historical; cognitive-psychological radical constructivism EMPIRICAL

The Yerkish . From Operational Methodology to Communication

Marco Bettoni A Swiss Distance University of Applied Sciences (Switzerland)

Origins of Yerkish – Purpose: Yerkish is an artificial language created in 1971 for the specific purpose of explor- ing the linguistic potential of nonhuman . The aim of this paper is to remind the The project research community of some important issues and concepts related to Yerkish that seem One day in the fall of 1970 Ray Carpenter, one to have been forgotten or appear to be distorted. These are, particularly, its success, its of the fathers of primatology in the United promising aspects for future research and last but not least that it was Ernst von Glaser- States, came to the almost regular Saturday sfeld who invented Yerkish: he coined the term “lexigrams,” created the first 120 of them golf meeting with von Glaserfeld bringing and designed the grammar that regulated their combination. Design: The first part of with him an intriguing idea: The Yerkes this paper begins with a short outline of the context in which the Yerkish language origi- National Research Center in nated: the original LANA project. It continues by presenting the language itself in more Atlanta1 (Georgia), the first and foremost detail: first, its design, focusing on its “lexigrams” and its “correlational” grammar (the institute of primate research in the USA, was connective functions or “correlators” and the combinations of lexigrams, or “correla- planning to investigate the possibility of tions”), and then its use by the chimpanzee Lana in formulating sentences. The second part communication between and great gives a brief introduction to the foundation of Yerkish in Silvio Ceccato’s Operational apes via a computer by means of a visual lan- Methodology, particularly his idea of the correlational structure of thought and concludes guage. The great apes (, , with the main insights that can be derived from the Yerkish experiment seen in the light ) would probably never learn a of Operational Methodology. Findings: Lana’s success in language learning and the success spoken language, Carpenter said, but they of Yerkish during the past decades are probably due to the characteristics of Yerkish, par- were quick and clever with their fingers and ticularly its foundation in operational methodology. The operation of correlation could be Alan and Beatrice Gardner had successfully what constitutes thinking in a chimpanzee and an attentional system could be what deliv- taught ASL (the American ers the mental content that correlation assembles into triads and networks. used by deaf people) to a chimpanzee called Research implications: Since no other assessment or explanation of Lana’s performances .2 has considered these foundational issues (findings), a new research project or program Despite impressive results in teaching sign should validate the above-mentioned hypotheses, particularly the correlational structure language to the great apes, in those years as of chimpanzee thinking. well as during the following two decades, lin- Keywords: Yerkish, artificial language, correlational grammar, operational methodology, guists and psychologists – who wanted to Silvio Ceccato, machine translation, chimpanzee communication. believe with Chomsky that language was a prerogative – doubted that “an ape can truly create a sentence” (Terrace et al. an ape also useful for understanding some of 1979, p. 891) and claimed that “they show no Introduction the abilities involved in human language and unequivocal evidence of mastering the conver- Could an ape participate in a chat session vice-versa? How can these questions be sational semantics or syntactic organization of over the Internet? At first sight the question answered in a scientific manner? language” (Terrace et al. 1979, p. 901). They may seem silly, but I claim that it could – at Today science seems to have overcome the also said that sign language did not have a least in part – be taken seriously and in this old behaviourist stimulus–response bond and proper syntax and therefore was not really a paper I will try to show why. To begin with, finds itself in a somewhat better position to try language. Moreover they suggested that the let us step back a little and have a look at to answer this and related questions. But about Gardners were like parents with a baby: they some questions that the scientific commu- 40 years ago, when Ernst von Glasersfeld cre- saw and heard demonstrations of linguistic nity would probably accept as more “sound.” ated “Yerkish” – an artificial language for use capabilities that no one else could see or hear. Is language no longer the exclusive domain by apes in computer-mediated communica- The Yerkes Center plan was to build a of man? Can an ape create a sentence? Are tion (CMC) with machines and humans – the communication system with a simplified lan- explanations of language learning and use by situation was much more uncomfortable. guage, a keyboard, and a small computer to

32 Constructivist Foundations 2007, vol. 2, nos. 2–3 http://www.univie.ac.at/constructivism/journal/

historical; cognitive-psychological radical constructivism EMPIRICAL

explore computer-mediated communication Beside providing food and drink, the com- between humans and apes. The computer puter could respond to correctly formulated would record everything the ape typed on the requests by playing taped music or sounds, keyboard and there would be no subjective projecting movies and slides as well as open- bias as to what the ape had or had not typed. ing and shutting the above-mentioned win- This plan seemed a great idea to von Glasers- dow. Above the keyboard was a sturdy hori- feld and when Carpenter asked him if he zontal bar that Lana had to hang on to in order would like to design the special language and to switch on the system (Fig. 1). the computer system for handling it, he immediately accepted. In his turn, von Gla- sersfeld recommended that his long-term The Yerkish language research partner Pier Paolo Pisani3, a compu- ter specialist, also join the effort. Language as a communicatory system has After a number of conferences among the three indispensable characteristics (Glasers- members of the project team, in early winter feld 1977a, p. 66): a) it has a set, or lexicon, of 1970 a proposal was submitted to the artificial signs; b) it has a set of rules, or gram- National Institute of Health requesting funds mar, that governs the creation of sentences as for a 4-year period. In spring 1971 the grant sequences of lexical entries; c) its signs are was awarded (NIH grants HD-060164and used as symbols (Glasersfeld 1974). RR-00165). The team immediately began The lexicon of Yerkish was developed by designing and building the system and a few von Glasersfeld starting from a list of things weeks later everyone was introduced to the that would presumably interest a young subject of the research, a young female chim- chimpanzee (and the experimenters) and panzee called Lana (born October 7, 1970).5 could be available in the project. The words of In the first phase of the project a Plexiglas this preliminary vocabulary were about 150, cubicle the size of a small room was built on but in the beginning only 25 were put on the to an existing wall that had a window to the first panel of keys. Each key had an abstract outside of the Yerkes Center. One of the Plexi- design representing not a letter but the “word- Figure 1: Lana at the lexigram board. glas walls was dedicated to the keyboard, a design” for a single concept. Ernst von Gla- Photo Ernst von Glasersfeld. square array unit of initially 5x5 keys, with sersfeld coined for these word-designs the space for other units to be added as Lana got name “lexigrams” and created them by means more proficient. By sequentially pressing the of non-representational design elements to characteristics like being able to eat, drink, keys of the keyboard, code signals standing for emphasize their symbol-character (Fig. 2) groom, tickle, give things or make things hap- words were sent to the computer, which con- and to prevent critical linguists from saying pen were collected in the lexigram class tained the vocabulary and the grammar of that Lana recognized them just because they “autonomous actor” and divided into four Yerkish, the automatic parser for checking the were familiar pictures. Whenever Lana sub-groups: “familiar primates” (lexigram correctness of sentences, and the rules for pressed keys the respective lexigrams were Lana and lexigrams for the first names of activating a dispenser in response to requests projected on to a row of small windows above technicians and experimenters, like Tim or that Lana was to formulate in Yerkish word the keyboard, one after the other from left to Shelley), “unfamiliar primates” (lexigram vis- symbols (Glasersfeld 1995, p. 11). The com- right. This helped Lana to see how far along itor), “nonprimates” (lexigram roach) and puter itself and the terminal with a keyboard she was in typing the sentence – seven was the “inanimate actor” (lexigram machine). Sev- for the researchers were placed just outside maximum length of a sentence. Moreover eral lexigrams were assigned to classes desig- the room: from here the experimenters could projecting the lexigrams in this row was used nating relational concepts like the class “par- interact with Lana by typing sentences that to flash messages from the human trainers to titive proposition” (lexigram of), the class were displayed above her keyboard and they Lana and to make conversations possible. “semantic indicator” (lexigram name-of) and could also see how she was behaving during After compiling the lexicon of Yerkish, the the class “attributive marker” (lexigram the computer-mediated communication ses- lexical items were divided into classes. Since which-is). sion.6 Yerkish was designed on the basis of a “corre- Like the lexicon, the grammar of Yerkish Next to Lana’s keyboard was the row of lational” approach to language (Glasersfeld was also “correlational”: in fact von Glasers- food and drink dispensers, activated through 1970), the lexigram-classes were defined in feld derived it from the correlational gram- the computer; they would provide all sorts of terms of the functional characteristics of con- mar implemented from 1960 to 1970 in his food and drink (like apple, bread, chow, cepts and not, as in a traditional lexicon, in projects for the machine translation of banana, milk, juice etc.) and it was hoped that terms of morphology and the roles they English sentences (Hutchins 2000). As a con- Lana would learn to feed herself by means of would play in sentences (noun, verb, adjec- sequence the Yerkish grammar was an inter- request sentences typed on the keyboard. tive, etc.). For instance, items with functional pretive device and consisted of the rules of a

2007, vol. 2, nos. 2–3 33

historical; cognitive-psychological radical constructivism EMPIRICAL

relating. In natural lan- guages correlators are triad indicated in a variety of ways, either implicitly or correlator explicitely. A correlator is always a binary func- tion in that it links two mental operands – LH correlatum RH correlatum expressed in language by either single words or word combinations – Figure 3: A correlator as a binary function. and thus forms a new unit (a triad) called a “correlation” (Fig. 3). Implicit correlators are indicated in phrases or 6. Please Tim give milk to Lana sentences merely by the (C_00 Please (C_05 Tim (C_21 (C_17 juxtaposition of the two give milk) (C_22 to Lana)))) lexical items they link, 7. Tim give apple which-is red to Lana machine name-of candy out-of and “explicit” correla- (C_05 Tim (C_21 (C_17 give (C_31 apple tors are indicated by (C_10 which-is red))) (C_22 to Lana))) specific words (such as 8. Please Tim move out-of room propositions, conjunc- (C_00 Please (C_07 Tim (C_21 move tions, etc.). In the fol- (C_22 out-of room)))) lowing we will use “cor- 9. Please Shelley move behind room Lana eat tickle into relator” both for the (C_00 Please (C_07 Shelley (C_21 move relational concepts and (C_22 behind room)))) Figure 2: Lexigrams table developed by von Glasersfeld in 1971. for the linguistic devices Compare the sentences of example 1 and that express them. 2. They require two different correlators In 1974 the Yerkish C_01 and C_02 because the performed activ- primitive syntax that governed which lexi- grammar used by Lana operated with some ity they link with the actor performing them gram sequences (i.e., sentences) were to be 30 correlators. Five of these are, for example, are different: correlator C_01 (example 1) considered correct (i.e., any input that it correlators that connect an operand of the connects an autonomous animate actor with could interpret) and which mistaken (any class “actor” with an operand of the class a stationary activity whereas correlator C_02 input that it could not). There were three “activity.” In the following examples we des- (example 2) connects an autonomous ani- classes of sentences: statements, requests, and ignate the correlators by the letter “C” and the mate actor with a transferring activity. questions. Requests were differentiated from number used to identify them in the LANA Sentences 3 and 4, although very similar, the others by first pressing a key called project; a correlation is then written as list of require two different correlators C_05 and “please”; questions had to begin with a ques- correlators and operands using Polish nota- C_06 because the intended effect they link tion mark. To know when to check the cor- tion (prefix notation with operators left of with the agent causing it are different: corre- rectness of Lana’s typing, the computer their operands). Examples of sentences and lator C_05 (example 3) connects a causative needed a signal to indicate the end of a sen- their correlations in Yerkish: agent with a change of position whereas cor- tence, like a period. 1. Lana drink relator C_06 (example 4) connects a caus- The correlational approach to language is (C_01 Lana drink) ative agent with a change of state. based on the assumption that sentences 2. Tim carry Lana express in language sequences of mental oper- (C_02 Tim (C_14 carry Lana)) ations (attentional operations) performed at 3. Please machine give M&M Use of Yerkish by Lana the cognitive level (Ceccato 1964, p. 14). The (C_00 Please (C_05 machine (C_017 give (chimpanzee) most important among the mental operations M&M))) are obviously those that establish connections 4. Please machine make movie Lana’s training began with a panel of three or among conceptual operands and thus build (C_00 Please (C_06 machine (C_18 make four keys for learning a set of preliminaries, up complex structures. These relational con- movie))) such as that it was the sequence of lexigrams cepts, that Ceccato called correlators (Ceccato 5. Please machine give piece of banana in the row of windows above the keyboard et al. 1961, p. 36), are connective functions (C_00 Please (C_05 machine (C_17 give that counted, not their position in the panel, used at the mental level in the process of cor- (C_026 (C_25 piece of) banana)))) or that it was always necessary to press the

34 Constructivist Foundations historical; cognitive-psychological radical constructivism EMPIRICAL

period key at the end of a sentence. She first She went out of the cubicle and to the other Silvio Ceccato and the learned to press a single key in order to side of the transparent wall – which from obtain a piece of food or an M&M candy. Lana’s point of view could quite reasonably correlational structure Lana progressed rapidly in her training and be conceptualized as “behind room.” Lana of thought within the first 2 weeks of training she watched her clear the dispenser and immedi- learned to concatenate keys to form a stock ately typed: “Please machine give piece of The correlational approach to language that sentence like “Please machine give M&M” banana” (see example 5). On this and many von Glasersfeld applied in developing Yerkish (example 3) or “Please machine make other similar occasions, Lana, by means of was based on investigations of mental activi- movie” (example 4). original, spontaneous, and appropriate ties that Silvio Ceccato had begun in 1939 When she got the first 25-lexigrams panel, utterances, made it quite clear that she was (Ceccato 1964/1966). Together with a group she quickly learned to watch the row of win- indeed capable of forming concepts and able of scholars living in Italy he proposed from dows above the keyboard to check what she to use the lexigrams in language. Lana dem- the beginning to study thought and its con- had typed. It took her no time to find out that onstrated that she was able to participate in tents in terms of operations (Ceccato 1951, when she made a typing error she could erase a manner of living that we call language, i.e., 1953). Because of this “operational approach” what she had typed by pressing the period key that she could experience a recursive coordi- or “operational methodology,” Ceccato's (which made the computer cancel the input nation of behavioural coordinations, a pro- group was called the "Italian Operational because it contained an error). Lana learned cess which allowed her to have a recursive School.” His research activity was devoted to not only to use several stock sentences appro- influence on what she was experiencing. understanding the basic structure and priately, but also to build novel sentences that In September 1974 Lana’s lexigram board dynamics of thought production, to the were syntactically correct. consisted of 3 panels of 25 keys each development of an operational solution to the Unfortunately the director of the project (Glasersfeld 1977b, p. 128). The total of 1577 problem of semantics (connection of thought was convinced that understanding in com- grammatical 6-lexigram strings produced by and language) and to applications of opera- munication with Lana could be proved sta- Lana in this month can be assigned to 125 tional methodology in machine translation tistically: as a consequence, in order to col- sentence types. Four types are requests for experiments. lect statistical evidence of her “skills,” Lana food and account for 1288 tokens. Of the The basic assumption of operational was subjected to repetitive tests like a rat in a remaining 289 tokens, 228 represent 76 types methodology is that the essential function (or maze. It was clear from simply watching her that were spontaneously formulated by Lana activity) for the constitution of any mental behaviour, that, like a human child, she lost – none of them were produced as a result of content is the function of attention. In fact, it interest after the nth repetition and pressed training. In some cases their occurrence was is easy to notice that without attention we do keys without looking. Her statistics therefore even a rather imaginative transference of a not have mental content, i.e., no mental life. tended to be worse than those of rats. On the meaning acquired in a very specific context Our clothes are in contact with our body: do other hand, she did things that no rat could to a substantially different context. we feel them? Not if we do not pay attention ever do. When Tim, the graduate student These and similar facts persuaded von to them. We are typing on the computer key- who worked with her in these experiments, Glasersfeld that Lana was well able to com- board: are we aware of our finger touching a repeated the same question for the nth time, municate by means of symbols and also key? Not if we do not pay attention to it. Sim- she typed in the response: “Please Tim move clearly indicated that understanding com- ilarly we do not notice the noise of traffic out- out of room” (see example 8). This was above munication with Lana could not be tested side or understand what someone in the all remarkable because Lana had encoun- statistically but shown only by the appropri- group is saying if we do not pay attention. In tered expressions such as “out of,” “in front ateness of individual utterances. Unfortu- other words, the dynamism of physical inter- of,” and “behind” only in the context of nately they did not convince the conven- action between our organism and our sur- boxes and wooden blocks on a table and the tional experimental psychologists involved roundings proceeds on its own account with- notion that her room was a kind of box you in the LANA project of the necessity to devise out constituting any mental content unless we could “move out of” was entirely her own. more appropriate research methods. direct our attention to the functioning of the On another occasion, when Shelley Later experiments in other projects (Sav- different organs of hearing, touch, etc. appeared outside the Plexiglas cubicle, Lana, age-Rumbaugh et al. 1980) suggested that Attention, however, is not limited to this rushed to the keyboard and typed: “Please Lana had difficulties in expanding her lin- function of making present the functioning of Shelley move behind room” (see example 9). guistic domain7 beyond the limits of the other organs; in fact, attention is not applied Shelley, who had no idea what it could mean, domain of interactions through a computer continuously but for discrete intervals of did not take any action so that Lana, who was in which she had participated. On the other time, ranging from a tenth of a second to a expecting a specific intervention, threw up hand , a , though he had never second and a half: after this time, attention both her arms in an unmistakable human- been taught, learned Yerkish very well and detaches itself and after a short pause can be like gesture of despair and once more typed even some English by simply listening and applied again. In this way, as it is applied and the same phrase. Eventually Shelley looked participating in the laboratory environment detached repeatedly, it fragments into dis- at the array of dispensers and noticed that during his mother Matata’s training sessions crete pieces (so-called “praesentiata” or the one for slices of banana had got stuck. (Savage-Rumbaugh & Lewin 1994). recepts) the functioning of other organs and

2007, vol. 2, nos. 2–3 35 historical; cognitive-psychological radical constructivism EMPIRICAL

builds an oscillation similar to alpha waves in thought, according to Silvio Ceccato is always birds, reptiles, dishwashers or chairs this kind the brain or to the rhythmic contractions of a triad, called a “correlation,” composed of of explanation is useful with the word items of the heart. This conception of a pulsating two correlates assembled together by a corre- a natural language only for the purpose of attention and of discrete microunits of mental lator (Ceccato 1961, 1967). This triad has a describing a catalogue. activity has been recently confirmed by neu- characteristic dynamism, an order of opera- However, for users and developers of a lan- rophysiological experiments suggesting that tional precedence in that the first correlate, or guage – for instance children acquiring it “the seemingly continuous stream of con- first mental construct is the first in time to be from their interactions or machine transla- sciousness consists of separable building constituted (or activated) and is then held tion researchers using it in experiments – the blocks” (Lehmann et al. 1998, 2000). present (active) during the constitution of the main purpose is not description but the inter- A third function of attention could be correlator, which in its turn is held present pretation and production of sentences, i.e., of called the “generating” function. Why? during the constitution of the second corre- combinations of items. For this reason the Because it allows attention not only to be late, or second mental construct. The corre- usefulness of the explanation depends on its applied to other organs but to be applied to lates can be concepts, percepts or entire ability to accurately specify in operational nothing (a state of simple vigilance, an empty thoughts but the correlator is always a purely (functional) terms the items involved. This attention) or to its own functioning instead, attentional microunit, a mental category. characterisation in functional terms is exactly thus generating discrete attentional frag- Correlation constitutes the dynamism of what the correlational approach provides by ments that are not pieces of hearing, touch, thought, of which the triad is the smallest means of a minute and rigorous discrimina- vision or other sensorial activity but purely unit. The larger units of thought are obtained tion of a word-item’s eligibility as correlatum attentional microunits (attentional states). by using a correlation as a term in another or correlator within a correlation (Glasersfeld We would however never build a seem- correlation, which in its turn can become a & Pisani 1968, pp. 1–2). ingly continuous stream of consciousness, if part of a third correlation, and so on, until a To reify a simple correlation into a linguis- there were not: greater or smaller correlational network is tic form, each single element must be desig- 1. “Categorization” as the function which assembled. Pronouns and other words with nated by means of at least two indications: enables the mind to produce concepts by recall functions then make it possible for one to say what it is (referential function) and combining attentional states into more complete correlational networks to be reused the other to say what function it performs in complex combinations (macrounits). as elements in other correlations. 2. “Perception” as the function which enables the mind to produce percepts by ABOUT THE AUTHOR applying some results of categorization to Language and thought Marco C. Bettoni was born in Legnano recepts. (Italy) in 1952. Since September 2005 he is 3. “Correlation” as the function which A fundamental function of language consists Director of Research & Consulting at the enables the mind to assemble concepts and in ensuring that thoughts can be reified. One Swiss Distance University of Applied Sci- percepts into thoughts. way of reifying thoughts is by designating ences (www.ffhs.ch) where his main The operation of categorization received them, i.e., by establishing a viable correspon- research topics are knowledge coopera- this name because it produces mental con- dence between the polyphonic structure of tion, knowledge networking, distance coop- structs that Ceccato, in honor of Kant has thought and a linear sequence of perceivable eration, distance- and e-learning and called “mental categories.” Thus mental cate- items. communities of practice. gories comprise those mental constructs Given a background of an operational After receiving his master degree in which are made only by combinations of dis- methodology, with its attentional model of mechanical engineering in 1977 (ETH crete attentional fragments and do not con- mental contents and its correlational model Zürich) he has worked until 1991 for indus- tain anything originating from observation. of thinking, we are now in a position to trial (Rieter, Siemens), banking (UBS) and Examples of mental categories are the more or explain language in a completely different academic (ETH) organizations in the less complex combinations (concepts) of way: an operational way! domains of machine design, engineering attentional microunits designated by words Traditional grammars explain, for education, IT management, IT development like “thing,” “object,” “beginning,” “end,” instance, vocabulary items (the lexicon) by and knowledge engineering (Artificial Intel- “part,” “whole,” “element,” “group,” “set,” assigning them as elements to classes such as ligence). “point,” “line,” “and,” “or,” “singular,” “plural,” “noun,” verb,” “adjective,” etc. by virtue of Between 1991 and 2004 he was Professor “space,” “time,” “number,” “1,” “2,” “3,” etc. some feature that is identified as common to for Knowledge Technologies at the Basel Each category is differentiated from the oth- all the members of a class. Since many mem- University of Applied Sciences (FHBB). In ers by the number of discrete attentional bers do not display all the required character- June 2003 ETH Zürich (www.ethz.ch) states (fragments) which it comprises and by istics of their class, grammars usually proceed appointed him as “guest researcher” for the way in which they are combined. by subdividing a class according to the specific investigating the role of knowledge-ori- The operation of correlating is what con- or “exceptional” features of certain items. One ented cooperation in Knowledge Manage- stitutes thinking. It assembles the attentional might call this the botanist’s, zoologist’s or ment. units in a binary tree. The basic structure of retailer’s approach: as with trees, flowers,

36 Constructivist Foundations historical; cognitive-psychological radical constructivism EMPIRICAL

the correlation (correlational function), able (Ceccato & Zonta 1980, p. 78). For means of appropriate Yerkish sentences she whether that of correlator or that of first (left example, consider the expressions “to eat an made it quite clear that she was not only capa- hand) or second (right hand) correlatum. In apple” and “to eat an hour” (for instance in: ble of forming concepts and of using lexi- order to supply these indications, “You may also need to eat an hour before train- grams but also able to participate in a manner can offer basically two means: on the one ing…”): without a general culture which of living that we call language, i.e., that she hand they use a particular, phonic or graphic allows us to distinguish between food items could experience a recursive coordination of material (spoken or written words), and on and time intervals the correlation expressed behavioural coordinations, through which the other hand they use the order of succes- in the previous sentences could not be cor- she could recursively influence what she was sion into which this material is put (word rectly produced or interpreted. experiencing. sequence). Only by providing these six indi- As a consequence of this tight connection The key question in her language acquisi- cations can we identify two expressions such to knowledge and experience, language can- tion is how Lana learned the appropriate syn- as “green bottle” and “bottle green” as two dif- not merely be considered as a strictly organ- tactic forms and word order for expressing ferent correlations or units of thought. ised and classified system of words and complex relations in Yerkish as well as a kind Mostly a correlation will be designated by phrases: it must also be approached as an of common sense background knowledge. employing two or three words (or whole sen- extremely intuitive arrangement of things, How did she correctly concatenate the lexi- tences in a correlational net), which is to say, intuitive in its production and intuitive in its grams? How did she learn to do that? Was it the required indications are distributed interpretation (Glasersfeld 1965, XIII–1). merely due to good training practice on the among two or three words, but usually the This is not to say that language does not part of the primatologists? Our hypothesis is correlations that occur more frequently are include logical functions and logical implica- that the success of Lana is primarily due to the indicated by only two words, one for the first tions, but it embraces very much more: for fact that she learned the grammar rules of and one for the second correlatum, whereas instance, interpretations that are “correct” Yerkish. How? By matching her conceptual the correlator remains tacit. How can we merely because they are much more probable abilities with the correlational structure of understand a correlation of this kind in which than others, given our experience of the world Yerkish. As a consequence we see the success there is no explicit word for the correlator? In we live in and our knowledge of how certain of Yerkish during the past decades (originally some cases the correlator is indicated by things are related (notional sphere). with Lana since 1973 and later with other changes in the form of the designation of one apes, such as Kanzi) as a demonstration of the of the correlates but in all other cases the indi- viability of the operational methodology that cation of the appropriate correlator has to be Conclusion is its foundation. We hence propose that deduced from a wide-spread knowledge, a Lana’s conceptual system be considered as a common cultural heritage behind any lan- Since the great apes are the closest relatives to correlational system in which the operation of guage, for which Ceccato has coined the terms human beings, experiments in teaching them correlation is what constitutes the chimpan- “Notional Sphere” (Ceccato 1961 et al., p. 62) a language can shed some light on the human zee’s thinking and an attentional system deliv- and “Constellation” (Ceccato 1961 et al., mind. Although Lana could not speak, she ers the mental contents that correlation p. 63), which were precursors of methods of learned to communicate in the Yerkish lan- assembles into triads and networks. Since no knowledge representation such as frames and guage. Lana was the first ape to work with a other assessment or explanation of Lana’s scripts in early Artificial Intelligence research computer keyboard, the first to show that performances has considered these funda- (Sowa 1984, p. 128). Knowing how certain chimpanzees could form syntactically correct mental issues, we strongly suggest that a new things are related allows the designation to be sentences, could recognize written symbols, research project or program be conducted to made more efficiently by reducing the num- could read and could complete incomplete investigate the above-mentioned hypothesis ber of explicit indications, thus making com- sentences appropriately. On many occasions of the importance of Yerkish in Lana’s success munication more rapid, flexible and adjust- within the context of the LANA project, by in language learning.

Notes matic translations projects, first in Italy cludes a six-minute feature on the and later in USA. chimpanzee Lana, seen during training 1. Yerkes National Primate Research Center, 4. http://www.ncbi.nlm.nih.gov/entrez/ sessions of the LANA project where Lana http://www.yerkes.emory.edu/ query.fcgi?CMD=Display&DB=pubmed communicates via her keyboard with re- 2. Alan and Beatrice Gardner about Washoe: 5. The Georgia State University Language searcher Tim Gill. The movie can be see Gardner, R.A. & Gardner, B.T. (1969) Research Center, http://www2.gsu.edu/ viewed at http://www.greatapetrust.org/ and (1971) ~wwwlrc/chimps.htm research/general/lana.php# 3. Piero Pisani, a computer scientist who had 6. “The Amazing Apes,” a TV program pro- 7. Also mentioned in Maturana and Varela worked with von Glaserfeld in his auto- duced in 1977 by Bill Burrud, which in- (1987), pp. 215–217

2007, vol. 2, nos. 2–3 37 historical; cognitive-psychological radical constructivism EMPIRICAL

References of the conference on computer-related Lehmann, D., Koenig, T., Pascual-Marqui, R. semantic analysis, December 3–5, 1965, D., Koukkou, M. & Strik, W. K. (2000) Ceccato, S. (1951) Language and the table of Las Vegas USA. National Science Founda- Functional tomography of EEG Ceccatieff. Hermann & Cie: Paris. tion & Office of Naval Research, U.S. Air microstates of visual imagery and abstract Ceccato, S. (1953) Consapevolizzazione Force, pp. XIII 1–24. thought: Building blocks of conscious dell’Osservare, mod. 3", Atti del Con- Glasersfeld, E. von (1970) The correlational experience. Brain Topography 12: 298. gresso di Metodologia. Ramella: Torino. approach to language. Thought and Lan- Lehmann, D., Strik, W. K., Henggeler, B., Ceccato, S. (1964/1966) Un tecnico fra i guage in Operations 1(4): 391–398. Koenig, T. & Koukkou, M. (1998) Brain filosofi. Vol. 1 & 2. Marsilio: Padova. Glasersfeld, E. von (1974) The Yerkish lan- electric microstates and momentary con- Ceccato, S. (1964) A model of the mind. guage for non-human primates. American scious mind states as building blocks of Methodos 16: 3–78. Journal of Computational Linguistics 1, spontaneous thinking: I. Visual Imagery Ceccato, S. (1967) Correlational analysis and microfiche 12. and abstract thoughts. Int. J. Psychophysi- mechanical translation. ORIGINAL Glasersfeld, E. von (1977a) Linguistic com- ology 29: 1–11. CITATION. Reprinted in: Nirenburg, S., munication: Theory and definition. In: Maturana, H. R. & Varela, F. (1987) The tree Somers H. & Wilks, Y. (eds.) (2003) Read- Rumbaugh, D. M. (ed.) Language Learn- of knowledge. The biological roots of ings in machine translation. MIT Press: ing by a Chimpanzee: The LANA project. human understanding. Shambhala: Bos- Cambridge. Academic Press: New York, pp. 55–71. ton. Ceccato, S. & Zonta, B. (1980) Linguaggio, Glasersfeld, E. von (1977b) The Yerkish lan- Rumbaugh, D. M. & Washburn, D. A. (2003) Consapevolezza, Pensiero. Feltrinelli: guage and its automatic parser. In: Rum- Intelligence of apes and other rational Milan. baugh, D. M. (ed.) Language Learning by beings. Yale Universitiy Press: New Haven. Ceccato, S., Beltrame, R., Glasersfeld, E. von, a Chimpanzee: The LANA project. Aca- Savage-Rumbaugh, E. S., Rumbaugh, D. M., Perschke, S., Maretti, E., Zonta, B., & demic Press: New York, pp. 91–129. Smith, S. T., & Lawson, J. (1980) Refer- Albani, E. (1961) Linguistic analysis and Glasersfeld, E. von (1995) Radical construc- ence: The linguistic essential. Science 210: programming for mechanical translation. tivism: A way of knowing and learning. 922–925. Feltrinelli: Milan & Gordon & Breach: Falmer Press: London. Savage-Rumbaugh, S., & Lewin, R. (1994) New York. Glasersfeld, E. von (2000) Silvio Ceccato and Kanzi: The ape at the brink of the human Gardner, B. T. & Gardner, R. A. (1971) Two- the Correlational Grammar. In: Hutchins, mind. John Wiley Publishers: New York. way communication with an infant chim- W. J. (ed.) Early years in machine transla- Sowa, J. (1984) Conceptual structures. Infor- panzee. In: Schrier, A. M. & Stollnitz, F. tion. John Benjamins: Amsterdam, pp. mation processing in mind and machine. (eds.) Behavior of non-human primates. 313–324. Addison-Wesley: Reading. Vol. 4. Academic Press: New York, pp. 117– Glasersfeld, E. von & Pisani, P. P. (1968) The Terrace, H. S., Petitto, L. A., Sanders, R. J. & 184. Multistore System MP-2. Scientific Bever, T. G. (1979) Can an ape create a sen- Gardner, R. A. & Gardner, B. T. (1969) Teach- Progress Report, Grant AFOSR 1319–67. tence? Science 206: 891–902. ing sign language to a chimpanzee. Science Georgia Institute for Research: Athens. 165: 664–672. Hutchins, J. (ed.) (2000) Early years in Glasersfeld, E. von (1965) An approach to the machine translation. John Benjamins: Received: 30 September 2006 semantics of propositions. In: Proceedings Amsterdam. Accepted: 28 January 2007

38 Constructivist Foundations