This post is an answer to Herman’s comment on my Spanish accents post.
Let’s start with some basics: Spanish has two rhotic sounds, which are usually called vibrantes in Spanish dialectology. The one called vibrante simple is an alveolar tap ɾ, and the one called vibrante múltiple is an alveolar trill r. The first sound is found in English, for example in American English data. However the first vowel is pronounced, the T is not a plosive t, aspirated or unaspirated (like in time or stop), or a glottal stop (like in but). It is an alveolar tap. The trill r is not found in English, but you can hear it when people try to imitate, say, the sound of a motorcycle engine.
The trill r appears word-initially and syllable-initially after another consonant, where it is spelled R (rato, Enrique). The tap ɾ appears syllable-initially in consonant clusters and syllable-finally, where it is spelled R (brazo, mar). These sounds are only phonemically contrastive when intervocallically; consequently, this is the only place where they are spelled differently: RR for r and R for ɾ (perro, pero).
Because of the phonemic neutralization I mentioned above, syllable-final ɾ can be pronounced r without any risk of misunderstanding. A good example of a situation where this is done for dramatic effect is the narrating style of radio soccer commentators in Spanish-speaking countries.
The syllable-initial consonant clusters in which ɾ can appear are /pɾ bɾ tɾ dɾ kɾ gɾ fɾ/ (prisa, brazo, tres, madre, cruz, grasa, frase). It can also form a cluster when syllable-final before another syllable-initial consonant (arma, arnés, arlequín, arcángel, ardiente, artículo). Likewise, r can form a cluster with a preceding syllable-final consonant, like for example /l n s/ (alrededor, enriquecerse, Israel).
Herman tells us:
As far as different phonetic environments are concerned, the realization of rhotics in different environments has always been quite difficult for me to grasp as a native English speaker and an adult language learner of Spanish. Notably, the realization of the rhotics in consonant clusters. In my experience with Spanish, essentially after any nasal consonant or lateral consonant, rhotics are realized as [r]. However, before nasal consonants and lateral consonants, my perceptions seems to either rhotics being realized as catch [r] or complete dropping depending on the speed of the utterance.
Herman’s impression about the realization being r after nasals and laterals matches mine. I can definitely hear the trill, and I can also percieve it articulatorily (alrededor, enriquecerse). After /s/, in my dialect of Spanish, where that /s/ is aspirated (realized as h), the r can be partially or totally devoiced due to voicing assimilation to the preceding consonant (Israel). The voicing assimilation can also go in the opposite direction and make the h a voiced ɦ. The same can happen in accents which do not aspirate /s/, making the allophone a z instead.
About ɾ before nasals and laterals (arma, arnés, arlequín), I’m not sure what Herman means by “catch r“. As I mentioned above, syllabe-final /ɾ/ can be realized as trill r without any risk of misunderstanding, and it does happen as a function of style in certain situations. I imagine that when he refers to dropping depending on the speed of the utterance, he means that it is dropped when the utterance is fast. I don’t think I agree with his impression, but I think I know where his impression is coming from. The fact is, ɾ has a vocalic component in Spanish in consonant clusters. You can see this in a spectrogram, and you can even percieve it articulatorily and auditorily if you pay close attention. The vowel in question is very short, appears before or after the ɾ depending on whether the consonant that forms a cluster with it happens before or after it, and is indistinct in quality, usually matching the vowel in the next or previous syllable.
Because of this, when a non-native student has trouble pronouncing Cɾ clusters (especially tɾ) in a way approximating a native pronunciation, I tell them to first insert a vowel between the consonant and the ɾ, and then shorten it more and more until a native-like cluster is produced.
This epenthetic vowel is also probably the reason /b d g/ are realized as fricatives or approximants when preceded by ɾ and followed by a vowel (pardo). In a technical sense, those voiced plosives are actually intervocalic. The same is true when these plosives are preceded by a vowel and followed by a ɾ (abrir, uno gris).
In saying words like arnés, arlequín (clusters ɾn and ɾl), in phrases articulated very fast, I notice that the epenthetic vowel between the ɾ and the following consonant tends to disappear, and the tap is not pronounced with a full contact-lack of contact sequence before the following consonant, but instead I perceive a contact, then the following consonant immediately after. We could say, borrowing terms usually used when referring to plosives, that the ɾ has the usual approach, but then it is released nasally or laterally.
Because of that, I understand why it may sound like outright dropping of the ɾ to Herman. On the other hand, before /m/ (arma), the ɾ is pronounced as a proper tap, at least in my production.
Then again, maybe in the examples Herman was listening to, the ɾ was actually dropped. Dropping word-final ɾ is definitely not unheard of in Spanish dialectology, so dropping syllable-final ɾ is not something unlikely either. In the Buenos Aires accent, the ɾ in porque is dropped quite often.
Word-final ɾ dropping is common in Andalucía, and I have also heard it from some speakers in Buenos Aires. This type of ɾ-dropping is especially common in the infinitive forms of verbs (hablar, comer, vivir).
Similarly, whenever a rhotic precedes [ð̞], I seem to hear 3 different realizations based on different factors which result in the following:
-The stress of the word in a phrase typically resulting in [r] before [ð̞].
-The speed of the utterance typically resulting in either [ɾ] or some kind of vowel quality change if the speed of speech is really fast (still can’t pinpoint what it is).
-Relaxed somewhat normal speed of speech which results in [ɾ].
This also makes sense to me. As I mentioned above, syllable-final r instead of ɾ is definitely possible, and at normal speed I also perceive ɾ in my production (pardo, cardo, fardo). However, I notice the ɾ being fronted to a dental or almost dental articulation, because of assimilation to the following consonant. When saying the word very fast, I notice that the tapped ɾ doesn’t always really reach the teeth to actually produce a tap, but is instead kind of an approximant. Because of this, I’m not surprised that Herman hears it as a change in the quality of the preceding vowel – maybe vocalization of the ɾ is the best analysis for what happens there, in which case one could analyze the result as a diphthongization of the preceding vowel. It’s definitely an interesting phenomenon and it reminds me of the centering diphthongs which developed in non-rhotic English as a result of the vocalization of historic /r/.
Shifting the focus on different dialects, there seems to be two different realizations of the rhotics in Mexican Spanish from central Mexico and Puerto Rican Spanish. In Mexican Spanish particularly from central Mexico, phrase final /r/ seemed to be realized as some sort of voiceless fricative. It seemed almost as if it was a type of voiceless lateral fricative with the tip of the tongue leaving just enough space for air to pass between the alveolar ridge and tip of the tongue. In regards to Puerto Rican Spanish, in an undergraduate Spanish phonology class I took, we observed that the realization of the rhotics was a type of voiceless fricative. I can’t remember exactly if it was a glottal fricative /h/ or something similar.
My question for you is: have you ever heard of any of these cases similar to these?
I have indeed heard the Central Mexican voiceless fricative allophone of phrase-final /ɾ/. It’s not just a voiceless ɾ; it does indeed sound like it has been fricativized. Auditorily, there is a similarity to the kind of voiced fricative realization of /r/ that you can hear in the Argentinian northwest, though voiceless and shorter; so I would agree with Herman’s impression of “the tip of the tongue leaving just enough space for air to pass”; i.e. a grooved fricative. It doesn’t sound lateral to me, though; but auditory impressions aren’t always right, and we should read up on it more and see if anyone has done any spectrograms, real-time MRIs, etcetera. You can listen to more audio from that speaker and from others from the Mexico City area here.
About the Puerto Rican thing, I had never heard of this before, so I did some quick research. I listened to three clips of the same speaker, here, here and here. His realizations of /ɾ/ did not seem to differ at all from mine: voiced and tapped. His realizations of /r/, however, were slightly different. My impression is that the r is preceded by a short glottal fricative, which can either be voiceless, which causes the r to be partly voiceless, or the whole thing can be voiced throughout; so, hr or ɦr.
I hope this answered your questions, Herman. I found it very interesting to try to find answers to them.