2 First use of the procedure

It is not surprising that experimental methods in phonetics were imported from other scientific fields, such as physiology, medicine, physics, acoustics, brain research, etc. (Moses Reference Moses1940, Gósy Reference Gósy, Wai Sum and Eric2011, Ashby Reference Ashby2016). The question posed in the 1880s was how tongue and palate, tongue and teeth participate in the very complex gestures that characterize a speech sound. The answer given to this question at that time was palatography, also known as stomatoscopy. This technique was invented by James Oakley Coles, an English dentist. Trying to help patients with defects of the palate, he experienced a need to observe the exact movements of the articulators and recognized the difficulty of teaching correct articulation to his patients without a suitable method. The method he invented was so-called direct palatography, published in the British Medical Journal in 1872. Another expert on dentistry, orthodontics and cleft palate therapy, the American Norman W. Kingsley used the indirect form of palatography, which means that he used an artificial palate in his experiments (Kingsley Reference Kingsley1879).

Palatography got soon around, famous linguists, phoneticians began to use the method, among others Grützner, Hagelin, Techmer, Rousselot, Sievers (see Abercrombie Reference Abercrombie1957, Tillmann Reference Tillmann, Koerner and Asher1995, Ashby Reference Ashby2016). According to Ashby (Reference Ashby2016), the palatographic method entered the linguistic-phonetic mainstream following Sievers’s (Reference Sievers1881) work. Studying the references and the subjective notes of the early papers of various European scholars does not provide sufficient information on the impacts and mutual influences that researchers exerted on one another. For example, Grützner (Reference Grützner and Ludimar1879) claimed that his method was independent of that of Coles (Reference Coles1872); however, he makes a reference to Coles’ paper, while Techmer (Reference Techmer1880) makes a reference to a third party (Gavarret Reference Gavarret1877), claiming that he had learnt Coles’ method from that source (see the details in Ashby Reference Ashby2016: 58–59). Rousselot (Reference Rousselot1897–1908) does not mention Coles’s name in his book, either.

As mentioned above, József Balassa was the first Hungarian phonetician who used palatography. He called the method ‘stomatoscopy’. Although Balassa was aware of the two types of the new methodology, he followed Kingsley’s procedure, using an artificial palate. His opinion was that Techmer’s ‘natural mode’ was more accurate but also more difficult than using the artificial palate. In Balassa’s (1887: 132) own words, in this method the tongue itself makes a drawing of its own movements. The results of his first investigations were published in a Hungarian journal entitled Nyelvtudományi Közlemények [Linguistic Papers] in 1887, and also in the journal Internationale Zeitschrift für allgemeine Sprachwissenschaft in 1889.

3 Methodology of palatography

The new technique of palatography spread relatively quickly among those interested in speech. There are two types of the method, direct (static or natural), proposed by Coles, and indirect palatography, proposed by Kingsley (e.g. Abercrombie Reference Abercrombie1957, Anderson Reference Anderson2008, Ashby Reference Ashby2016). Direct palatography was designed to make visible the contacts between the speaker’s tongue and the palate. In this method, various marking materials are painted on either the palate or on the tongue. Following articulation, the points of contact can be observed and recorded. In the case of indirect palatography, which soon followed the previous type, the researcher uses an artificial, custom-made palate, which is removed from the speaker’s mouth after the articulation of a particular sound. The artificial palate will preserve the contact points between the speech organs under study. There was only a seven-year difference (in terms of publication date) between the two types of palatography (Coles Reference Coles1872 vs. Kingsley Reference Kingsley1879). In Hungary, however, the chronological order of using the palatographic technique took the opposite path: first it was the artificial palate that began to be used and then came the application of the natural mode. Irrespective of the type of palatography, there have been substantial changes over time in the materials used, the procedure, and the participants in the experiments.

3.3 Procedure

In the case of direct palatography the palate was painted and then the speaker was asked to articulate a speech sound to be analysed either in isolation, between two [aː] vowels or in short syllables or words. After articulation, some of the mixture was wiped off from the palate. Grützner and Techmer made also linguograms, that is, they painted the tongue before the production of speech sounds (see Grützner Reference Grützner and Ludimar1879, Techmer Reference Techmer1880). Carruther (Reference Carruthers1900) applied his mixture either to the palate or to the tongue. In Hungary, it was only after 1970 that researchers began to make linguograms (Figure 1), and even Hegedűs (Reference Hegedűs1941–43), an outstanding phonetician, made only palatograms.

Figure 1 Palatogram (left) and linguogram (right) of the Hungarian consonant [s] produced between two [aː] vowels by a young female speaker (Gósy Reference Gósy1981).

The artificial palate was placed into the oral cavity, and when the speaker articulated the desired speech sound, it was immediately removed. At the place where the speaker’s moist tongue touched the material of the artificial palate (e.g. vulcanized rubber) the layer of chalk (or any other material) disappeared (wipe-off areas), and the original (black) surface became visible, showing the contacts of the articulators, characteristic of the speech sound in question between the articulators.

With direct palatography it proved difficult to draw (record) the visible contacts of the various speech organs. The authors – like Coles – used a small mirror to observe and draw the patterns on a paper sheet, on stone, or on plaster models. Coles produced both palatograms and linguograms of the ‘letters’ (as speech sounds were called by Coles). He admitted that he himself had no idea about ‘the way’ certain sounds are produced (see Coles Reference Coles1872: 181). Carruther asked his father to draw the contacts looking at his palate; see Ashby (Reference Ashby2016: 62). Grützner (Reference Grützner and Ludimar1879) recorded the contact patterns transferred to the palate, using mirrors to look at the patterns. With indirect palatography it proved easy to record the visible contacts of the speech organs. After the removal of the artificial palate, the researcher had time to observe and analyze the patterns. Thus, the indirect method provided a simple way to make visible the contacts of the speech organs following articulation and ensured enough time for the researcher to observe the results. Balassa (Reference Balassa1887) reported that he had made drawings based on the patterns the artificial palate showed in the case of each speech sound. Hegedűs (Reference Hegedűs1941–43) refined the indirect method of palatography in order to obtain more accurate drawings and more objective values for the contacts on the palatogram. He drilled small holes into the surface of the palate (10 mm from each other), which provided reference points for further analysis.

Hagelin (Reference Hagelin1889) was the first who used photography in indirect palatography. The Czech phonetician, Josef Chlumský (Reference Chlumský1914), a disciple and follower of Abbé Rousselot and the founder of the Czech experimental phonetic laboratory (see Šturm Reference Šturm2019), refers to Hagelin’s photography, whom he followed in his own work. He emphasized, however, that he did not follow the common practice of retouch on his photos and called attention to the importance of drying the powder before applying it to the artificial palate. Although Rousselot (Reference Rousselot1897–1908) made a reference to Hagelin’s use of photography in palatography, he claimed that taking pictures of the patterns had no benefit since the contours were not sharp enough to allow exact measurement. Hegedűs (Reference Hegedűs1941–43) used the artificial palate in combination with photography. For various reasons (including practical difficulties), however, photography was not often used until the 1950s, while subsequently it became common practice in direct palatography.

4 Consonant and vowel findings by Hungarian researchers

Balassa’s (Reference Balassa1887) research had some understandable shortcomings that were due to the fact that he was a pioneer in Hungarian experimental phonetics, and these were the first steps in using palatography in this country. Figure 2 below shows 17 original palatogram pictures from one of Balassa’s papers (Reference Balassa1887: 135). The letters on the left side under the palatograms represent Techmer’s phonetic symbols, while those on the right side represent the equivalent Hungarian letters (these letters practically correspond to Hungarian phonemes).

Figure 2 Balassa’s palatograms (1887: 135). (Original title translated into English: Stomatoscopic pictures of Hungarian sounds.)

Beside using a modern technique to explore the exact articulation gestures performed in producing speech sounds, Balassa has to be credited for the fact that in studying speech sounds he freed himself from the effects of spelling. For example, it was not only [k] and [ɡ] that he listed as velar consonants, but also the velar nasal [ŋ], which is not a phoneme in Hungarian but a variant of the alveolar nasal consonant when it precedes and is coarticulated with velar stops. Balassa did not tackle the issue of context-dependency systematically. However, he provided some palatograms to illustrate velar consonants in back vowel contexts (Figure 2: picture 6) and in front vowel contexts (Figure 2: picture 7).

Although short and long vowels are both phonetically and phonologically different in Hungarian, Balassa found that the palatograms for these vowels were the same. Similarly, the consonants differing only in voicing patterns (which are phonologically relevant) were represented by the same palatogram. Figure 2 contains (short) front vowels and only one back. The explanation is that the open vowels ([ɔ aː ɛ]) do not show visible contacts on the palate, and they can easily be examined by looking into the oral cavity (1887: 131, 136).

Balassa (Reference Balassa1886) defined various regions on the palate that correspond to the places of articulation of Hungarian consonants. The places of the tongue, contacting the palate, were marked by Roman numerals while Arabic numerals referred to regions corresponding to the places of articulation of the consonants on the palate. He examined the places of articulation of the consonants by looking at the marks of contacts on the palatograms and identified them according to predefined regions. After analysing the palatograms, he realized that in his book on phonetics he misidentified the fricatives [ʃ ʒ] as palatals, while in fact they are post-alveolars.

Although this paper is designed to focus on the work of Hungarian scholars who used palatography from the end of the 19th century, it is impossible not to mention in this connection Abbé Rousselot, the acknowledged founder of experimental phonetics. In 1897 he published four palatograms based on Hungarian speech sound articulation (see pictures on pages 609, 611, 652 and 653 of Rousselot Reference Rousselot1897–1908). Two of them were known at the beginning of the 20th century (Gombocz Reference Gombocz1908: 193) while the other two were found more than 70 years later (Gósy & Olaszy Reference Gósy and Gábor1985). One of Rousselot’s pictures shows the Hungarian voiceless velar consonant ([k]) in various contexts (Figure 3), indicating the importance of context-dependency at that early date. Abbé Rousselot intended to demonstrate that the modifications in articulation gestures were due to the effects of the velar stop consonant on the vowels in question. (The speaker is unknown.)

Figure 3 Palatograms of consonant [l] (No. 3, on the left) and patterns of various front vowels coarticulated with [k] (No. 4, on the right) produced by a Hungarian speaker, shown in Rousselot’s book (1897–1908: 611, 652).

Gombocz (Reference Gombocz1908) analysed all short Hungarian consonants, all short vowels (without their long counterparts) and two long vowels ([aː] and [eː]) that have no short counterparts in standard Hungarian. He made a palatogram of the velar nasal [ŋ], like Balassa, he pronounced it in the word engem [ɛŋɡɛm] ‘me’ (Gombocz Reference Gombocz1908: 198). In his paper he presented the picture of only one long consonant, the trill, without any explanation. He used existing monosyllabic and disyllabic words, consisting of one or two of the following consonants: [p b m v f] and the vowels [ɛ ɔ aː]). He analysed the vowels and the consonants in two different phonetic positions, in word initial and in word medial positions (e.g. nyak [ɲɔk] ‘neck’ vs. anya [ɔɲɔ] ‘mother’) as well as in various phonetic environments. Figure 4 shows some of his palatograms.

Figure 4 Gombocz’s palatograms (IPA symbol is shown under the picture).

Gombocz noted the differences between the voiced and voiceless members of consonant pairs. Although the voiced and voiceless pairs shared the same palatogram picture, only one member of the pair is shown in the pictures, while the other one is indicated either by a dotted or a white line (see the palatogram for [ɡ] and the dotted line indicating [k] in Figure 4). Gombocz made some comparisons concerning the contact areas between palatal and velar vowels. He realized that phonetic position changes the size of the contact areas in the palatograms, but he did not make measurements.

As mentioned earlier, Csűry was not a phonetician but a dialectologist who decided to use palatography in analysing dialectal speech sounds. He came from north-eastern Hungary, the region called Szamoshát, and thus he studied his own, dialectal articulation. The goal of his investigations was to shed light on some controversial or unknown speech sounds in that dialect. Csűry copied Gombocz’s method so that their findings could be compared. Based on the palatograms, he determined the exact articulation for a controversial front vowel of this dialect, which is slightly more close than [eː] but more open than [iː], a sound that is closer to the former than to the latter. Csűry demonstrated that the environment influenced and modified the articulation of the consonant [n] in the vicinity of some fricatives. In addition, he ascertained that the [n] and [t] consonants were postalveolars in this dialect.

Csűry found some specific differences in articulation between his dialect and Gombocz’s, who came from West Hungary. Figure 5 shows the palatograms of the consonant [t] pronounced by two speakers speaking two (geographically) distant dialects. The place of articulation for [t] is nearer to postalveolar in north-east Hungary while it is dentialveolar in West Hungary (Gombocz Reference Gombocz1908, Csűry Reference Csűry1936).

Figure 5 Palatograms for the consonant [t] made by Csűry (left) and Gombocz (right), representing different dialects. The dotted line indicates here the contacts of [t] in word-initial position.

Hegedűs was the first to make exact measurements on the palatograms using a planimeter. (A planimeter is an instrument to determine the area of an arbitrary two-dimensional shape, like the artificial palate in this case.) In his early work, Hegedűs (Reference Hegedűs1941–43: 69) provided a detailed description on how to use the planimeter. Figure 6 shows some of Hegedűs’s palatograms made in the Phonetics Laboratory of Panconcelli Galzia in Hamburg in 1939. The small black points in the palatograms show the small holes Hegedűs (Reference Hegedűs1941–43) drilled into the surface of the palate 10 mm from each other, which provided reference points for measurements. His paper, reporting the results of his experiments was published some years later.

Figure 6 Palatograms made by Lajos Hegedűs (Reference Hegedűs1941–43). The consonants are shown in IPA symbols. The dotted lines indicate the contacts of (phonologically) long [r] and (phonologically) long [ɲ], respectively.

Hegedűs summarizes his measurements (expressed in mm²) in a table containing the contact areas visible on the palate, representing the author’s pronunciation (he noted that his pronunciation corresponded to standard Hungarian, i.e. he did not speak a dialect). The whole area of his artificial palate was 3.350 mm². Here I will cite some of the values he obtained in his measurements (Table 1).

Table 1 The contact areas visible on the palate in the case of some Hungarian speech sounds, measured by Hegedűs (Reference Hegedűs1941–43: 69).

Based on the analysis of the measured values Hegedűs drew some important conclusions. He discovered that with stops and fricatives the area of contact between the palate and the tongue was larger in word initial than in word medial positions. The larger contacts could be observed if the stop or fricative was followed by a palatal vowel, particularly [i]. Contact patterns were found to be larger in the case of (phonologically) long vowels and consonants compared to (phonologically) short ones. Finally, based on the measured values, Hegedűs confirmed that the contact areas were larger with voiced stops and fricatives than with their voiceless counterparts.

The natural method of palatography received a boost when the so-called palatograph was introduced in the 1950s in Hungary. This is an instrument containing a special mirror of a suitable size and design which would fit into the oral cavity without hurting the mouth or the tongue, a camera, lightbulbs and a sort of holder showing the letter (and the context) corresponding to the articulated speech sound (Figure 7). It is possible that Lajos Hegedűs started using this palatograph when he worked for the Hungarian Linguistics Institute (in the 1950s).

Figure 7 Palatograph from the 1950s.

After another 30 years the use of palatography was employed again by some of the researchers at the Linguistics Institute. The instrument was used in several research projects, including one designed to study disordered articulation of some fricatives using both palatograms and linguograms. This research was carried out in the 1970s and 1980s at the Phonetics Laboratory of the Institute, which owned the instrument. A conspectus containing palatograms and linguograms for each Hungarian speech sound was compiled by Bolla (Reference Bolla1982), using the institute’s palatograph. However, he did not make any measurements.