Have you ever had the following experiences? Trying to find a lost document in computer folders; or trying to track a lost memory in your mind; or searching for a runaway pet. One way or another, people in those circumstances do not simply dive into relentless search. This would be purposeless! Instead, they immediately look for tips, tracks and footprints that can possibly guide one's personal investigation. Even though those remnants may take on variable formats and shapes, somehow people know what they are looking for. Believe it or not, sometimes, a linguist needs to engage in a similar investigation when dealing with vowels.
Acoustic analysis shows that, depending on contextual features, vowels may no longer be produced; they may disappear. That speech event is contrary to a common understanding for most Brazilian Portuguese speakers: words in Portuguese always end in vowels, despite some sounds such as /s/ and /m/. In other words, they think that words in Portuguese never end in consonant sounds, such as /t/ or /g/. The turning point is that speakers can track those vowels again, both by comprehension skills and by acoustic footprints.
For discussing what happens when a vowel disappears - what can be called “elision” - it is relevant comprehending as common vowels look like, their main characteristics. In the following spectrograms - visual transcriptions of speech signal - the pink band delimits cycles of an effective final vowel production. Figure 1 selects part of the production of “caco de” and Figure 2 selects part of the production of “caco cortante”. Both highlight the final /o/ sound from “caco”.
Focusing on the green frame in pictures, it can be seen that both speech wave signals have 7 or more cycles. A cycle can be identified by the repetition of a pattern, starting from a intersection point with the horizontal line until the next similar point, passing over the highest and the lowest signal pikes. Moreover, both vowels indicated in the spectrograms have a dark band in the bottom (the fortis), highlighted by the blue frame. The fortis in the bottom indicates vibration of vocal folds. These features - wave cycles and existence of bottom dark band - are essential for acoustically characterizing vowel sounds.
However, it is not always this way, for vowels may disappear. Acoustically, this can be seen in the following spectrograms, that are the deleted counterpart of the previous vowels. Thus, Figure 3 and Figure 4 present the elision of the final /o/ sound from “caco”.
As can be seen, in Figure 3 the green frame shows the production of only 2 cycles. Even though there is a bottom dark band (blue frame), the selected speech signal cannot be considered an effective vowel production, for Acoustic Phonetics research considers 3 cycle waves (or less) a misproduction. Also, for Figure 4, there are not wave cycles (green frame) and there is a faint bottom fortis (blue frame).
The question is why it happens. What are the main influences for such acoustic event? According to Câmara JR., there is an important difference in phonetic quality - articulatory movements and auditory consequences - between stressed and unstressed vowel sounds. The Linguist says unstressed vowel sounds present less phonetic quality than stressed vowels. Consequently, the first is not so relevant for distinguishing phonetic characteristics, and can be easily reduced or deleted.
Another important aspect for allowing vowel elision is the area where each vowel is produced inside the vocal tract. A study conducted by Viegas & Oliveira shows that there are more chances of elision for closed (or high) vowels, such as /i/ and /u/. These vowels are produced with the elevation of the tongue till very close to the roof of the mouth. Acoustic Phonetics assumes that these vowels have a lower intrinsic “time”, that is to say they are naturally shorter and tend to disappear easily.
Moreover, depending on the immediate context, vowels tend to disappear. Research such as from Dias & Seara shows that vowels tend to disappear in unvoiced contexts. They can be, for example, the end of words and sentences, where the vowel is side by side with a pause (a section without vocalic production). Also, when vowels are close to strong unvoiced consonants, such as from occlusive - /p/,/t/ and /k/ - and fricative sounds - /s/ and /ʃ/ -, elision is more common.
At this point, you may be asking yourself: how can we still understand each other without vowels? Essentially, because vowels disappear but left behind them footprints that can be interpreted acoustically.
Dias & Seara present some of this footprints that are, in general, “parts” of what could have become a vowel. For example, in Figure 3 some cycles can be distinguished, even though there are not dark bands (fortis) in the spectrogram. Also, Figure 4 presents a dark band in the bottom of the image, even though there is not cycle production.
An important aspect about languages - generally unknown by speakers - is that speech production and speech comprehension are different things. That is to say that, even though you suppose someone said “o gato gordinho”, that is just an abstraction made by your brain trying to fill the “lacks”. What the person could actually have said was /gat/ instead of /gato/, but the meaning is not committed by such alteration.
In fact, elisions and other speech events may happen because human brain knows how to deal with them. Our complex neurological system connects all linguistic levels - phonology, morphology, syntax, semantics, prosody, etc - to understand and interpret sound impulses received by our auditory system, connecting production and perception. Briefly, if one level has some missing parts, such as deleted vowels, our mind looks for other tips to interpret the signal, like morphology and syntax. Getting those information together, our mind supposes what could actually fill those linguistic “lacks” for effective communication.
Unfortunately, there are few studies in the area of vowel elision, especially for final-word vowels. That happens because these sounds are extremely variable, due to their fragility facing linguistic changes and perturbation. Recent studies have been approach events, such as elision and vocalic harmony, as a frame of different influences coexisting together. These are probably the ongoing approaching for future research in the area of vowel elision for Brazilian Portuguese.
BERTI, L.C. Relação entre Produção e Percepção de Fala: Coerência com o Parâmetro Fonético-Acústico. Cadernos de Estudos Lingüísticos, Campinas, 50(1):45-67, 2008.
DIAS, Eva Christina Orzechowski; SEARA, Izabel Christine. Redução e apagamento de vogais átonas finais na fala de crianças e adultos de Florianópolis: uma análise acústica. Letrônica, Porto Alegre, v. 6, n. 1, p. 71-93, jan./jun., 2013.
MENESES, Francisco de Oliveira. As vogais desvozeadas no Português Brasileiro : investigação acústico-articulatória. 2012. Dissertação de Mestrado - Universidade Estadual de Campinas (UNICAMP). Instituto de Estudos da Linguagem.
OLIVEIRA, B. de; SILVA, S. F. da. Redução das Vogais Átonas Finais no Falar de um “Manezinho”. Revista uox, n. 02, 2014/1. P.31-41.
SCHWINDT, L.C. Revisitando o Estatuto Prosódico e Morfológico de Palavras Prefixadas do PB em uma Perspectiva de Restrições. Revista de Linguística. Alfa, São Paulo, 52 (2): 391-404, 2008.
VIEGAS, Maria do Carmo; OLIVEIRA, Alan Jardel. Apagamento da vogal átona final em Itaúna/MG e atuação lexical. Revista da ABRALIN, v. 7, n. 2, p. 303-322, jul-dez. 2008.