A palatographic study of Shughni consonants

1.1 Phoneme inventory

The consonant inventory in Table 1 reflects the proposed distinctions between alveolars and dentals which will be argued for below.

1.2 Previous descriptions

The articulatory description of Shughni has not received much attention in the phonetic literature. Although the Soviet linguist V.S. Sokolova conducted instrumental research on many Iranian languages, including Shughni, in the 20th century, she did not always make the data she used available. While Sokolova’s works remain the most comprehensive phonetic sources on Pamir languages, it is clear that some topics were not covered in depth. For example, Sokolova (1953) treats consonants of the Shughni-Rushani subgroup as generally the same across all varieties, and the only exception concerns the palatalization of velar plosives. After commenting on /k ɡ x ɣ w/, she states that “other phonemes do not have any significant peculiarities^[1]”.

1.3 Peculiarity of velar fricatives

Shughni velar fricatives have been known to have an unusual phonetic quality ever since the first description of Shughni by Shaw (1877). Shughni /x/ – as in the word /xɐb/ ‘night’ (cf. Persian شب /ʃab/ ‘evening, night’) – was characterized by Shaw as “the German ch of ich sibilated so as almost to resemble an English sh” (p. 98). Later, Morgenstierne (1928: 36) noted that Shughni /x/ is “certainly easier to confound… with š than with x”. In later descriptions of Shughni phonetics, much attention was paid to /x ɣ/. For example, Sokolova (1953: 137) dedicated a whole paragraph to their articulation:

As it was noted by previous scholars, the phonemes x̌ and ɣ̌ acoustically remind one not only of x and ɣ (Russian dorsal fricatives) but of š and ž [/ʃ/ and /ʒ/. – Y. M.] as well, having a “š”- or “ž”-like quality. Observations have shown that the articulation of the phonemes x̌ and ɣ̌ is the same as that of the second focus [i.e., of the velar narrowing, typical of velarized sounds. – Y. M.] of Russian ш and ж [/ʃˠ/ and /ʒˠ/. – Y. M.]. If one refrains from using the tip of the tongue when producing Russian š and ž, the result will be typical Shughni x̌ and ɣ̌. The difference between the Shughni and Russian dorsal fricatives lies in the shape of the narrowing: in the production of Russian x and ɣ, the tongue is flat [Russian плоскощелевые, lit. ‘with a flat narrowing.’ – Y. M.] while during the production of Shughni x̌ and ɣ̌ (as well as in the second focus of Russian š and ž), it is grooved, which makes these sounds slightly hissing.

On the contrary, Karamshoev (1963: 69) stated that Shughni /ɣ/ is produced with the tip of the tongue raised upwards, creating a secondary place of articulation (“focus”) and causing its peculiar “noisiness”.

It is expected that such claims should be backed up by articulatory evidence. However, the only palatograms in Sokolova’s book (Karamshoev’s grammar of Bajuwi lacks any) are of the consonants of Wakhi, another Iranian language of Pamir.

1.4 Dental–alveolar contrast

No series of Shughni plosives, nasals, or laterals is distinguished solely by the dental–alveolar contrast. However, among fricatives, interdental /θ ð/ contrast with /s z/. The exact place of articulation of the latter has been a matter of debate. According to Edelman and Dodykhudoeva (2009), the Shughni coronals /t d ʦ ʣ θ ð s z n r l/ are all dental; nevertheless, Olson (2017) considers /ʦ ʣ s z n r l/ alveolar and the only dentals are said to be /t d θ ð/. Finally, a recent grammar by Parker (2023: 53) only treats /t d ʦ ʣ s z n r l/ as alveolar.

Interestingly, this discrepancy seems to reflect the backgrounds of the scholars. In the Soviet linguistic tradition, which was significantly influenced by the studies of the Russian language, coronals were “often mistakenly called dental, disregarding the target of the tongue, the teeth or the palate” (Vinogradov et al. 1960: 53). At the same time, the scholars whose mother tongue was English (e.g., Parker 2023) might have been more inclined to consider the Shughni coronals alveolar, cf. common descriptions of English coronals (e.g., Ladefoged and Johnson 2014; Roach 2009).

2.1 Palatography as a method

The main goal of palatography is to help identify the contact areas resulting from the tongue touching the palate or the teeth, i.e., the place of articulation. It is achieved by covering the tongue with non-toxic paint and photographing the roof of the mouth after the studied sound has been produced.

2.2 Procedure performed in this study

To obtain images of the tongue–palate contact areas for the Shughni phonemes, a mixture of sunflower oil and charcoal powder was used as paint; it was applied to the tongue using cotton swabs. Each stimulus was first explained in Russian. After successful recognition, the target Shughni word was produced and then the paint was washed from the mouth with tap water. If the word was mispronounced or the speaker accidentally produced extra words after the target one, the procedure was repeated for the same stimulus (Table 2).

Table 2:

Shughni vowel chart (Makarov 2026).

The list of the stimuli is given in Table 3. They are monosyllabic words with the target phoneme being the only lingual consonant; vowels are either /ɐ/ or /a/, which helps reduce the effects of coarticulation. Voiced/voiceless counterparts are assumed to have the same place.

2.3 Participants and collected data

Participants were seven speakers of Shughni (5 males, 2 females; mean age = 19.4, SD = 2.6), see Table 4 for more information. All of them lived in Khorugh except for A29, who came from Porshinev (roughly 15 km from Khorugh); every participant also spoke Russian. Souvenirs were offered in recompense for participation in the study.

2.4 Results

A total of 299 palatograms was collected, which included several duplicate photographs of the palate for each token (so as not to end up having only a bad picture). The clearest and most informative shots (70 in total) were manually selected for subsequent analysis. The palatograms demonstrating the main places of coarticulation observed in this study are given below.

2.4.1 Phoneme /t/

All of the participants except for A28 produced /t/ as a dental plosive with the frontmost contact right behind the incisor line, see Figure 1.

Figure 1:

Palatograms of Shughni /t/ as in /tɐp/ ‘shove’. (a) A28 (alveolar), (b) R28 (dental).

2.4.2 Phoneme /ð/

All participants produced /ð/ as an interdental fricative with the frontmost contact with the edges of the incisors, see Figure 2.

Figure 2:

Palatogram of Shughni /ð/ as in /mɐð/ ‘these’ produced by M28.

2.4.3 Phoneme /s/

The production of Shughni /s/ was subject to more variability compared to /t/ and /ð/. Five speakers produced it as an alveolar sound, and for two (N27 and H28) it was a dental sound, which can be seen from the contact with the incisors, see Figure 3.

Figure 3:

Palatograms of Shughni /s/ as in /bɐs/ ‘enough!’. (a) R28 (alveolar), (b) N27 (dental).

2.4.4 Phoneme /ʣ/

Shughni /ʣ/ looks like a typical dental sound with the frontmost contact with the frontal incisors, see Figure 4.

Figure 4:

Palatogram of Shughni /ʣ/ as in /ʣa/ (nonsense word) produced by R28.

2.4.5 Phoneme /n/

For all participants except for A28 /n/ was dental, see Figure 5.

Figure 5:

Palatograms of Shughni /n/ as in /na/ ‘not (negative prefix)’. (a) A29 (dental), (b) A28 (alveolar).

2.4.6 Phoneme /l/

Except for M28, Shughni /l/ was produced as a dental sound, see Figure 6. No dominant pattern of what part of the tongue is lowered was found, see Table 5.

Figure 6:

Palatograms of Shughni /l/ as in /lap/ ‘very; plenty’. For figures c & d, the side of the lowered tongue is given in parentheses instead of the place (dental). (a) N27 (dental), (b) M28 (alveolar), (c) A29 (right), (d) S25 (left).

Table 5:

Tongue–palate contact areas resulting from the production of /lap/ ‘very; plenty’ (by participant) including data on the side of the tongue lowered.

Participant	Frontmost contact	Side lowered
S25	incisor line	left
N27	incisor line	n/a
R28	incisor line	left
M28	alveolar ridge	n/a
H28	incisor line	n/a
A28	incisor line	right
A29	incisor line	right

2.4.7 Phoneme /ʃ/

The frontmost contact of Shughni /ʃ/ was always further back compared to /s/, see Figure 7 and Table 6 for details.

Figure 7:

Palatograms of Shughni /ʃ/ as in /mɐʃ/ ‘we’. (a) N27 (alveolar; painted edges of incisors are not related to /ʃ/ and were caused by an accident after the production of the target word), (b) M28 (behind the canine line, the postalveolar), (c) H28 (behind the 1st molar line, postalveolar).

Table 6:

Frontmost contact areas during the production of /mɐʃ/ ‘we’ and /bɐs/ ‘enough! (by participant).

Participant	Frontmost contact for /s/	Frontmost contact for /ʃ/
S25	alveolar ridge	canine line
N27	incisor line	alveolar ridge
R28	alveolar ridge	canine line
M28	alveolar ridge	canine line
H28	incisor line	1st molar line
A28	canine line	1st molar line
A29	alveolar ridge	canine line

2.4.8 Phoneme /ʤ/

Shughni /ʤ/ has the frontmost contact either behind the incisor line (S25 and H28) or the alveolar ridge (N27, R28, A28, and M28), see Figure 8.

Figure 8:

Palatograms of Shughni /ʤ/ as in /ʤɐm/ ‘plus’. (a) N27 (alveolar), (b) S25 (dental).

2.4.9 Phoneme /r/

Shughni /r/ was produced as a postalveolar sound with the frontmost contact behind the canine line, see Figure 9.

Figure 9:

Palatogram of Shughni /r/ as in /pɐr/ ‘feather; wing’ produced by H28.

2.4.10 Velar fricatives

Shughni /ɣ/ was produced as a typical velar sound with the frontmost contact behind the 1st molar line, see Figure 10.

Figure 10:

Palatograms of Shughni /ɣ/ as in /maɣ/ ‘sheep’. (a) S25 (behind the 3rd molar line), (b) R28 (behind the 1st molar line).

3.1 Palatal zones and place categories

In this study, palatograms are analyzed based on the modified classification from Firth (1948: 861), reproduced in Figure 11. Dental sounds, unlike alveolar sounds, are expected to contact the incisors; postalveolars are expected to be produced with tongue–palate contact behind the canine line. Identifying a sound as dental or alveolar depends on the frontmost contact observed in the palatogram.

Figure 11:

Classification of palatal zones from Firth (1948: 861).

Nevertheless, the organization of the oral cavity is individual, and sometimes amendments to this classification are required. For example, if the incisor and lateral incisor lines are roughly in the same place because of dental crowding, it does not mean that the speaker does not differentiate between dentals and alveolars. In the case of A28, the two lines coincide but /s/ and /ʃ/ are nevertheless articulated differently: the latter is further back in the mouth, crossing the canine line (cf. Table 6). This situation showcases the relative orientation of articulatory gestures and the importance of considering speakers’ mouth anatomies individually. To account for such cases, the alveolar ridge is used as a reference point instead of the lateral incisor line.^[2]

3.2 Revised classification of Shughni consonants

The results of this study suggest that neither of the existing descriptions of Shughni consonants is accurate. While Edelman and Dodykhudoeva (2009) argued that /s z r/ are dental, according to this study /s z/ turn out to be alveolar and /r/ postalveolar. Olson’s (2017) treatment of /ʦ ʣ n r l/ as alveolar has found little-to-no support; the same can be said about Parker’s (2023) account with the addition of /t d/, incorrectly classified as alveolar.

The proposed consonant chart can be found in Section 1.1. To sum up, the phonemes /t d θ ð ʦ ʣ n l/ are dental, /s z ʧ ʤ/ are the only alveolars and /r/ is postalveolar.

3.3 Production of fricatives

As for the articulation of the coronal fricatives, Figure 3b shows no unpainted gap in the middle (the same for H28). This seems quite unusual compared to the palatograms of the dental fricative in Figure 12 or, for instance, in Bolla (1981: 56), which all have such gaps.

Figure 12:

Palatogram of Russian /z̪ˠ/ as in зам /z̪ˠamˠ/ ‘deputy’ produced by a 22-year-old male speaker of Standard Russian.

It might be the case that the dental articulation of /s/ in Shughni requires a tighter narrowing. Alternatively, the two examples (N27 and H28) can manifest hyperarticulation effects.

As for the velar fricatives, there is no sign of the tip of the tongue touching the front part of the palate, so Karamshoev’s (1963) suggestion regarding the front tongue–palate contact (see Section 1.3 above) is not confirmed. Similarly, Sokolova’s (1953) suggestion that the hissing is caused by the shape of the velar constriction finds no support since the shape of the narrowing during the production of /ɣ/ is reminiscent of Russian /x/ (Figure 13), and Russian /x/ lacks the notorious hissing. These facts suggest that an ultrasound or MRI research should be conducted as they provide more information about the tongue shape.

Figure 13:

Palatogram of Russian /x/ from Bolla (1981: 69).

As a side note, Shughni /x/ reminds one of /x/ and /ʃ/ produced simultaneously, just like Swedish /ɧ/ (sj), though Shughni /x/ is not as highly labialized, cf. Figure 14 below. However, the literature on Swedish /ɧ/ can certainly provide valuable insight into the Shughni velar fricatives.

Figure 14:

Lip position during the production of Shughni /x/ as in /xɐb/ ‘night’ pronounced by S25. (a) Side face view, (b) full face view.

3.4 Production of affricates

The production of affricates poses a problem for classificatory charts like Table 1 since they are viewed as complex sounds, and its stop and fricative components can have different places of articulation, cf. Ladefoged and Johnson (2014: 45). Utilizing direct palatography, it is impossible to access the place of the frontmost contact for both components. Following the convention used in this study, Shughni /ʣ/ is classified as dental because of the initial dental occlusion /d/, which is the first component of /ʤ/. However, Shughni /ʤ/ can only be attributed to dentals in two cases (see Section 2.4.8), the rest suggesting that it is alveolar. The proportion of the observed alveolar articulations of /ʤ/ can be explained by the retracting influence of the postalveolar fricative component /ʃ/. It must be noted, though, that the alternative analysis of /ʣ/ as alveolar and /ʤ/ as postalveolar is also possible and in fact can be bolstered by some phonological factors. This is not, however, a goal of the present study.

3.5 Intra- and inter-speaker variation

The majority of the data in this paper were collected in 2023 but in 2022, a preliminary study was conducted. Participant S25 was involved in both studies. Interestingly, in 2023, S25 produced /t/ as a dental plosive, whereas it was clearly alveolar in 2022; the same applies to /s/, see Figures 15 and 16.

Figure 15:

Palatograms of Shughni /d/ produced by S25. (a) 2023 (dental; as in /tɐp/ ‘shove (sth.into sth.)’), (b) 2022 (alveolar; as in /bɐd/ ‘then’).

Figure 16:

Palatograms of Shughni /s/ as in /bɐs/ ‘enough!’ produced by S25. (a) 2023 (dental), (b) 2022 (alveolar).

Articulation may change over time as a result of many factors, e.g., physiological developments, second language influence, etc.; and the observed shift highlights the importance of panel studies in phonetics. While the exact reason why the consonants had different places of articulation in 2022 and 2023 remains unclear, there is room for conjecture. One reason could be a physiological change affecting S25’s mouth, which can still develop until early adulthood (Yu and Klein 2020). Another possibility is code-switching influenced by certain social factors (e.g., acquaintance with the researcher by the second year and hence lesser formality).

Finally, the observed “change” may actually result from variation within the Shughni-speaking community. It is obvious that variation in place of articulation is not typical of some phonemes (e.g., /r/ or /ʣ/) and typical of other (/s/ or /ʤ). The discussed phonemes belong to the latter group. Furthermore, assuming that a phoneme has a constant place even for one speaker and in the same phonetic context is a simplification, especially for the easily confused coronals. It is therefore not surprising that the same speaker switches from dental gestures to alveolar and vice versa. However, it is not clear whether the change in the same direction affecting both /t d/ and /s/, constituting the natural class of coronal obstruents, is a pure coincidence or a reflection of the articulatory shift affecting the natural class in general.