tests, if she could comprehend sentences in the everyday context of daily life, then we would conclude that Kanzi and Panbanisha’s comprehension of complex sentences was not based on speech, as we believed, but on the basis of subtle and unknown cues inherent within the contexts afforded by daily living circumstances, routines, etc., at the Language Research Center. Observer agreement Two caretakers independently scored an initial set of 10 sentences randomly selected from the videotapes in order to establish validity of scoring criteria. There was 100% agreement between the two scorers on response identifications. A Cohen’s Kappa was then calculated for another randomly selected set of 50 sentences (25 for Panbanisha and 25 for Tamuli) from the videotapes. Results revealed a high level of agreement between the two scorers for response identzjication, Cohen’s Kappa= 0.89; an acceptable level of agreement for error identzjication, Cohen’s Kappa=O.64; and an acceptable level of agreement for categorizing the behavior following the demonstration, Cohen’s Kappa = 0.60 (Cohen, 1960). Results Panbanisha responded correctly on 112 (77%) of the 145 sentences recorded. Tamuli was correct on only 9 (6Oh). Panbanisha refused to respond to two sentences; she was never completely incorrect in any of her responses. Tamuli, on the other hand, was given an ‘incorrect’ score on 92 (63Oh) responses and refused to respond to 15 sentences. Figure 1 provides a descriptive presentation of the type of utterances and the performance scores for each bonobo. Table 1 provides the raw scores and total number of sentences within each sentence type. Data from 263 sentences given to Kanzi when he was 5.5 years old are included in Fig. 1 and Table 1 for purposes of comparison (Savage-Rumbaugh, 1987). These results indicate that both Kanzi and Panbanisha differed dramatically from their nonlanguage-exposed sibling with regard to comprehension of spoken English sentences (x2 (10,553) = 41.94, P-C 0.001). Kanzi and Panbanisha performed similarly well even though their total vocabularies were somewhat different (x2 (5408) = 6.09, p > 0.05). For a complete corpus of sentences presented to Panbanisha and Tamuli see Tables 2-7. It is interesting to note that on 20 occasions Tamuli went to the keyboard immediately after hearing the sentence and touched one or more of the lexigrams. Panbanisha never did this, apparently recognizing that the speaker was not asking for a communicative response but a behavioral one. Tamuli’s correct responses to nine of the sentences suggested that she may have learned some spoken English wcrds prior to this time. She was therefore presented with a test of


Table

307

1. Summary of sentence comprehension data for Panbanisha and Tamuli

Sentence type

Panbanisha % (#C/F)

Tamuh (#C/F) %

Kanzi* (#C/F) %

Action-object Action-location

45150 9112

90 75

5150 2112

10 17

105/107 23123

98 100

Subtotal Action-recipient-object Action-object-recipient Action-object-location

54162 19124 6107 25142

87 79 86 60

7162 1124 o/o7 1142

11 04 00 02

128/130 3104 17/19 33136

98 75 89 92

Subtotal Other

50173 8/10

68 80

2173 O/l0

03 00

53159 65174

90 88

112/145

77

91145

06

2461263

94

Grand total

Number correct (#C), Frequency (F), and Percent correct (%) for different sentence types presented to Panbanisha and Tamuli. *Kanzi’s data is from Savage-Rumbaugh (1987).

single word comprehension, similar to that given to Kanzi and Panbanisha at approximately the same age, in which three objects or photos were presented in an array. One represented the target, and the other two were randomly selective alternatives. Eight words (carrot, plastic bag, book, water, cooler, backpack, potty, cereal) were selected from the sentences that Tamuli had appeared to understand. Each of the words was presented verbally four times, three times with objects as the stimulus set and once with photographs (targets were presented randomly within blocks of eight trials). The caretaker was blind to the position of the correct answer as she delivered each word. Tamuli had been familiarized with the procedure prior to this assessment through presentation of words, objects and photos not included in the test. Tamuli responded at chance levels with both real objects and photos. 100

r

90

c

80

t T

;;

c

40

50

0 R R

30 20

:

10

7 -

0 A0

AL

AR0

AOL

AOR

OTHER

SENTENCE TYPE m Kanzi*

m Panbanisha

I Tamuli

*Kanzi’s data taken from Savage-Rumbaugh (1987).

Fig. 1. Overall performance on test of English comprehension.

308


Table 2. Action-object requests Correct responses by Panbanisha Bite the apple Bite the dog Let’s do the shots Find the book Find the bubbles Get the backpack Get the blanket Get the bowl Get the bubbles Get the can opener Get the carrot Get the clay Get the hammer Get the ice Get the keyboard Get the leaf Get the oil Get the spoon Get the soap Get the straw Get the string Get the thermometer Get the vacuum

Get the water Grab the leaves Hammer the soap Make the apples I need the knife I need the toothpaste Open the bubbles Open the clay Open the cooler Open the oven Open the toothpaste Open the umbrella Open the wipes Do you see the dog? Do you see the hat? Do you see the paper? Throw the ball Turn (oil) the lights Turn (on) the TV Turn (on) the vacuum Use the telephone Wash the cup Correct responses by Tamuli

Find the book Get the backpack Get the plastic bag

Open the cooler Where’s the water? Incorrect responses by Panbanisha

Brush the doggie See the picture of Liz Get the pineneedles

Wash the apples Wash the peach Incorrect responses by Tamuli

Bite the apple Bite the dog Brush the doggie Lets’s do the shots Find the bubbles Get the bowl Get the bubbles Get the can opener Get the carrot Get the hammer Get the ice Get the keyboard Get the leaf Get the oil Get the pineneedles Get the spoon Get the soap Get the straw Get the string Get the thermometer Get the clay Get the vacuum Grab the leaves

Hammer the soap Make the apples I need the knife I need the toothpaste Open the bubbles Open the clay Open the oven Open the toothpaste Open the umbrella Open the wipes Get the dog Get the hat Get the paper Which is picture of Liz? Throw the ball Turn (olI) the lights Turn (on) the TV Turn (on) the vacuum Use the telephone Wash the apples Wash the cup Wash the peach


309

These results suggest that Tamuli had not learned these words during her linguistic experience. Therefore, there may be other explanations for her performance on the sentences in question. For example, serendipitous placement appears to explain Tamuli’s appropriate response to one sentence. When asked to ‘Find the book’, Tamuli ran to look Table 3. Action-object-location request Correct responses by Panbanisha Carry balloon to bedroom Get wipe out of backpack Get yogurt from the fridge Get monster out of the water Look for surprise in fridge Put the apple in the bowl Put the bowl in the backpack Put carrot in the backpack Put carrot in plastic bag Put the chow in the backpack Put clay on the paper Put the collar on Put the hat on your head

Put mushrooms in trash Put peach on umbrella Put raisins in jelly Put the shoe on your foot Put the soap in the cup Put the straw in the ball Put the straw in the paint Put the towel on your head Put the water in trashcan Put the water on the chow Put water on toothbrush Throw the rock at the noise Correct responses by Tamuli

Put carrot in the plastic bag Incorrect response by Panbanisha Hide soap behind tree Hide pineneedles under car Hide the snake under the cube Put blanket in the machine Put book in the plastic bag Put the bubbles in the wipes Put the milk in the hat Put the orange in the bowl Put the paper in trash

Put paper on the vacuum Put raisins on the book Put the rock on the towel Put the shirt on Put shoe in the plastic bag Put the soap on the vacuum Put the soap on the dog Take the hat off your head Incorrect responses by Tamuli

Take balloon to the bedroom Get wipe out of the backpack Get yogurt from the fridge Get monster out of the water Hide soap behind the tree Hide pineneedles under car Hide snake under the cube Get surprise from the fridge Put the apple in the bowl Put blanket in the machine Put book in the plastic bag Put the bowl in the backpack Put the bubbles in the wipes Put the chow in the backpack Put the clay on the paper Put the collar on Put the hat on your head Put the milk in the hat Put mushrooms in the trash Put the orange in the bowl Put the carrot in backpack

Put paper in the trash Put paper on the vacuum Put peach on the umbrella Put raisins on the book Put raisins in the jelly Put the rock on the towel Put the shirt on Put shoe in plastic bag Put the shoe on your foot Put the soap in the cup Put soap on the vacuum Put soap on the dog Put the straw in the ball Put the straw in the paint Put the towel on your head Put water in the trashcan Put water on the chow Put water on toothbrush Take the hat off your head Throw rock at the noise

310


Table 4. Action-recipient-object requests Correct responses Call somebody on the phone Let’s give doggies a rock Get me some mushrooms Get me the snake picture Give me the apple Give me the onions Give me the shot Show me the cereal Show me the flashlight Show me the soap

Panbanisha

Show me the ball Show me the bunny Show me picture of chow Show Shelly the vacuum Show me the rubberband Show me the clay Show me the mirror Show me the tomato Show me the monster hand Correct response by Tamuli

Show me the Cereal Incorrect responses by Panbanisha Give me the banana Give me the gorilla picture Give Linda the rock

Give Jeannine the shot Show me the carrot Incorrect responses by Tamuli

Call somebody on the phone Give the doggies a rock Get me the mushrooms Get me the snake picture Give me the apple Give me the banana Give me the onions Get the shot for me Give me the gorilla picture Give the rock to Linda Give Jeannine the shot Show me the carrot

Show me the flashlight Show me the soap Show me the ball Show me the bunny Show me the picture of Chow Show Shelly the vacuum Show me the rubberband Show me the clay Show me the mirror Show me the tomato Get me the monster hand

Table 5. Action-location requests Correct responses by Panbanisha Draw on the paper Go to the potty Hide in the cabinet Hide under the blanket Open the play yard

Can I tickle your tummy? Can I tickle your butt? Can I tickle your mouth? Let’s go see Maryann Correct responses by Tamuli

Go to the potty

Can I tickle your butt? Incorrect responses by Panbanisha

Let’s dig in the bark Knock on the door

Can I tickle your head? Incorrect responses by Tamuli

Let’s dig in the bark Draw on the paper Hide in the cabinet Hide under the blanket Knock on the door

Open the play yard Can I tickle your head? Can I tickle your mouth Can I tickle your tummy Let’s go see Maryann

311


out the window and picked up a book lying in her path. Two other sentences to which Tamuli responded appropriately involved a plastic bag, an object which she found highly attractive and which she often incorporated in her incorrect responses as well. On the remaining six occasions, Tamuli may have performed the correct response because contextual information in the situation helped her realize the requested action. Table 6. Action-object-recipient requests Correct responses by Panbanisha Give some milk to Linda Take the hot dog to Tamuli Offer some carrot to Linda

Find a grape for Shelly Give some ice to the doggie Give some apple to Linda

Incorrect response by Panbanisha Get the string from Linda Incorrect responses by Tamuli Give some milk to Linda Take hot dog to Panbanisha Offer some carrot to Linda

Find a grape for Shelly Get the surprise from Shelly Give some ice to the doggie Give some apple to Linda

Table 7.

Other requests (not categorized)

Correct responses by Panbanisha Slap monster mask with hammer Go to Fridge and get banana Tell me Yogurt at keyboard Get stick and poke snake house

Let’s tickle with monster hand Get Linda with the snake Hit orange with the hammer Tickle Pam with monster hand

Incorrect responses by Panbanisha Make snake bite the banana

Make doggie bite the orange

Incorrect responses by Tamuli Lets tickle with monster hand Get Linda with the snake Hit orange with the hammer Go to fridge and get banana Slap monster mask with hammer Table 8.

Say Yogurt at the keyboard Get stick and poke the hole Make doggie bite the orange Make snake bite the banana Tickle Pam with monster hand

Categorization of responses after demonstration following trial termination Panbanisha

Tamuli

N

N

Demonstration followed by: Imitation Partial imitation No. response Molded No. demonstration

10 03 01 10 121

17 12 19 37 60

Total

145

145

N = Total number of sentences within each category.

312


Partially correct responses Performance on partially correct sentences tended to differ between the two bonobos. Panbanisha usually performed correctly, but occasionally made errors relative to one element of the sentence. Nineteen of Panbanisha’s 3 1 errors were attributed to performing an incorrect action on appropriate objects. Panbanisha chose the wrong object on only five sentences and made a locative error on five more. Finally, on one sentence she chose not to take an object from a person and she chose the wrong recipient on another sentence when she gave herself the shot instead of Jeannine. Tamuli’s scores of ‘partially correct’, on the other hand, can best be characterized not by the errors she made but by the elements she identified correctly. Twelve of the 29 sentences in this group were judged partially correct because they were indicative in nature. Tamuli only had to point, touch or displace an object for 10 of the sentences, touch the keyboard for one of the sentences, or initiate bodily movement to indicate travel on another. The fact that she responded in this manner indicates that she understood the demands of the test situation, i.e. that she needed to do ‘something’. Therefore, Tamuli got the (simple) action correct but failed to identify the correct object or location to complete the sentence. On seven other partially correct sentences, Tamuli was extremely interested in an object that was part of the array before her; she was therefore oriented to the object when, or even before, the sentence was uttered. Similarly, as part of five sentences Tamuli was asked to incorporate an activity, object or recipient of action that she favored and may have interacted with independent of being asked. For example, while sitting near a tree, Tamuli was asked to ‘hide the soap behind the tree’; Tamuli began to play chase around the nearby tree. In another four sentences, Tamuli performed the correct action with the correct object (‘give apple to Linda’, ‘draw on paper’, ‘give milk to Linda’, ‘put rock on towel’) but failed to identify the correct location or recipient. With five more sentences, the action was incorrect but the object and recipient/location were correct. In these cases the objects were either in Tamuli’s hands or placed directly in front of her, therefore increasing her chances of selecting them. Tamuli knew she had to interact with these objects in some way and the setup of the situation afforded the appropriate actions or recipients/locations. In each of the 29 aforementioned cases, in fact, the context appeared to support the partially correct response given by Tamuli but failed to help her in completing the entire request. Responses to caretaker demonstrations After incorrect, partially correct, and no response trials, caretakers demonstrated the correct response on 24 occasions for Panbanisha and 85 occasions for Tamuli. Table 8 gives the breakdown of the bonobos’ responses following these demonstrations. Panbanisha imitated the correct response on twice as many sentences as Tamuli. Caretakers helped the bonobos perform the correct response after an equal percentage of demonstrations. Behavioral responses to the test situation Behavioral differences other than responses to sentences per se between Panbanisha and Tamuli were apparent during the test situations. For example, Tamuli had a very short attention span and, on many occasions, would reach for objects before the sentence was completed. Seldom did Panbanisha begin her response before the caretaker completed delivery of the sentence.


313

Panbanisha could respond to verbal indicators to attend such as ‘Are you ready to listen?’ by focusing her attention appropriately. Caretakers more often called Panbanisha’s name to initiate a trial (51 times versus 34 for Tamuli), whereas Tamuli often remained inattentive to such inquiries, appearing unable to distinguish these requests from the ongoing stream of vocal communications. Sometimes she would shift her attention briefly as if to acknowledge that she was being addressed, but then would immediately lose interest once the caretaker began to address her with a spoken request. Therefore the caretakers would physically turn her head to establish mutual regard (signaling that attention was required as the trial was initiated) 87 times versus the 42 times this means was used with Panbanisha. On the average, utterances were repeated 2.61 times per delivery for Panbanisha and 3.26 times per delivery for Tamuli. A point biserial correlation between number of times the sentence was repeated and whether it was judged to be correct or not revealed a significant difference for Panbanisha (rni,i = 0.51, t = -6.996, p .c 0.001). In other words, the sentence was repeated more often when she was having trouble understanding what was being asked. There was no significant relation between repetition of sentences and performance for Tamuli. Upon examination of the videotaped trials, other distinguishable behavioral differences were observed between Panbanisha and Tamuli during the sentence presentations. After a sentence was delivered, Panbanisha would look over the array of items or look around the room for the desired object. Her gaze, as indicated from head position, was always directed away from the caretaker while she searched for the appropriate item needed to complete the response. This observation was also reported for Kanzi (Savage-Rumbaugh, 1987). Only after Panbanisha had made her choice did she, on occasion, look towards the caretaker for acknowledgement. On the other hand, Tamuli constantly engaged in eye contact with the caretaker delivering the sentence. For example, when asked to ‘show me the cereal’, she reached out directly in front of her while her eyes remained on the caretaker. Discussion Panbanisha’s and Tamuli’s performances in the sentence-comprehension task reported here unquestionably differed. Panbanisha was clearly using more than the contextual information that was available to Tamuli to formulate her responses. Panbanisha additionally appeared to have access to the linguistic input that human caretakers provided. Tamuli does possess the communicative skills necessary to function successfully in the ape world. Her early exposure to other apes and their communication systems allows her to learn to attend and focus on environmental cues that are important for her to know. During the 7 months that Tamuli was placed in the enriched linguistic environment she learned how to respond to many subtle behavioral cues. Both Tamuli and Panbanisha were able to follow glances and gestures quite well, and both were familiar with the contextual routines of the laboratory environment. However, even though Tamuli learned to follow eye movements, hand movements and postural changes to help her in producing the appropriate behavior, these cues appeared to facilitate a correct response on only a few occasions. This implies that following the subtle movements was not enough for her to understand the intent of the speaker. Despite all that she had learned, she failed to learn what was needed for success on the comprehension test. Panbanisha and Kanzi also possess the skills that allow them to exist successfully in the ape world. However, their early rearing also involved close interaction with humans who

314


spoke and used symbols with them, and who supported the young apes’ development as intentional, symbolic communicators. By being immersed, as it were, in a symbol-using environment during the period of greatest brain plasticity (Purves, 1994) all of the components necessary for language comprehension (and production) were put into place for Kanzi and Panbanisha. The neural substrates of the bonobo brain appear sufficiently developed to accept incoming linguistic information; if such information is available early enough in life, the bonobo can reorganize and adapt it for successful functioning within a human culture, albeit at the level of a young child. Bonobos, who share 99% of their genetic makeup with humans (King and Wilson, 1975; Andrews and Martin, 1987) have complex social environments and exhibit many of the same behaviors as humans (Andrews and Martin, 1987; Savage-Rumbaugh and Lewin, 1994). They enter the world with a readiness to learn how to survive in it. Part of this readiness involves neural-cognitive systems that can access, process and store information from the environment. How and in what form this information is presented therefore determines the various neural connections and integrations. According to Hebb (1949) learning in an enriched environment enhances the neuronal complexity of the brain during early development. Purves (1994) also offers a similar interpretation of early learning and its effects on the brain. He believes that learning in a particular environment results in the maintenance of selected and sustained neuronal connections that are particularly beneficial to the organism and a depletion of neuronal connections that are not. The idea that ‘neural Darwinism’, a theoretical construct proffered by Purves, takes place during early development suggests that much plasticity occurs as a function of environmental conditions. Cortical regions that are specifically useful to the animal or human are therefore used more frequently. These regions have also been known to grow more quickly during maturation than other areas (Purves, 1994). Further confirmation of this idea comes from the investigation of differential energy consumption of various cortical regions which indicates differences in electrical activity of cells. Areas of high electrical activity grow the most during early developmental stages (Riddle et al., 1993). Development of linguistic skills in humans appears dependent on the proper exposure to competent language-using models during early life. If deprived of this exposure, various linguistic disabilities are observed (e.g. Itard, 1932; Curtiss, 1977). However, since the exposure to language at infancy is nearly universal in our species we must look to the nonhuman primate model to help us further understand the relationship between language and environment. The findings of this and other studies (Savage-Rumbaugh et al., 1990; Brakke and Savage-Rumbaugh, 1995) suggest that early exposure to language-using models is critical and necessary for the appearance of speech comprehension in both common chimpanzees and bonobos. In conclusion, language acquisition, perhaps as well as the development of many other cognitive/perceptual abilities (Purves, 1994) appears largely dependent on the type of environment experienced during the first year of life. The fact that bonobos can comprehend spoken English provides support for theories positing that cognitive functions in humans and animals are continuous rather than of entirely different origins and natures. Biological ‘programming’, as nativists portray language competence (Chomsky, 1968), is not unique to the human brain, but appears to share a common underlying substrate with that of the ape brain.


315

~c~~~~~e~ge~e~~~-This research was funded by the National Institutes of Health grant NICHD-06016 which supports the Language Research Center of Georgia State University. The college of Arts and Sciences of Georgia State University also supports this research and the Language Research Center. Some of the data reported here were presented at the International Primatological Meetings in Strasbourg, France in August of 1992 and at the Eastern Psychological Association Meetings in Boston, MA in April of 1992. The authors wish to thank Jeannine Murphy, Linda McGarrah, and Phil Shaw for helping to collect the data.

NOTES ‘The lexigram keyboard consisted of 256 printed symbols, each the equivalent of an English word. 21n similar tests, Kanzi mastered all 57 words on which he was tested at 3.5 years, and later mastered 149 of 194 words tested 2 years later. By the time the latter testing was conducted, Kanzi was demonstrating comprehension of simple novel sentences (see Greenfield and Savage-Rumbaugh, 1990; Savage-Rumbaugh et ul., 1987, 1990).

REFERENCES Andrews, P. and Martin, L. (1987) Cladistic relationships of extant and fossil hominoids. Journal of Humun Evolution 16, 101-108. Bakeman, R. and Adamson, L. D. (1984) Coordinating attention to people and objects in mother-infant and peer-infant interaction. Child Development 55, 1278-1289. Bates. E. (1993) Comprehension and production in early language development (commentary on SavageRumbaugh, et al., ‘Language comprehension in ape and child). Monographs of the Society for Research in Child Development, 58 (334, Serial No. 233). Bates, E., Benigni, L., Bretherton, I., Camaioni, L., and Volterra, V. (1979) The Emergence of Symbols: Cognition and Communication in Infancy. Academic Press, New York. Bates, E., Bretherton, I., and Snyder, L. (1988) From First Words to Grammur: Zndividual Dtflerences und Dissociable Mechanisms. Cambridge University Press, New York. Benedict, H. (1979) Early lexical development: comprehension and production. Journul of ChiZd Languuge 6, 183-200. Brakke, K. E. and Savage-Rumbaugh, E. S. (1995) The development of language skills in bonobo and chimpanzee: I. Comprehension. Language and Communicution 2, 121-148. Bridges, A. (1979) Directing two-year-old’s attention: some clues to understanding. Journal of Child Languuge 6, 211-226. Bruner, J. S. (1975) The ontogenesis of speech acts. Journal of Child Language 6, 183-200. Bruner, J. S. (1983) Child’s Talk: Learning to use Language. Norton, New York. Chapman, R. and and Kahn, L. (1978) Comprehension strategies in two- and three-year-olds: animate agents or probable events? Journal of Speech and Hearing Research 21, 74676 1. Chomsky, N. (1968) Language und Mind. Harcourt, Brace & World, New York. Cohen, J. (1960) A coefficient of agreement for nominal scales. Educational und Psychological A4easurement 20, 3746. Curtiss, S. (1977) Genie: A Psycholinguistic Study of a Modern-day Wild Child. Academic Press, New York. Dale, P. S. (1991) The validity of a parent report measure o f vocabulary and syntax months. Journal Speech and Hearing Sciences 34, 565-57 Dale, P., Bates, E., Reznick, S. and and Morisset, C. (1989) The validity of a parent report instrument of child language at 20 months. Journal of Child Language 16, 239-249. Greenfield, P. M. and Savage-Rumbaugh, E. S. (1990) Grammatical combinations in Pan paniscus: process of learning and invention in the evolution and development of language. In Lunguage und Zntelligence in Monkeys and Apes: Comparative Developmental Perspectives, eds S. T. Parker and K. Gibson, pp. 54t3578. Cambridge University Press, Cambridge. Hayes, C. (195 1) The Ape in our House. Harper and Row, New York. Hebb, D. 0. (1949) The Organization of Behavior. Wiley, New York. Huttenlocher, J. (1974) The origins of language comprehension. In Theories in Cognitive Psychology, ed. R. L. Solso, pp. 33 1-368. Lawrence Erlbaum Associates, Hillsdale. Itard, M. G. (1932) The Wild Boy of Averyon. Meredith Publishing Company, New York.

316


King, M. C. and Wilson, A. C. (1975) Evolution at two levels in humans and chimpanzees. Science 188, 107-116. Lenneberg, E. H. (1967) Biological Foundations of Language. Wiley, New York. Lock, A. (1980) The Guided Reinvention of Language. Academic Press, London. Lock, A. (1991) The role of social interaction in early language development. In Biological and Behavioral Determinants of Language Development, eds N. A. Krasnegor, D. M. Rumbaugh and R. L. Schiefelbusch, pp. 287-300. Erlbaum, New Jersey. Miller, G. A. (1981) Language and Speech. Freeman, San Francisco. Purves, D. (1994) Neural Activity and the Growth of the Brain. Cambridge University Press, Cambridge. Riddle, D. R., Gutierrez. , G., Zheng, D., White, L., Richards, A. and Purves, D. (1993) Differential metabolic and electrical activity in the somnatic sensory cortex of the developing rat. Journal of Neuroscience 13, 41934213. Rumbaugh, D. W. (1977) Language Learning by a Chimpanzee. Academic Press, New York. Rumbaugh, D. M., Savage-Rumbaugh, E. S., and Scanlon, J. L. (1982) The relationship between language in apes and human beings. In Primate Behavior, eds J. L. Fobes and J. E. King, pp. 361-385. Academic Press, New York. Rumbaugh, D. M., Hopkins, W. D., Washburn, D. A., and Savage-Rumbaugh, E. S. (1991) Comparative perspectives of brain, cognition, and language, In Biological and Behavioral Determinants of Language Development, eds N. A. Krasnegor, D. M. Rumbaugh and R. L. Schiefelbusch, pp. 1455164. Erlbaum, New Jersey. Savage-Rumbaugh, E. S. (198 1) Can apes use symbols to represent their world? In The Clever Hans Phenomenon; Communication with Horses, Whales, Apes, and People, eds T. A. Sebeok and R. Rosenthal, 364, pp. 35-59. New York Academy of Sciences, New York. Savage-Rumbaugh, E. S. (1984) Pan paniscus and Pan troglodytes: contrasts in preverbal communicative competence. In The Pygmy Chimpanzee: Evolutionary Biology7 and Behavior, ed. R. L. Susman, pp. 3955413. Plenum Press, New York. Savage-Rumbaugh, E. S. (1986) Ape Language: From Conditioned Response to Symbol. Columbia University Press, New York. Savage-Rumbaugh, E. S. (1987) A new look at ape language: comprehension of vocal speech and syntax. In Nebraska Symposium of Motivation: Comparative Perspectives in Modern Psychology, ed. R. A. Dienstbier, 35, pp. 201-255. University of Nebraska Press, Lincoln. Savage-Rumbaugh, E. S. (1991) Language learning in the bonobo: how and why they learn. In Biological and Behavioral Determinants of Language Development, eds N. Krasnegor, D. M. Rumbaugh, M. StuddertKennedy and R. L. Schiefelbusch, 84, pp. 209-233. Lawrence Erlbaum, Hillsdale. Savage-Rumbaugh, E. S. and Lewin, R. (1994) Kanzi: The Ape at the Brink of the Human Mind. Wiley, New York. Savage-Rumbaugh, E. S. and Rubert, E. (1992) Language comprehension in ape and child: evolutionary implications. In Neurophilosophy and Alzheimer’s Disease, eds Y. Christen and P. S. Churchland, pp. 3048. Foundation Ipsen Pour la Recherce Scientifique, Springer-Verlag. Savage-Rumbaugh, E. S., Rumbaugh, D. M. and McDonald, K. (1985) Language learning in two species of apes. Neuroscience and Biobehavioral Reviews 9, 653-665. Savage-Rumbaugh, E. S., McDonald, K., Sevcik, R. A., Hopkins, W. D. and Rubert, E. (1986) Spontaneous symbol acquisition and communicative use by pygmy chimpanzees (Pan paniscus). Journal of Experimental Psychology: General 115, 2 11-235. Savage-Rumbaugh, E. S., Sevcik, R. A., Brakke, K. E., Rumbaugh, D. M., and Greenfield, P. M. (1990) Symbols: their communicative use, comprehension, and combination by bonobos. In Advances in Znfancy Research, eds C. Rovee-Collier and L. P. Lipsitt, 6, pp. 221-278. Ablex, Norwood. Savage-Rumbaugh, E. S., Brakke, K. E., and Hutchins, S. S. (1992) Linguistic development: contrasts between co-reared Pan troglodytes and Pan paniscus, In Topics in Primatology, Vol. 1, Human Origins, eds T. Nishida, W. C. McGrew, P. Marler, M. Pickford and F. B. M. de Waal, pp. 51-66. University of Tokyo Press, Tokyo. Savage-Rumbaugh, E. S., Murphy, J., Sevcik, R. A., Brakke, K., Williams, S. L., and Rumbaugh, D. M. (1993) Language comprehension in ape and child. Monographs of the Society for Research in Child Development, 58, (334, Serial No. 233). Sevcik, R. A. and and Savage-Rumbaugh, E. S. (1994) Language comprehension and use by great apes. Language and Communication 14(1), 37-58.


Snow, C. E. (1977) The development of conversation between mothers and babies. Journul of Child Lunguuge 4, 1-22. Werker, J. F., and Desjardins, R. N. (1995) Listening to speech in the 1st year of life: experiential influences on phoneme perception. Current Directions in Psychological Science, 4(3), 76-8 1. Cambridge University Press, Cambridge.