WOLFGANG VON KEMPELEN AND HIS
„MECHANISM OF HUMAN SPEECH


| SLOVAK |BRATISLAVA - PRESSBURG - POZSONY |

 

Wolfgang von Kempelen (1734-1804), a native of Bratislava (Pressburg), was in his time an unsurpassable inventor, mechanic and automata constructor, polyhistor, artist and dramatist, even counsellor of Vienna imperial court. He is known to have acquired a world fame namely due to his ingenious and rumoured chess automaton, constructed in 1770. However, many others of his inventions and real miracles have been known to international public. Let us mention briefly a pressurized water-supply, which sometimes supplied Bratislava castle with water (the remains of „Kempelen's dwell“ may be found now at the slope of the castle hill), a pendulum bridge crossing the river Danube, ingeniously constructed fountain in Schönbrunn, allegedly functional till today, a printing machine for the blind (enabling a.o. the Fräulein von Paradies, a musician and writer who lost her sight, to type-set her works herself), then water pumps for drawing the water from flooded shafts of salt-mines in Transylvania, which he directed, etc. A Great Soviet Encyclopaedia values Kempelen most as an author of the project of canal Danube-Adriatic Sea. Only to display his extensive talent and organizational capacities we mention, that in emperor's service he excellently accomplished e.g. a leadership of reconstruction of Buda castle, a movement of Trnava University library and printing plant to Buda or even the wipe-out of terroristic gangs in Banat, with which Maria Theresa could not have managed and a settlement of this region by new inhabitants. The empress awarded this service with a lifetime revenue of 1000 guldens for her councellor von Kempelen. However, before long a new sovereign came to the throne and Kempelen lost his rent.

Also Kempelen's landscape engravings were famous - he even became a member of Imperial-Royal Fine Arts Academy - as well as dramas, of which Perseus and Andromeda was the most known. However, today we know, that the most precious of all what Kempelen's miraculous hands made, was his „speaking machine“, although it was not so successful with the public than the chess automaton. Let us mention, that while the Turk the Chess-player was constructed in six months since Kempelen promised it to Her Majesty Maria Theresa, the construction of the speaking machine demanded more than twenty years and still he could not consider it to be finished. This machine had to produce the letters and syllables, even the words and short sentences almost in every European language. Related to the work a treatise was published by J. V. Degen in Vienna in 1791 in German and French titled Mechanismus der menschlichen Sprache nebst Beschreibung einer sprechenden Maschine - Le Méchanisme de la parole, suivi de la description d'une machine parlante (Mechanism of Human Speech with the Description of a Speaking Machine), which proves, that Kempelen's speaking machine was neither a mystification nor a mechanical toy, while the principle, which underlied a construction of a chess automaton still remained unknown. In any case the book Mechanism of Human Speech confirmed outstanding Kempelen's capacities to reason profoundly in abstract philosophical concepts and to affiliate them with exact technological and constructional thinking.

The first attempts to imitate human speech are very old, coming even from antiquity. On the Lesbos isle the sounds of a human voice came out of the mouth of Orpheus' oracle. In 999 a French monk Gerbert - later a pope named Silvester II - made a bronze head, which „spoke“. A scholar-philosopher and a scientist Albertus Magnus is also said to have constructed a „speaking head“. Presumably all were only mystifications, resulting in the 18th century, noted by a particular fancy for mechanical toys and automata, sometimes of an astonishing perfection. Nevertheless, in the 18th century some serious scientific attempts to imitate human speech by means of machines appeared, joined with a study of real speech processes. The representatives were W. von Kempelen as well as Ch. T. Kratzenstein, professor of physiology at the University in Hall and in Copenhagen. The seriousness of the trials is proved by the treatises left by their authors. In 1773 Ch. T. Kratzenstein constructed a machine, which mechanically produced vocals. His treatise was published in 1780 titled Christiani Theophili Kratzensteinii, ... Tentamen resolvendi problema ab Academia scientiarum imperiali petropolitana ad annum 1780 publice propositum: 1. Qualis sit natura et character sonorum vocalium a, e, i, o, u, tam insigniter inter se diversorum; 2. Annon construi queant instrumenta ordine tuborum organicorum, sub termine vocis humanae noto, similia, quae litterarum vocalium sonos exprimant, in publico Academiae conventu, die XIX septebris 1780, praemio coronotum, Petropoli (French translation Sur la naissance de la formation des voyelles. Journal de Physique, 21, 1782, p. 358-380). Nonetheless, the extensive treatise Mechanism of Human Speech by W. von Kempelen represents a unique work in the history of phonetics, which profoundly influenced later development of the science of sonic aspect of language and which thus deserves as great publicity as possible. The Slovak translation of the work supplemented by a rich commentary was published in 1990 by Tatran publishing house (translated by Slavomír Ondrejoviè and Peter Ïurèo), the Hungarian one was issued one year earlier. Let us consider more closely a thematic range and solutions offered by the book.

In opening chapters W. von Kempelen enters the contemporary discussion on origin and nature of a language. Although in a preface he modestly mentions, that he only exploits an occasion to introduce several ideas, which occured to him while constructing the speaking machine - „hence, take them just as such - the incidental ideas“ -, we find here astonishingly actual claims. The very first sentence of the book suggests much: „A speech in the broadest sense of the word has a capacity to communicate one's feelings and ideas to others by means of signs“. It is not by accident, that this definition introduces the feeling in the first plane. Speech is comprehended by Kempelen more extensively - in a semiotic sense he discerns „simple“ signs, used by animals and „complex“ ones, which have to be learned and the system of which is opened.

According to Kempelen if animals can „think“, they can „speak“ as well. Hardly a Kempelen's supposition can be objected, that it is a dog among animals who possesses the most evolved images and who can express them best (in more detail compare S. Ondrejoviè: Wolfgang von Kempelen a pes. Kynologická revue, 1992, No. 7). His observations of communicative behavior of animals presented in the book must have been very revealing in his time. When Kempelen wrote his book, such a thinker as Herder claimed the monkies being dumb and developed out of this presumption far-reaching teleological conclusions („The monkies would disgrace a human voice“).

In the second chapter of the book the writer conveys an admirably informed survey on scientific discourse of his contemporaries concerning the origin of a language, in which A. Court de Gébélin, Ch. de Brosses, J. P. Süssmilch and J. G. Herder participated, as well as many others. The history of linguistics discloses, that the latest two ardently debated whether a speech is of a human or of a divine origin. Kempelen inclines towards a Herder's opinion and supplements his explanations with arguments based on his own observations. His attitude is worth mentioning. While Süssmilch, advocating an opinion on divine origin of language, arguments that man lacking the speech would not be able to reach a demanded level of rational development and that man lacking the developed reason would not be able to create a speech, hence as a result he had to have received it from his Creator, Kempelen supports an idea, that a human reason is born simultaneously with a speech and he accentuates an evolutional character of a language progress from more simple forms to the most complex ones. The modern attitude actually says, that a thinking in a specific human form, a conceptual thinking, is born only with a language. Certainly not in a sense, that a language would have been the only one basis, from which a thinking is born. Our conceptual thinking sources in a pre-conceptual (sensory-imagery associative thinking a.o.) one, which does not possess a specific appearance of language (compare Poršnev, B. E.: O zaèiatkoch ¾udských dejín. Bratislava 1979; Horecký, J.: Príspevok k teórii o vzniku a vývoji jazyka. Slovo a slovesnost, 42, 1981, s. 109-114).

In his book Kempelen also copes with an opinion of Court de Gébélin, who claimed the languages being of divine origin, because all they derived from the only one „divine protolanguage“. The words which nominate particular things were not chosen by man accidentally, but following „certain rules“, which have to be discovered by etymology. However, Kempelen objects not only Gébélin, but all advocates of protolanguage theories, that the occurence of some words and their variants in several languages does not prove their common origin. He deducates very accurately it may be the result of mixture of languages. W. Kempelen was well-versed in linguistic knowledge and unlike majority of contemporary scholars who were collecting the „primordial words“, he mastered the Hungarian language, what was not common in his time. This supported him in his skeptical view towards the idea on common origin of languages. E. and K. Zwirner, the founders of phonometry even presume, that it was just this Kempelen's critical attitude towards the comparative linguistics, what determined profoundly the evolution of phonetics: „We must mention, that roots of experimental phonetics are in Kempelen, i.e. earlier than historical-comparative methods evolved and not only the first works on physiology of speech, but also phonetical research of the whole 19th century thank for their origin to these thoughts“ (Grundfragen der Phonometrie. New York 1966, p. 86). Having compared Hungarian and German words of the basic vocabulary and namely cardinal numbers of some Asian and African languages he fully rejects the theory of protolanguage. The indication, that Finno-Ugric Hungarian language is not relative with majority of European languages, was justified, but some of his claims concerning the (non)relationship of languages would not be agreed by all. At least he would have got into a conflict with the advocates of „nostratic“ hypotheses. „Nostratic“ („boreal“) hypothesis inspired first of all by works of V. M. Illiè-Svityè adjoins a series of language families, e.g. Indo-European, Kartvel, Uralic, Altaic, Dravidian, Semito-Hamitic a.o. to one primordial language. A similar attitude had also N. S. Trubeckoj, an author of a known book Grundzüge der Phonologie, who maintains, that the languages are getting closer by common contact and it has no sense to look for their common roots. We may leave this question opened for later research and considerations and let us turn to at least one particularly modernly sounding point from Kempelen's explanations: „If we want to explore the differences among the languages, it is not enough to stay by words but the syntax and structure of a language must be considered as well“ (p.48). These are more fixed and less prone to changes. Turning to history of linguistics we find out, that Kempelen was not the first one to establish the condition of comparing the structure of languages for postulation of their relationship. Presumably it was an orientalist Job Ludolf (1624-1704), an author of Ethiopian grammar and dictionary, who at the end of the 17th century articulated a principle, that we must find not only several common words to declare the languages as relative but we have to prove the similarity of their grammatical structure. A. Arens writes in his book Sprachwissenschaft (Der Gang, ihre Entwicklung von der Antike bis zu Gegenwart. Freiburg - München 1956): „We must admit, that this first knowledge was left unused till the second decade of the 19th century“ (p. 88). The same may be said about Kempelen's contemplations on languages relationship.
Kempelen turns with direct question not lacking irony to advocates of the theory of protolanguage: what was the language like, how did it sound and to which of currently living languages did it come closest? He deducates himself genuinly, that „we lack the evidence for more responsible conclusions“. Also on other places he accentuates the importance of the grammar, even of syntax - the research of the vocabulary is insufficient.

These considerations merge to the third chapter, in which Kempelen presents another bizarre proof resulting from the belief in divine origin of a language and in the primacy of Hebrew. It is a treatise Alphabeti vere naturalis hebraici brevissima delineatio... by F. M. B.ab Helmont, according to which the shape of written letters in the Hebrew language fully corresponds with the position of a tongue producing these letters. Having accomplished a detailed research of functions of speech organs and a tongue W. Kempelen cannot help but appreciate this treatise as a product of „ill phantasy“ and as „uncomprehensive and funny“ (Slovak translation p. 14). However, let us mention, that the relation between the vowel and a letter (grapheme) will come in the foreground also much later, even in highly serious works. For instance authors of a well-known Visible Speech R. K. Potter, G. A. Kopp, H. C. Green (New York, 1947) also tried to designate particular vowels or vowel types by simplified spectrograms according to which a machine should have been able to produce a vowel. If they were successful, the deaf would get a real possibility to „speak“. But all trials to transform a spoken speech into a series of letters and vice versa, to synthetize them out of symbolical or graphic records fail due to a problem of invariants (compare Brekle, H. E. - Wildgen, W.: - In: Mechanismus der menschlichen Sprache nebst Beschreibung einer sprechenden Maschine. Faksimile-Nedruck. Stuttgart-Bad Cannstadt 1970, p. VII-XLV.). By these kinds of trials one is looking for models for a code process, similarly to reading or writing. Kempelen constructing his speaking machine pursued a more modest and realistic aim - to realize „only“ a physical synthesis of vowels disregarding other kinds of media (i.e. also graphemes).

The third chapter presents exhaustive descriptions of particular respiratory and speech organs and their functions. Kempelen's system required to deal with all sounds produced by these organs, so we may find here descriptions of smacking, grinding one's teeth, kissing, snoring etc. sometimes resulting in a certain kind of physiologism. It is an important indication of his, that none of the organs - the vocal cords excluding - evolved precisely for a speech, similarly to fingers, which were not determined to play the flute nor eyes to read. For Kempelen this is one of proofs, that man must have been creating his speech and getting accustomed the said organs continually for this activity.

As to the question to what extent the descriptions of speech organs and their operation are adequate we may turn to E. Brücke, professor of Vienna University and a court councellor, who referred to Mechanism... in his book Grundzüge der Physiologie und Systematik der Sprachlaute für Linguisten und Taubstummlehrer (Vienna 1876) 65 years later saying, he had never read a better book concerning the physiology of speech sounds. Nevertheless, the most important is, that it was Kempelen, who revealed a today well-known basic function of cavities for creation of speech sounds. Before him it was presumed, that all important motions for sound production are taking place in larynx and the processes performed in cavities over larynx are of a secondary importance. However, the physiologist D. Dodart found out (1703), that the significance of a trachea for a speech sounds production resides in carrying the air from lungs to larynx (since Galen it was believed, that trachea actually takes part in voice creation). Kempelen's experiments cleared the way for a new era of phonetic research. The fact, that his speaking machine based on the imitation of speech tract was really functional, positively influenced the phonetics. G. Ungeheuer estimated an importance of Kempelen's speaking machine in a work Elemente einer akustischen Theorie der Vokalartikulation (Berlin - Göttingen. Heidelbert 1962): „This successful vocal synthesis steered an attention of the 19th century scholars to acoustic function of cavities and suppressed the observations of the functioning of uvula, soft palate, spurious vocal cords etc. The way to deeper phonetico-acoustic view on phenomena of vocal articulation was made free“ (p. 3).

The fourth chapter is devoted to a speech sounds (letters) production. The terms (and notions) of a vowel and a letter were freely exchanged in the 17th and 18th centuries, what is reflected also in W. Kempelen's terminology and explanations. However, in some cases they are distinguished and their mutual relation is examined. Next to any speech sound the ways of its articulation and the most frequent articulate faults are examined in detail. The supplement to consonants is a remarkable section about combinability of consonants at the beginning and end of the words. The evidence come from many European languages including Slovak („Slovan“) one.

In the fifth chapter we learn more about the construction of a speaking machine, preceded by extensive periods of exploration. Kempelen writes: „It is enough to say I have repudiated so many components, that a strong horse would scarcely draw them away“ (Slovak translation p. 210).

Kempelen's machine was not in fact a „speaking machine“ in a real sense of the word, but a mechanism for production of speech sounds, words even sentences. An accurate hearing was needed to operate the machine, because vocals and many consonants had to be continually controlled. A degree of openess of a mouth cavity and the period of oscillations of a sound with consonants were not mechanically determined. Among the operator, listener and the machine a sort of a game with certain rules took place. It was not caused by an inadequacy of a machine as a synthetic model, on the contrary, the production of speech sounds in people is controlled by the same feedback of listening. But, the synthetic way is not visually steered, what eliminates the possibility for deaf-mute to speak. However, Kempelen had not even set this goal for himself, although his interest in such issue may be traced from several text passages (Slovak translation p. 55).

Kempelen's speaking machine may be interpreted as a cognitive model. Disregarding purely psychical components following parts (stages) may be identified in a communicative process: neurophysiological, organogenetical (articulative), auditorial (external, middle and internal ear) and again neurophysiological (competent sections of central nervous system) in perception. Kempelen's machine serves as a model for articulatory and partly acoustic stage, and his significance resides especially in this double modelling. A pure articulatory modelling would not possess a genuine scientific value, it would be simply only an imitation of observed articulatory processes. Acoustic modelling uncovered a relevance of properties of cavities and provided a scientific information on speech processes. This information is a relatively universal one, hence of a high degree of predictive power. One may use it for description of articulatory processes in terms of general acoustics or specific properties of the cavities (H. E. Brekle - W. Wildgen: op. cit.). With Kempelen this principle was not theoretically justified, he only faced a technical problem to reach his set aim - to simulate a human speech - with the material at his disposal: pipes, tubes, wooden boxes, cartoon etc. Nonetheless, the simulation of articulation under the given conditions made him to make a said step, justified only in 1829 by R. Willis (G. Ungeheuer: op. cit.).

Later, too, some machines appeared modelling the human vocal tract and possessing the bellows thrusting the air, reeds and acoustic resonators imitating the lungs, vocal cords and cavities. Many of them may be labeled more or less as imitations of Kempelen's machine. For instance „speaking machine“ of Posch (1828), of J. Faber (1842) or of Reis (1937) (more compare Flanagan: Speech Analysis, Synthesis and Perception. Berlin - Heidelberg - New York 1965, p. 196). Some of these machines allegedly pronounced words very well. Following the exact instructions in the book Mechanism... sir Ch. Wheatstone constructed a similar apparatus and presented it in Dublin in 1835. Inspired by Wheatstone even A. G. Bell, later inventor of telephone, tried to imitate it. Kempelen's mechanism was exhibited for the last time on May 16, 1923 by sir R. Paget in London (J. L. Flanagan: op. cit.).

Eventually we may affirm, that today speech synthesizers often utilize basically similar principle of sound reproduction, only instead of mechanical devices the electronical ones are used. The vocal cords are replaced by an impulse generator serving as a source of speech sounds; for consonants like „s“ or „t“ the noise generator is used. Electrical analogues of real sounds are directed into electrical devices of two kinds. The first one is based on articulation modelling, on imitation of diverse parts of human vocal tract by electronic devices. The vocal tract represents in itself a series of interconnected sections, consisting of capacitor and inductive part. Control signals in synthesizers are constructed according to analogy with physiological positions of tongue and lips. The first speaking device of this kind able to pronounce the whole sentences was constructed in 1957 by B. Rosen who named it DAVO (Dynamic Analogue of Vocal Tract). It consisted of 13 control units, any of which imitated the relevant part of vocal tract. The quality of speech reproduced by DAVO is said to have been outstanding (G. Ched: Sound. From Communication to Noise Pollution. New York 1970, s. 39).

The devices for speech synthesis of the second kind are based on acoustic modelling, where not the physiological structure of a vocal tract is followed, but such parametres as formant frequency, amplitude, sound duration etc. are used. Unlike the method of articulation modelling these are the trials to reproduce a sound out of spectrographic records. The first acoustic synthesizer publicly performed was made by an English scholar B. Lawrence in 1952.

The development of synthesizers continued since then first of all in laboratories of the USA, England, Japan, Russia, in other countries including ours, too. The rules are being sought, which would exactly determine mutual influences among the speech sounds and the ways, by which the transitions between them realize. These rules are supplemented by others, which have to secure a „logic“ of a speech, an accurate distribution of accents, a usual rhythmical plan etc.

Mechanism of Human Speech reveals, that Kempelen had a sound knowledge in philosophy of language and his considerations concerning the relationship of languages are at least remarkable. With this knowledge he could have compete successfully with other scholars, who took part in a competition of the Berlin Academy of Science in 1769 on a thesis „Can people with their innate capacities create a speech?“. We must realize, that in the time of the publication of the Mechanism... F. Bopp, one of the founders of comparative Indo-European linguistics, was only to be born in several months. However, especially the phonetic and logopaedic component of the book attract attention. Even if observation and study of speech organs physiology and their function boasted a long tradition in Kempelen's time, he lacks direct predecessors. A lot was explored by Leonardo da Vinci in a monumental work Quaderni d'anatomica, but Kempelen could not know him, because Leonardo as a phonetician remained unknown till the end of the 19th century (compare S. Ondrejoviè: Leonardo da Vinci o ¾udskom hlase a hovorení. Vesmír 67, 1988, No. 10, p. 577-579). Whether we agree with the Zwirners, that W. von Kempelen was a founder of experimental phonetics or we stay by abbé Russelot (1891), whether we consider him the real founder of logopaedia as a science and of practical cybernetics or not, we cannot deprive this Bratislava native of a significant position in a history of all mentioned branches. In any case the Mechanism of Human Speech remains undoubtedly the peak of all, what people knew about speech sound formation 250 years ago.*

 

Slavomír Ondrejoviè