Notably, during the first task, the values were all slightly lower compared to the other tasks, possibly due to the uncertainty associated with reading an unfamiliar text. Table IV and shows the average values of the CoG and StD, divided by speaker and linguistic context.

First, we analyse the “real” dental affricates in solnce, konce and koncov. R1 and R2, both from Saint Pietersburg, exhibit very similar CoG values, suggesting a central articulation of the fricative moment. In contrast, R3 from Volgograd has higher values than R1 and R2, indicating a more anterior pronunciation of the phoneme. Lastly, R4 from Moscow shows the highest CoG values, nearly suggesting an interdental articulation. Regarding the StD values, R1 shows a wide dispersion of energy, while R2, despite being from the same city, presents significantly lower values than R1. R3’s values display minimal variation and are very similar to those of R4.

Kazakh speakers, on the other hand, exhibit notably greater variability in their values, both across different speakers (with a variation of more than 1000 Hz for the CoG) and within the values of the same speaker.

K1, who uses the Russian language daily, maintains a high average CoG value, exceeding 7500 Hz, suggesting an interdental articulation. His CoG values also show considerable variation approaching 1000 Hz.

In contrast, K2 demonstrates even higher variation, exceeding 1000 Hz, and his average CoG is around 5700 Hz, indicating a more central articulation of the fricative moment. Interestingly, the CoG of the strengthened fricative in solnce, produced as /'ʦonʲʦi/, is higher than the other affricates articulated by K2, signifying a more anterior realization of the phoneme. The StD (5500 Hz) are lower than those of K1, suggesting less diffusion articulation energy. Remarkably, this value is nearly identical to that of the strengthened fricative, with a difference of only 43 Hz.

Moving to K3, which primarily used the Russian language in academic contexts, there is almost 1000 Hz of variation in the CoG values. The average CoG value is 5000 Hz, showing a central articulation. Interestingly, the dental affricates produced as voiced also have a central place of articulation. However, the CoG of the voiced affricate in solnce is lower than the voiceless phoneme in the same context, while the CoG values of the voiced affricates in konce and koncov are higher than the voiceless counterparts in the same contexts.

Analysing the articulation energy, the StD values for the voiceless affricates produced by K3 are lower than those for both K1 and K2, with an average value of 4000 Hz, while the voiced value is even lower (3700 Hz).

K4, who frequently uses the Russian language, shows a notable variability in his CoG values. The highest value is observed in konce, reaching almost 9000 Hz, while the lowest is in solnce (7000 Hz). These values suggest an interdental articulation of the phonemes. In contrast, the StD values for K4 are similar to those of K1 and K2 (5350 Hz).

Moving to the affricates formed by the voiceless dental occlusive /t/ followed by the voiceless fricative /s/ in sčitat’sja (be considered), udastsja (manage to), otkazat’sja (to give up). Table IV indicates that the CoG values of these affricates produced by Russian speakers are slightly higher than those of Kazakh speakers. It also shows that the StD values, on the other hand, are lower than those observed for the “real” affricates analysed previously.

R1 maintains an average CoG value higher than 6000 Hz, indicating a more anterior articulation of the phoneme compared to the “real” affricates. Although his StD values show variability of almost 300 Hz, the average value (5090.1 Hz) is nearly identical to that of the “real” affricates (5098 Hz).

R2 has a slightly higher average CoG value (6425.9 Hz) than R1, but his StD value is lower (3051.4) than R1’s. It is worth noticing that R1 and R2, both from Saint Pietersburg, exhibit more similar articulation patterns for the “real” affricates other than the ones analysed in this context.

R3, from Volgograd, has an average CoG value of 6944.5 Hz, which is 200 Hz higher than that of the “real” affricates. However, his StD value (3721.8) is 300 Hz lower.

R4, unlike the other speakers, does not produce the affricate in udastsja in any of the three reading tasks. His average CoG value is very high (9388.9 Hz), more than 500 Hz higher than the “real” affricates, while the StD value (3831.8 Hz) is 450 Hz lower than those observed in other contexts analysed.

These findings show that all Russian speakers produced the voiceless dental affricates formed by /t/+/s/ with a place of articulation more anterior and less articulation energy dispersion than the “real” affricates analysed above.

Moving to K1 from the north region of Kazakhstan, we observe a place of articulation that is more posterior than the previously analysed affricates, as indicated by the lower CoG value. Interestingly, the StD values for both contexts are similar to each other.

In the case of K2, there is a notable variability in CoG values, exceeding 1300 Hz, and the average value (7488.8 Hz) shows a more anterior articulation compared to the articulation of the real affricates (which was 5700 Hz). Additionally, the StD is 5201.4 Hz, indicating a narrower dispersion of articulation energy compared to the other context.

In line with the patterns observed in K2’s production, the articulation of K3 of the voiceless affricates follows a similar trend: the average CoG value is higher than that of the “real” phonemes, while the StD value is lower. Notably, the StD value closely aligns with that of Russian speakers. It is worth noting that K3 produced the affricate in udastsja as voiced /d͡z/, and both the CoG and StD values are lower than their voiceless counterparts. This is surprising, given that in languages distinguishing between /d͡z/ and /͡ts/, the standard deviation of the voiced phoneme is typically higher than that of the voiceless affricates (Meluzzi, 2020).

Analysing K4, it becomes clear that this speaker has a degree of articulatory uncertainty, as reflected in the variability within his values. Similar to previous observations, K4 has an almost interdental articulation, since the CoG of the affricates in sčitatsja and otkazat’sja are around 7500 Hz, while the one in udastsja is 9000 Hz. However, the StD values indicate that this speaker has a similar dispersion of articulation energy across all three affricates.

The final set of affricates analysed are those formed by a voiceless dental occlusive in word-final position followed by a voiceless fricative in a word-initial position, as in tut severnyj veter (“and so the North wind) and ot svoej (“from its”).

Among the two Russian speakers from Saint Pietersburg, R1 and R2, there is a noteworthy degree of similarity in their CoG values, while their StD values differ by 1000 Hz. Specifically, R1 presents an average CoG of 6667.3 Hz, surpassing the CoG value of the previously analysed consonant group by 500 Hz, and exceeding the real affricates value by 1000 Hz, signifying a more anterior articulation. Conversely, the average StD is nearly 400 Hz lower than in other contexts.

Similarly, R2 also has a high CoG value (6618.7 Hz), surpassing the CoG value of the previously analysed consonant group by 200 Hz and exceeding the “real” affricates value by nearly 700Hz. Unlike R1, R2 has also a higher StD value than that observed in other contexts.

Following a similar pattern as R1, R3 showcases an anterior place of articulation with the lowest StD value compared to the other two contexts. Additionally, his CoG values are also 1000 Hz higher than those of R1 and R2.

Contrarily, R4 production leans towards an almost interdental articulation. However, both the CoG and StD values are lower than those observed in the previous contexts. Interestingly, this speaker demonstrates a similar energy dispersion to that of R1 and R3.

Regarding the Kazakh speakers, K1 has produced this type of affricate with a more anterior articulation compared to the previously analysed affricate, as indicated by CoG values that are more than 2000 Hz higher. However, his StD is lower than those of other contexts, although it aligns with the values of R1, R3 and R4.

K2 follows a similar trend to K1, with an average CoG value higher than that of the other types of affricates and a lower average StD value, which is nearly identical to K1’s value differing by just 173,2 Hz.

In this context, K3 articulated only voiceless phonemes, and only in ot svoej, dividing the two words in tut severnyj not producing an affricate. The average CoG value of the phoneme analysed suggests a more anterior place of articulation compared to the other contexts, aligning with the other speakers. However, the StD value, is notably low, indicating little energy dispersion during the production process (2581.8 Hz).

As for K4, the values show a very anterior articulation of the affricates. Nevertheless, the average CoG value (7957.3 Hz) is lower than that of the consonant group contexts analysed earlier. However, K4’s StD value follows the trend observed in others, being lower than those of all the other contexts analysed before.

It is worth noticing that the participant from Moscow R4 articulates highly anterior affricates, as evidenced by the highest values among those analysed. This observation raises an interesting possibility to investigate whether this could be a case of idiosyncrasy, given that regional varieties within a city can exhibit distinctive phonetic characteristics (Kochetov, 2017).

In addition to investigating the place of articulation, our analysis explored the overall production of dental affricates and examined the two distinct phases of occlusion and friction within the phoneme.

Our particular focus was on the duration of the occlusive moment, as this aspect exhibits substantial variability in the production of dental affricates across different languages.

To obtain these values, we extracted data from the spectrogram and then normalized by word duration. Table V shows all the average values.

Concerning the “real” dental affricates, both speakers from Saint Pietersburg R1 and R2, have nearly identical temporal values, with a difference of 0.03 ms. However, when we focus on the occlusive moment, R1’s is shorter than R2’s. Interestingly, R2 produced s strengthened fricative, forming an affricate in the word solnce, which is longer than the other phonemes, yet its occlusive phase is similar to that of R1 concerning the overall word duration.

In the case of R3, the affricate produced in solnce is longer than what is observed in other Russian speakers. Nonetheless, its occlusive moment closely resembles that of R1 when considering the entire word. In contrast, the phoneme in konce is 12 ms longer than the one in solnce, but it remains notably brief in comparison to the other Russians. Conversely, the occlusive moment is longer than the moments produced by R1 and R2. Furthermore, the affricate in koncov closely resembles the one articulated by R1.

Interestingly, R4 produces the dental affricate in solnce with a shorter duration compared to everyone else. However, it is noteworthy that the occlusive moment in this case is longer than that of any other Russian speaker. Moreover, the phoneme analysed in konce is 10 ms longer than that produced by R3, although the occlusive parts articulated by these two participants are nearly identical. As for the affricate in koncov, it appears to be short, and the occlusive moment does not exhibit a significant difference compared to the other speakers.

In summary, among all the Russian participants, it is evident that koncov emerges as the shortest affricate in terms of duration, except for R2, who produced the most temporally variable phonemes and articulated solnce as the shortest.

One prominent observation in the durations of the Kazakh participants is that they tend to be longer overall, although they still align with the trend observed among the Russian speakers, where the affricate in the word koncov is the shortest, except for K3, who articulates the voiced affricates as shorter than the voiceless ones.

In the case of K1, the phoneme in solnce is longer than the Russians. However, the occlusive moment in this instance is nearly identical to that of R4. Moreover, K2 articulates this phoneme very similarly to R4, including the occlusive part. This is very interesting considering that this speaker does not frequently use the Russian language. K4, on the other hand, produced a long affricate, yet the occlusive moment is comparable to that articulated by the Russians. Notably, the voiceless dental affricate in solnce produced by K3 is 100 ms longer than the voiced one, but the occlusive part of the voiceless phoneme is shorter than that of the voiced one.

As previously discussed, koncov has the shortest affricate produced by Kazakhs (except for K3). K1 and K4 articulate a phoneme that closely resembles that of the Russians, although the occlusive part of K1 is shorter than the fricative phase. K2, on the other hand, produces a very brief affricate, but its occlusive moment is relatively long compared to the entire word. In the case of K3, this participant articulates a lengthy voiced dental affricate, with a durative occlusive moment.

The analysis of the duration of “real” affricates reveals that Russian speakers tend to have more similar results.

Interestingly, K1 has different values from the Russians, despite using Russian in daily communication. In contrast, K2, who rarely uses Russian, demonstrates similarity to a Russian speaker. K3 shows variability, likely influenced by the addition of sonority in his affricates, but his values are similar to those of the other speakers. Lastly, K4 presents the most variable values, although the occlusive moments he articulates closely resemble those of the Russian speakers.

Analysing the dental affricates formed by the consonant group occlusive /t/ + fricative /s/, the initial observation is that their duration is similar to that of the “real” phonemes.

The first affricate analysed is the one in sčitat’sja. R1 and R3 produced the occlusive part of the phoneme with a duration like the “real” ones. In contrast, R2 articulated a short occlusive part in relation to the fricative one, which is longer. R4 produced a shorter phoneme, with the occlusive moment almost identical to the fricative one. Kazakh speakers, on the other hand, show a duration for the affricate in this context that is generally more similar to each other and even to the Russian speakers. K1 and K4 articulated the phoneme with a very short occlusive part, whereas K2 and K3 produced this part almost as long as the Russians.

The second word analysed is udastsja, for which we have fewer occurrences. This is because R2, R4 and K1 did not unify the articulation of the two phonemes into the affricate in the fourth task, likely due to the complexity of the consonant group. This context is also the most variable among those analysed. R1 articulated the affricate 100 ms longer than R3, and the occlusive part of this phoneme is also longer compared to the one produced by R3. In the case of K3, who articulated a voiced affricate in this word, there is a very short occlusive moment compared to the duration of the entire phoneme. A similar pattern is observed in the articulation of the occlusive part of K4.

The last word analysed in this consonant group context is otkazat’sja. The phonemes articulated by the Russians in this word were very variable, but overall, their durations were shorter compared to the other words analysed. It is noteworthy that in the phonemes produced by R1 and R4, the fricative part is longer than the occlusive one, while R2 and R3 articulated the “standard affricate” where the occlusive part is longer than the fricative.

The first phoneme analysed is the voiceless dental occlusive /t’/. In total, there were 11 occurrences of this phoneme in 4 different linguistic contexts, amounting to 33 occurrences across the three reading tasks. The contexts are:

None of the participants articulated the expected 33 occurrences. Specifically, the Russian speakers did not produce the ones in the word-ending position: instead, they articulated a long vowel. In total, we analysed 208 palatalized occlusives, with 126 produced by the Russians (plus the occlu sive articulated as an affricate by R2); and 82 voiceless occlusive palatalised by the Kazakhs (plus 12 affricates).

It is worth noting that there were 3 unexpected occurrences: those in kutalsja, the masculine past form of “cover himself”; kotoryj, a relative pronoun; and zakutannogo “wrapped up”. All these occurrences are shown in Table VI divided by speakers

First, we analysed the place of articulation of the voiceless dental occlusive when produced with palatalisation. To do so, we extracted the average values of the release moment using the software PRAAT (see Table VII). The fricative phases of those phonemes articulated as dental affricates were also analysed, and these are shown in Table VII with the '*' symbol.

The first noticeable thing is that the Russian speaker R1 produced the lowest number of palatalised phonemes, only 27. This number of occurrences is similar to K1’s, which articulated 25 palatalised occlusives. This can be explained by the fact that R1 moved from Saint Pietersburg to Italy 10 years ago, and during the interview task, he confessed that she tried to eliminate all the traits typical of Russian when speaking Italian.

She felt that these traits could create a linguistic divergence between him and the interlocutor (cfr. Bratkowsky, 1980 for the palatalisation of German consonants and their perception in Russian speakers). This fact has led to possible linguistic accommodation, which is the process whereby a listener makes adjustments in response to the speaker’s behaviour.

We proceed now with the analysis of the expected occurrences.

R1 articulated the voiceless dental palatalised occlusive similarly in all the linguistic contexts observed. In veter, he produced a phoneme with a CoG of 5486.5 Hz, which is closer to the value of zametili (5830.0 Hz) rather than the ‘-te-’ context (6157.7 Hz). This suggests that this speaker articulates the vowel following the palatalised occlusive in veter similarly to the standard pronunciation /vjetjir/.

R2 exhibits slightly more variability than R1. The average value of veter (5241.0 Hz) indicates that he articulates the vowel more like /jɛ/ than /i/, as it is closer to the value of the ‘-te-’ context (5191.1 Hz) than the zametili one (4824.2 Hz). This speaker also produced the affricate when reading the word dut’, which showed a higher CoG value of 200 Hz compared to the other contexts. The StD of R2 is 1000 Hz less than R1’s, indicating that even though these two speakers are from the same city, they disperse articulatory energy differently when producing palatalised occlusives.

In the word veter produced by R3 from Volgograd, the CoG value is similar to that of zametili, which is the highest among all contexts. To better understand this, we analysed the F2 of the /i/ following the occlusive in zametili. The second formant (F2) is related to the degree of backness: the more front the vowel, the higher its value. In R1 and R2, the F2 value is 1800 Hz, while in R3 it is higher than 2000 Hz, indicating that he not only produced an anterior consonant, but also affected the following vowel. This speaker also produced an affricate in the ‘-te-’ context, but the average value of the fricative moment is 750Hz lower than those of the release part of the occlusive This suggests that he articulated this affricate in a more central position.

The average StD value shows less energy dispersion compared to R1, but it is similar to R2. The lowest is the ‘-te-’ one (3319.6 Hz), not significantly different from the value of the affricate in the same context (3834.7 Hz).

R4 follows the trend seen above where she produces the highest values of all speakers: the highest is the ‘-te-’ context, with a CoG of 8540.9 Hz. The value of the occlusive in the word veter is more similar to the zametili one, suggesting that she produces a standard pronunciation.

The average StD values are higher than those of R2 and R3, but they are close to the values of R1. As said above, the presence of palatalised occlusives in kutalsja, zakutonnogo and kotoryj was unexpected. In the cases of kutalsja and zakutonnogo, the reason for the palatalisation can be attributed to the vowel following the consonant. These vowels were expected to undergo reduction, a phenomenon typical of Russian, and should be reduced in /ʌ/. However, during the transcription of the recordings, it was noted that some of the speakers articulated /æ/ instead of /ʌ/. When /æ/ was preceded by a /t/, the speakers tended to palatalise it in /t’/.

The average CoG values of R2 (3841.9 Hz) and R4 (4394.1 Hz) suggest that they produced a more central palatalised occlusive, possibly due to a co-articulation phenomenon with the central vowel /æ/. In the word kutalsja all speakers realised a palatalised occlusive with a similar point of articulation, as there is little variation between the values, which range between 3200-4000 Hz. Interestingly, these are low values, suggesting a more central place of articulation compared to the contexts where the phoneme is followed by /jɛ/ and /i/. However, this central articulation could be attributed to the more central vowel /æ/.

Finally, the case of kotoryj is unexpected, as the palatalised occlusive is followed by the vowel /o/. However, only one occurrence was found out of a total of 209 occlusives analysed. This was produced by R2, suggesting that it represents a different articulation from the standard one. Since the phoneme was followed by the back vowel /o/, the CoG and StD values are lower (CoG 3613.4 Hz, StD 3102.5 Hz).

The Kazakhs, on the other hand, show less confidence in producing the palatalisation, as they articulated 45 palatalised occlusive less than the Russians, and 12 affricates.

Starting with K1, his average CoG values are higher than those of the Russians. The highest is 7059.4 Hz in zakutannogo, with an almost interdental articulation. The StD is also high, indicating a great dispersion of articulatory energy. In the ‘-te-’ context, the CoG of the palatalised occlusive is 6731.5 Hz, while the one of the affricates is 6101.1 Hz. More central are the occlusives in zametili (CoG 5573.8 Hz; StD 5222.7 Hz), veter (CoG 5444.3 Hz, StD 5401.9 Hz) and kutalsja (CoG 4903.7 Hz, StD 5785.8 Hz). It is important to note that veter is similar to zametili, suggesting that this speaker articulates veter in a standard way. In the word-final context, the CoG value is 3928.9 Hz, nearly 3000 Hz lower than that of zakutonnogo. However, its StD value is 4728.4 Hz, which aligns with the other contexts observed. In conclusion, K1 exhibits diverse values, but they do not significantly differ from those of Russian speakers.

Secondly, K2 is the speaker with the lowest number of occurrences, having produced voiceless dental occlusives without palatalisation, totaling 11 phonemes analysed. His CoG values vary from 5368.6 Hz in veter to 6682.8 Hz in zametili. Interestingly, these values suggest that he articulates veter with a more anterior position than the standard pronunciation. The StD values for these occurrences are similar to each other, as 5943.4 Hz is the one of veter and 5737.1 Hz the zametili one. The StD of the ‘-te-’ context is similar to that of zametili. In this context, the CoG of the occlusive is 6230.4 Hz, while the CoG of the affricates is 7315.4 Hz, indicating an almost interdental articulation, and the speaker disperses more articular energy when producing them. In the word-final context ‘-t’’, K2’s values are intermediate between the ‘-te-’ context and zametili. The CoG in this case is 6453.9 Hz, and the StD 6520.6 Hz, which is higher than in the other contexts.

Continuing with the analysis of the Kazakhs, K3 exhibits significant variability in his articulation. In the ‘-te’ context, he produces the most anterior palatalised occlusive (5822.4 Hz), while the least anterior is the word-final context, like K1, with a CoG of 4471.2 Hz. Interestingly, K3 frequently substitutes occlusives with affricates. In the -t’ contexts, the CoG is 4985.5 Hz, higher than the palatalised occlusive in the same context; whereas in the ‘-te-’ context it is 5286.4 Hz, which is lower than the occlusive. He also articulated an affricate in veter, where the CoG value is 5335.8 Hz, higher than the palatalised phoneme in the same context (5156.4 Hz). In zametili, K3 only articulated the palatalised occlusive, with a CoG value (4968.2 Hz) similar to veter, suggesting a standard pronunciation in this context. The StD values reveal that the speaker disperses more articulatory energy when producing occlusives compared to affricates in the same contexts.

Finally, K4 articulates the occlusives similarly to R1, with the highest CoG value in the ‘-te’ context (6854.6 Hz) and the lowest in the ‘-t’’ context (5591.7 Hz). The veter value (6115.3 Hz) is more similar to zametili (6012.5 Hz), suggesting a standard pronunciation. The CoG values of the affricates produced K4 are articulated more posteriorly. The StD is also similar to K1, with a maximum difference between the two speakers of 100 Hz. The StD of the affricates is lower than that of the palatalised consonant in the same linguistic contexts.

Following the same methodology as for the dental affricates, we also analysed the duration of the palatalised occlusives. The data was extracted from the third reading task and then normalized (Table VIII).

First, we analysed the expected occurrences. For all the speakers, zametili was the shortest value, and the occlusion phase was not significantly different for R1, R2, R3 and K3. However, R4 and K1 displayed a very rapid occlusion phase, which was more relevant compared to the release phase. On the other hand, K1 and K1 showed a longer occlusion phase.

The ‘-t’’ context appears to be heterogeneous among the expected occurrences. The Russian participants all had similar occlusive moments, while the overall duration of the entire phoneme varied. R3 produced the longest one (301 ms), followed by R4 with 281 ms, R1 with 269 ms, and R2 had the shortest value at 172 ms, which resembled the value found in zametili.

Interestingly, the affricate produced by R2 was longer than the palatalised occlusive, but its occlusive moment was similar to that of the palatalised phoneme.

Kazakh speakers exhibited slightly higher values. K1’s duration is similar to R3, with an imperceptible difference of 2ms; however, the occlusion phase than that of the other participants. K2 has the longest duration (334 ms), but the occlusion moment is similar to that of the Russians. K3 and K4’s durations were more akin to the Russian speakers. The affricates of K3 and K4 were long (378 ms for the former and 274 ms for the latter), but their occlusion phases were like R2’s.

In the ‘-te’ context, the palatalised occlusive had the longest value for all participants, except for K1, who produced the longest phoneme in the ‘-t’’ context. R1, R2 and K1 had very brief occlusion phases. R4 and K3 had the occlusive moment identical in proportion to the entire phoneme, and the latter was similar to K4. R3 produced a very long palatalised occlusive (422 ms), and the occlusive moment was also lengthy, with these val ues being slightly higher than that of the affricates produced by K2, who did not articulate any palatalisation in this context during the third reading task.

In the veter context, R1 had similar values to R2, but the occlusion phase of the latter was very brief, while the moment of R1 was quite similar to R3, albeit slightly higher than R4. Among the Kazakhs, K1’s values are similar to K2, but the occlusive part is longer than the release phase. K3’s duration is 57 ms shorter than K4, but the occlusion phase is similar between the two. Lastly, the affricate produced by K3 was shorter than the palatalised occlusive counterpart.

Analysing the unexpected occurrences, we have two instances of kutalsja by R3 and K1. R3’s value is slightly higher than K1’s, but they both show the same duration of the occlusion phase, which is higher than the release moment. The same pattern, where the release phase is shorter than the occlusion, is also found in kotoryj by R2 and in zakutannogo by R4 and K1. On the other hand, in this last context R2 articulates a very short occlusion in relation to the entire phoneme.

In summary, Kazakh speakers exhibit a higher degree of variability in the occlusion moment, where the release phase is sometimes shorter than the occlusive one. Notably, when considering the two speakers who use Russian daily, K1’s articulation shows significant variability, while K4’s values closely resemble those of Russian native speakers.

As for the phonemes seen before, for the analysis of the fricatives we used PRAAT to identify and annotate the phonemes with the tags “FRIC” and “FRIC PAL”. We then extracted the CoG and StD values (Table IX).

The data reveals considerable variation among the speakers. For Kazakhs, R2 and R4, the /ʃʲː/ in the final-word position has a significantly lower CoG value than the initial-word position, with a difference of nearly 1000 Hz. One possible explanation for this variation is co-articulation, as the /ʃʲː/ is followed by the vowel /i/ in the word sčitat’sja. To support this hypothesis, we examined the F2 values of the vowels /i/ in sčitat’sja and /a/ in plašč. The average F2 value for /a/ was 1753.64 Hz, while for /i/ it was 2457.24 Hz. This indicates that /i/ is articulated more anteriorly than /a/, which influences the articulation of the preceding or following fricative.

Russian speakers did not exhibit significant differences between palatalised and non-palatalised fricatives. However, there was notable variety among their values. For instance, R1 had the most anterior and energy-dispersive phoneme in the word čto, while for R2, the other speaker from Saint Pietersburg, this was one of the lowest values. R3 produced fricatives with CoG values consistently above 4000 Hz, and he was the only speaker with notably high values for the palatalised fricatives, even though the differences between these and the non-palatalised ones were negligible. This similarity also extended to the StD values. R4, on the other hand, had the highest values among all participants, following the patterns observed in the previously analysed phonemes. His values displayed significant variation, with a difference of almost 2000 Hz, where čto had a CoG of 6627.3 Hz and plašč of 4951.4 Hz. The StD values showed a 900 Hz variation, compared to the 500 Hz variation observed for the other Russian participants.

Regarding the Kazakh participants, their values displayed significant variation. In most cases, speakers had the highest CoG values in čto, similar to the trend observed for R1 and R2, except for K2, who had the highest value in sčitat’sja. Notably, the Kazakh speakers dissipated more articulatory energy compared to the Russians, except for K3, who had values more akin to the Russian speakers. These values also showed less linearity compared to the CoGs. As for the lowest StD values, all the Kazakhs had them in the word rešili, while K1 had his lowest value in plašč, like R2 and R4. The highest StD value depended on the contexts, and it is worth noting that K1 and K4 had it in šel, like R4. Another interesting observation is that K2 had his highest StD value in the čto context, similar to R1 and R3, with a value of 5037.7 Hz, the highest among all the participants.

The CoG and StD values also exhibit significant variation within the same speaker. K1, for instance, has a 1200 Hz variation, while K2 shows a 2000 Hz variation. In K2’s case, rešili has the lowest value, and sčitat’sja is the highest. This is interesting since both fricatives are followed by the /i/ vowel, indicating that K2 is aware of the difference between palatalised and non-palatalised fricatives and is trying to emphasize this contrast. K3 and K4 exhibit values more similar to the Russian speakers. It is worth noting that K3 has a low CoG value in solnyško, but a high StD value, suggesting a posterior articulation with a considerable amount of energy dispersion.

In conclusion, there is no clear and consistent distinction between the two phonemes, except in the case of K2, where he hyper-articulates the difference. These results confirm the findings of Spinu, Kochetov & Lilley (2018), who compared palatalised and non-palatalised fricatives and concluded that there is no distinctive difference between them.

The duration of the fricatives was also analysed, following the same methodology used for the other phonemes. The values are presented in Table X.

Firstly, it is evident that solnyško is the shortest fricative for all the participants, except for R1, who articulated solnce instead of the expected word. The shorter duration of solnyško can be attributed to the fact that this fricative is preceded by /ɨ/, which has an average F2 of 1650 Hz. This F2 value places the articulation more posteriorly than that of /ʃ/. Additionally, /ʃ/ is followed by /k/, an occlusive with a longer voice onset time (VOT) as discussed by Zmarich, Bortone, Vayra, & Galatà (2013). This long VOT can lead to a more posterior articulation or aspiration. Since there was no aspiration evident in the spectrograms, the participants articulated it in a more posterior manner. The fricative in solnyško is positioned between two posterior phonemes, which contributes to its faster articulation, emphasizing the surrounding phonemes.

All participants produced the non-palatalised fricative in a word-initial position as the longest in duration. Specifically, R2, R3, R4, K2, K3 and K4 articulated the longest fricative in šel, while R1 and K1 did in čto. The second longest fricative varied among participants: for R1 and K1, it was in šel, while for R3, R4, K3 and K4 in čto, and for R2 and K2 in plašč. Kochetov and Lobanova (2007) studied the Russian language in Komi-Permyak and found that fricatives in a word-final position are generally shorter than those in a word-initial position. This finding is partially reflected in our participants, as two speakers articulated the fricative in plašč as longer than the one in čto.