Acoustic Analysis of the Pair /s/-/z/ Produced by Modern Greek Learners from Spain: Comparative Interlevel Study
Author: Parrón Bellido, Íker (Universidad of Salamanca) ︱Editor: Amelia Kobylarz
The opinions expressed here are solely those of the author and do not represent the views of U-Lingua, its editors, or any affiliated organisation. Any copyright issues should be taken up with the author themselves.
The study of human languages’ sounds is a multidisciplinary research field that contains a vast amount of subdisciplines. The primary ones are phonetics and phonology. Within this journey through the science of acoustic signals, various approaches can be taken. Phonology focuses on the abstract representation of linguistic sounds, directing its efforts toward identifying the articulable phonemes across all or specific natural languages. A key characteristic of phonemes is their ability to distinguish meanings, as demonstrated by minimal pairs such as desk and disk or buy and pie. Phonetics, on the other hand, examines the physical aspects of the sounds, viewing them as a combination of acoustical waves produced following several articulatory patterns. This process of production generates phones, which, as opposed to phonemes, cannot distinguish meanings. By studying the interlinguistic differences regarding the summary of sounds that are comprehended in a particular language in comparison with another one, we can easily observe the existence of several sounds that are not in the other language orthose different, but we perceive as the same. This phenomenon, known as phonological deafness (Polivanov, 1931; Troubetzkoy, 1939), plays a critical role in second language (L2) learning. When students cannot perceive the differences between the sounds of the L2 and those of their native language (L1), confusion between phonemes can occur, potentially altering sentence meanings (De Almeida Sandes, 2010). For example, between Modern Greek and Spanish, despite their phonetic similarity in the majority of the sounds, there is a highlighted pair that is distinguished phonologically in Greek, but phonetically in Spanish: it is /s/ - /z/ (Κλαίρης et al, 2004; Quilis, 2012). Differentiating these sounds is essential in Modern Greek, as they represent inflectional contrasts. Specifically, it is essential for this differentiation to the point of representing flexive contrasts in Modern Greek: /s/ is used for the past simple conjugation, while /z/ is used in the past continuous (διάβασα – διάβαζα). Consequently, correct pronunciation in such a case is critical. Both sounds are acoustically characterized by the presence of aperiodical soundwaves on the oscillogram and by their amplification on the top of the spectrogram (3000 Hz<). The voiced sound can be differentiated from its voiceless counterpart by looking if there is a voicing formant at the bottom of the spectrogram and by calculating its duration. As noted by Nirgianaki (2014), the voiced phoneme requires less time to produce than [s], as we can see on the following spectrogram:
Image 1. Spectrographic and oscillographic representation of [z] and [s] via Praat. Source: self-made.
This study examines the ability of Spanish speakers learning Modern Greek as a second language (L2) to distinguish linguistic features, focusing on the acoustic analysis of spectrographic representations using Praat. The acoustic samples were recorded with a Tascam DR-05X recorder at the University of Salamanca (Spain). The study considers the participants' proficiency levels in Greek: A1 (2 participants), B1 (1 participant), C2 non-native (1 participant), and a control participant who is a native speaker of Greek. The analysed materials consist of laboratory speech obtained from the reading of a phonetically calibrated text. Each participant completed this task, yielding 10 acoustic samples per person, so it is expected to obtain canonical realizations of [z] in all levels. However, from all the contexts that could appear the /z/phoneme, we will only look at those where Spanish speakers wouldn’t articulate allophone [z], which appears in syllabic coda followed by voiced consonants (Quilis, 2012).
From the first informant, a native Greek speaker, all realizations were [z], as expected. Moreover, the mean duration of all the [z] materializations was calculated to be 0.0506 seconds. This serves as a reference to compare differences in the articulation of [z] or [s] among the other participants. The informant at level C2, who was expected to articulate [z] when required, showed results consistent with the native control informant, except for one case, which was attributed to a conjugation error. The mean of the duration of the [z] sound for the C2 informant was 0.0767 seconds, suggesting a potential difficulty in its articulation or the unfamiliarity of this sound within the speaker's phonological system. The situation differed significantly for the B1 and A1 level informants. The cases of apparition of [z] decreased significantly, with zero cases recorded for the B1 informant and one of the A1 informants. The B1 speaker obtained a mean duration of 0.1397 seconds, which represents a significantly higher value than the native [z] duration. Similarly, the other informant with zero [z] articulations exhibited a mean duration of 0.1310 seconds. The last, A1 informant, who produced other allophonic solutions such as the palatal sibilants [ʃ] and [ʒ], likely influenced by their Asturian dialectal background (Academia de la Llingua Asturiana, 2021), produced only one instance of [z] with an extended duration of 0.0945 seconds. The mean duration in this case was similar to the previous two informants, measuring 0.1356 seconds. Notably, some participants attempted to reinforce [s] to approximate [z] by introducing supporting phones before the sibilants, such as [d], [t], [k], or [ð]. These do not correspond to [z] with morphological differentiation capacity but indicate an awareness of the speakers of the difference between [z] and [s] in specific sequences.
In conclusion, while the data suggest that the articulation of [z] can be challenging even for advanced Spanish learners of Modern Greek, improvement in its pronunciation is evident as linguistic competence increases. Contrary to the initial hypothesis, there is no evidence that phonemes distinguishing morphological features yield better results than those without such values. Future research could be improved by increasing the number of informants or recording additional acoustic samples to achieve more accurate and reliable results.
References
[1] Academia de la Llingua Asturiana. (2021). Normes ortográfiques (8ª ed.). Academia de la Llingua Asturiana.
[2] De Almeida Sandes, E. I. (2010). Teorías y modelos de adquisición y aprendizaje de los sonidos de una LE y un breve análisis de las dificultades de los estudiantes brasileños de ELE. Universidade Presbiteriana Mackenzie: Universidade de São Paulo, 151–164. Retrieved from https://cvc.cervantes.es/ensenanza/biblioteca_ele/publicaciones_centros/PDF/saopaulo_2010/11_almeida.pdf
[3] Κλαίρης, Χ., Μπαμπινιώτης, Γ., Μόζερ, Α., Μπακάκου-Ορφανού, Α., & Σκοπετέα, Σ. (2004). Γραμματική της Νέας Ελληνικής. Δομολειτουργική-Επικοινωνιακή (10η έκ.). Ελληνικά Γράμματα: Αθήνα.
[4] Nirgianaki, E. (2014). Acoustic characteristics of Greek fricatives. J. Acoust. Soc. Am., 135, 2964–2976. Retrieved from https://www.researchgate.net/publication/262226543_Acoustic_characteristics_of_Greek_fricatives
[5] Polivanov, E. (1931). La perception des sons d’une langue étrangère. Trav. Cercle Linguist. Prague, 4, 79–96.
[6] Quilis, A. (2012). Principios de fonología y fonética españolas. Editorial Arco: Madrid.
[7] Trubetzkoy, N. (1939). Falsa apreciación de los fonemas de una lengua extranjera. In Principios de fonología (D. García Giordano & L. J. Prieto, Trans., 1973). Cincel: Madrid.