Stenotype Italia s.r.l. ::: Official web site :::

On the basis of study and rational analysis of a language, its basic principles and the valid elements necessary to realize any technique of rapid writing can be identified. By applying these principles strictly we can manage to create a relationship between the various languages, from the phonetic-orthographic point of view, which makes it possible realize a common method of computerized management for stenotyping.

     In 1987 at the Intersteno Congress held in Florence, Stenotype Italia had a stand in which it publicly demonstrated the use a of real-time computerized stenotype machine in Italian: operating system SM Dos, computer with XT processor.
     I repeat that this was in 1987, exactly 17 years ago.
The method used already in 1987 to achieve real-time was the method created by myself; from it, on its architecture, the relevant software has been realized first in Dos and now in Windows.
     In 1987 none of the Software for the English language provided in practice for stenotyping of the spoken language in real-time. This represented a long-sought goal which only now, with the evolution of technology and with increasingly fast computers, it has been possible to attain.

     What is the trick? you may ask. Is it possible that Prof. Melani is so expert as to reach, over ten years before the Americans, wizards in stenotyping and electronics, this important goal? And using means that are, it might be said, so antiquated?
     That Prof. Melani is expert is true, as I, the person directly involved, can assure you, but the trick exists, and it is that of the language!
     When I began to create my method of stenotyping (it was in 1979) I had a great advantage: that of being able to count on the experience of the past, starting from the Michela method created in Italy in 1880 and still today in use at the Senate of the Republic of Italy; then the American Stone Ireland method currently in use, with some variations, in the united States and created in the early years of the 20th century, and lastly the Granjean method developed and introduced in France around 1920.
     My target however was that of realizing a method that would be fully compatible with electronic processing, a technology from which the old methods had not been able to benefit, for anagraphic reasons.

This solution however was obligatory because the English language (as well as the French) is in practice composed of two distinct, different languages: an oral one and a written one, between which exist phonetic rules that are not entirely objective and in which the vowels often assume different values. An example is provided by the word TITANIC, where the first I is pronounced AI and the second is pronounced short I. Without any logical reason. And, as we know instead, computer logic cannot accept ambiguous information.

The software realized in the 70s was designed to carry out this function, that of automatically transcribing the notes entered phonetically. Note that, when I say “entered phonetically”, I am referring, to give an example, to the world “room” in English, which will be entered on the machine as RUUM and not ROOM, as it is spelled.
The notable difference between phonetics and spelling, the different orthographic interpretations of the vowels, made it practically impossible to use any other approach.
The software is not required to carry out complicated processing, but the problem consists of words not included in the dictionary. In this case the software, not finding any correspondence to the stroke entered, transcribes the stroke as is in code, in the code that the machine produces in transcribing a discourse.

In the orthographic method the word is constructed, just as it is with the typewriter.
If we wish to write the word CASA (house) in stenography, we strike the four letters of the word. We do the same with the stenotype machine.
No dictionaries of any kind are used. We have only the letters available. In stenotyping we write CASA in a single stroke, pressing contemporaneously the corresponding letters: CASA.
How can the software write a word correctly if there is no dictionary in which the word is included? It’s easy.
There exists in the software a table of meanings: A = A, C = C, S = S, etc.
Consequently, the words are not inserted into the table of meanings, but only the elements of them, that is the vowels and consonants. Obviously, if the vowels had different sounds as in English, this principle could not be applied.
It is therefore easy to understand that all words and thus also those of the various dialects can be written without the need to insert them in the data base, because, as already stated, words are built with the existing basic elements which are in the meaning table.

That is, there are two types of language which can be classified as:
- orthographic languages;
- non- orthographic languages.
Italian, Spanish and Portuguese, as well as many others, as Greek, are basically orthographic languages, i.e., they are written exactly as they are pronounced; the relationship between phonetics and writing is identical.
The non-orthographic languages, on the contrary, are distinguished by the fact that the written language does not correspond orthographically to the phonetics, but differs from it substantially. Important examples of these languages are, as previously mentioned, English and French.

Applying the procedure of traditional stenography, stenotyping, for the English and the French language, uses a phonetic transcription of words. The function of the software is then that of re-transcribing the words with the correct spelling.

There is another problem which is not a negligible one. Since many words in English and in French are pronounced in almost the same, if not identical manner (e.g., in English, peace, piece, peas, pees; in French: parle and parlent) they are entered using the same strokes, thus creating the so-called “conflicts” which the software resolves by proposing the two, or three, or four differently spelled words from which, during the stage of correction, the operator will select the correct one.
We may now turn to the software for orthographic languages previously mentioned. The problem of the dictionary is eliminated, because the words are pronounced exactly as they are written. To explain more clearly: the vowels, as well as the consonants, almost always have the same sound. There are no mute vowels or consonants and thus the way a word sounds and the way it is written are almost entirely identical.

Furthermore, I was able to make use of experience derived from the first attempts, in the 1970s, of Stenograph, which had created a computerized machine that recorded on tape and a software based on the same principles as present-day software, so as to attain automatic transcription of the stenoscript with the creation of a list of strokes (those available with the stenotype machine) which corresponded to a dictionary of words.
This involved (and still today involves) the realization of an enormous dictionary of all of the strokes corresponding to all of the existing words, which would thus allow all ordinary words to be deciphered. I say “ordinary” because in almost all of these dictionaries for stenotyping, most proper names of any kind (family names, names of cities, rivers, mountains, states, etc. ) are missing, and they are infinite in number.

The table of meanings for the Spanish language, for example, is composed of 32 letters and special signs; there are 42 signs of punctuation, special signs, complementary signs (e.g., the signs for greater than and less than) and codes; 10 units of measurement; 10 numbers; 29 codes serve to indicate the prefixes; 24 the terminations; 29 the composite endings, and all of them include the various declinations (examples: ísimo, ísima, ísimos, ísimas; or gramación, gramaciones); lastly, there are approximately 1400 for coded abbreviations.

     In all, then, about 1550 codes. Not many when we consider that, for example, the word “mucho” also serves for “mucha, muchos and muchas” and they are considered 4, but the word is entered by striking the same key and merely changing the ending. Basically, it is a question of 800/1000 codes at the most. This number is the same for the tables of meanings in the Italian and the Portuguese languages as well.
     For non-orthographic languages there are instead 40/50 thousand codes. A big difference indeed!
     Obviously, the program which is used for an orthographic language is more convenient in all aspects.
     It is not subject to the quantity of words inserted in a dictionary and to their memorizing. It will not incur “conflicts”, and all of these problems are detrimental to achieving “real time”. And all words, of any kind, including proper names, can be written.
     In this regard it is enough to consider that three French competitors in the International Championship of Lausanne (1998) in the professional stenotyping class, in which a transcription time of 4 hours was allowed, were classified second, third and fourth at 470 syllables.

     In the fast transcription contest instead, with a maximum speed of only 310 syllables, one of them was classified tenth, consigning the text after more than 15 minutes, another 12th with consignment after over 20 minutes and the third was not classified because she failed to finish within the consignment time, which was 24 minutes.
     In the latest international championship of Hanover, again in the Fast competition, the Italians who were using the orthographic method were classified in the first five places (another two Italians came in seventh and eighth), consigning the text in a few seconds, while the American Karla Boyer, several times world’s champion and champion in the American competitions, was classified only ninth, consigning the text in 3 and a half minutes. For Italian stenotyping, to have 7 winners out of the first 8 is the demonstration that this achievement does not depend only on the professional skills of the competitors, but also clearly demonstrates the rationality of the method and of the software derived from it.
     Note that, at Hanover in the world’s championship contest (where up to 4 hours are allowed for consigning the text), the American was second, behind the super-champion Javier Nunez Hidalgo, classified at 409 syllables and thus at a speed much greater than the 210 syllables a minute of the Fast competition. Like the French, who in the Fast competition came in eleventh, twelfth and fourteenth.
     These are official results that demonstrate the absolute superiority of the orthographic method. Of course the merit is mainly that of the language, but perhaps it also depends a little on the intuition that, for the orthographic languages, there existed a more rational approach to managing transcription.
     It is clear that all this automatically results in greater reliability of the orthographic method in realizing the subtitles transmitted in direct on television.
     It is in this way that subtitles must be produced on TV in order to fully exploit the potentialities of the stenotype machine which produces, to be precise, simple files of text, i.e., pure ASCI code.

Any excessive delay in transmitting subtitles results in the well-known problem of the subtitles of previously reported news continuing to appear on the screen. And this is what must be avoided as far as possible. Already a brief amount of time is needed by the stenotypist to strike the keys, and this is only natural. Technology must not waste more time, and today this is certainly possible.
Lastly, as regards the technique of vowel recognition, for which there exists a very substantial international financial commitment for its development, even considering that it is a valid system for dictating texts, there are problems of another kind, hard to solve, if we wish to use it for “real time”. One of these problems consists of the fact that the software recognizes (and still imperfectly) only one voice. Moreover, it too requires a vocabulary of ordinary words and then it “stumbles” and writes other words that are entirely different, if the word is not pronounced perfectly or if the discourse includes proper names of any kind or other words not included in the dictionary.

Stenotyping (better the one derived from orthographic languages) is thus the only form, for the moment, the most reliable and the fastest in existence for realizing subtitles in direct transmission.

When at last, after extensive reflection and analysis, I decided to begin to develop my method, I threw out all the old methods and designed a brand-new one.
What differentiates the various methods and the various kinds of software? I believe it is appropriate to say that, after the arrival of the software derived from my innovative method, which differs substantially from the traditional one used for English, they can be divided into two categories, which refer to two different modes of interpreting a language from the phonetic-orthographic point of view.