I recently came across a fascinating research paper by Eugenio Martínez-Celdrán of the Universitat de Barcelona about “chimeras” of Spanish phonetics. A chimera is, essentially, an unrealistic notion. The paper presents three case studies in which traditional phonetic description of the Spanish sound system has under- or overshot the mark.
The three case studies include:
- Stops, fricatives, and approximants (a discussion of the voiced series /b, d, g/)
- Devoicing and voicing (primarily phonetic outcomes of /p, t, k/)
- Apico-dental vs laminal-denti-alveolar (phoneme /d/)
The paper’s main strength is its empirical foundation: arguments are well supported by sound spectra and palatography. A secondary strength is its implicit reminder that the phonetics of any language is much less tidy than the teacher or student might wish it to be, and viewpoints are often held out of convenience rather than because they reflect empirical conclusions. Hence the need for edited volumes such as the one from which this article is excerpted: Selected Proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology (2008), edited by Laura Colantoni and Jeffrey Steele (Cascadilla). Over the past decade or so, the field of “laboratory phonology” has grown monumentally, due to an upswing in interest in accurate phonological description as well as increasingly easy access to “laboratory” equipment by ordinary individuals rather than well-funded university departments. For example, sound spectrography, a longstanding keystone of laboratory phonological analysis, can now be done on just about any laptop using free downloadable software.
Of the three chimeras Martínez-Celdrán covers, I find the first one most applicable to a pronunciation lesson. The Spanish voiced obstruent series /b, d, g/ includes three allophones, traditionally fricatives [β, δ, γ], which pervade spoken Spanish and are therefore essential to a good “accent.” Martínez-Celdrán rounds up evidence that the three allophones are better described as approximants— i.e. more like [j] and [w] than like [v] and [z]. Over the years, various phonetics classes of mine have applied playful nicknames to these allophones, such as “leaky” and “relaxed.” On sillier days it has even been proposed that accurate pronunciation of the series requires a decent level of intoxication to ensure that the vocal apparatus (particularly the tongue) is adequately relaxed. Indeed, the designation “fricative” does not seem right, since even the less observant student knows that the pronunciation of nada as [náδa] is not commonly heard. In light of this case study, the descriptors “leaky” and “relaxed” are far more accurate than we thought and, in fact, may be the most practical way to describe an approximant without actually delving into thick articulatory terminology, formant structure, palatography, etc.