[ Note to regular phonoloblog readers: this is a follow-up on my last Language Log post, on the Chilean Spanish pronunciation of Pinochet. ]

Here is a key for redirected Language Log readers who may not be familiar with phonetic terms and transcriptions. The phonetic value of any other symbols used below should be transparent. (IPA = International Phonetic Alphabet, APA = [North] American[ist] Phonetic Alphabet):

• The sound that is usually represented orthographically by “ch” in English (as in e.g. chain) is a voiceless postalveolar affricate and is represented as either [ʧ] (IPA) or [č] (APA).
• The sound that is usually represented orthographically by “sh” in English (as in e.g. shame) is a voiceless postalveolar fricative and is represented as either [ʃ] (IPA) or [š] (APA).
• The sound that has no single usual orthographic representation in English but is sometimes represented as “zh” (as in e.g. measure) is a voiced postalveolar fricative and is represented as either [ʒ] (IPA) or [ž] (APA).

---

To double-check my claims about different American Spanish varieties, I first consulted the superb book by D. Lincoln Canfield pictured and Amazon.com-linked on the right. According to the book description, “[t]his book represents the culmination of a lifetime of research in the spoken Spanish dialects of the Americas by one of the foremost experts in this field.” Indeed. Canfield makes the useful organizational decision to devote a separate chapter to the discussion of pronunciation patterns found in each country on the continent (including the U.S.), though he is clear about the fact that differences between varieties of a language do not necessarily respect national boundaries (pp. 20-21). Each chapter includes a map (in some cases, multiple maps) highlighting certain key (geographically-definable) pronunciation patterns. It’s an amazing piece of work, mercifully short (130pp.), and at $14 from Amazon.com, a real steal. (Makes a great gift, too!)

But: this book was published a full generation ago (1981), so it’s getting a little out of date. To supplement this, then, I also consulted the two books pictured and Amazon.com-linked below the fold.

The first is the book by José Ignacio Hualde that I mentioned in my Language Log post (this was the only book I happened to have on hand at the time). This book was published just last year so it’s much more recent, but it is also more of a textbook about Spanish sounds (phonetics and phonology) generally and less well-suited as an easy reference for finding out where things are pronounced how. Still, there’s lots of information about different dialects throughout the book, and it comes with an audio CD on which “[a]ll the sounds discussed in this book are demonstrated”, according to the book description. I highly recommend this book to anyone with an interest in Spanish language and linguistics and/or phonetics and phonology. (And if you don’t trust me, trust Ralph Penny, who writes: ‘This is a masterful and wide-ranging book. Every student and scholar of Spanish phonetics and phonology should read it and keep it by them.’ Now that’s high praise.)

The second is the excellent book by John M. Lipski that is kind of in between the other two in several respects. It was published in 1994, about half-way between the others. Like Canfield’s book, the bulk of Lipski’s book is also organized country-by-country (though Spanish in the U.S. is not discussed, at least not separately). (See Note 1 below.) The rest of the contentful part of the book consists of a 136pp. introductory part on “the evolution of Latin American Spanish” and a 65pp. bibliography (compare Canfield’s 9pp. and Hualde’s 10pp. bibliographies); Lipski is also more comprehensive in that he discusses syntactic, morphological, and lexical characteristics of different varieties in addition to phonological/phonetic ones. Lipski also provides a useful historical overview for each country.

Why am I bothering to give you all this bibliophilial detail? Well, as Bill Poser and Geoff Nunberg (and everyone else at Language Log) will readily appreciate, books are cool. But it’s also by way of saying: these are just a few of the sources that I have consulted to back up what I said in my Language Log post and what I’m about to say below.

---

First, more details about Chilean Spanish. On p. viii of the Preface, Canfield writes:

In conversation with a radio announcer from Valparaíso, Chile, I noted that he consciously pronounced /č/ as [š]. When questioned, he stated that he considered it “más suave [= “softer”].” What he thought of as meticulous articulation actually involved the elimination of an occlusive element, a change that occurred in French during the thirteenth century and that is now taking place in cities of northern Chile, in Panama, in Cuba, and elsewhere.

In other words, the relevant change has happened in at least one other Romance language, and it is happening in several different varieties of Spanish. I’m not sure what Canfield means by “consciously pronounced” and “meticulous articulation”, though. Did the radio announcer really try to pronounce a fricative instead of an affricate? Or was he simply aware — especially when being questioned about it specifically! — that these sounds are different in a specific way and that the fricative he uses is non-standard? My guess is that the radio announcer’s “softer” answer is his aesthetic ex post facto reason for why he “chooses” not to use what should otherwise be the preferred, standard affricate pronunciation.

In the chapter on Chile (pp. 31-33), Canfield notes that one “rather general” fact about Chilean speech is that there is an “alveolar rather than palatal articulation of the /č/” — that is, that it sounds more like [ts]. One of Canfield’s maps (the one on the left in this unauthorized reproduction) indicate that this alveolar pronunciation extends from near Antofagasta in the “Big North” through the Center and South portions of the country and into parts of North Patagonia (see this map for the identification of these regions). Valparaíso, where the radio announcer of the Preface is from, is right in the middle of this large swath of the country. Canfield explains further:

There is an interesting trend in some northern towns that may be attitudinal, but since it is also becoming very popular in Panama and to an extent in Cuba and Puerto Rico, it may indicate further development similar to the one that occurred in French in the thirteenth century: the loss of the occlusive element in /č/, which is then heard as [š]. As has been noted (in the Preface), speakers may consider the simple sibilant “más suave.”

Canfield’s other map (the one on the right) shows the handful of small areas, fairly spread apart in the North, where the fricative [š] is found — and Valparaíso is in the southernmost of these areas. (This explains most of this comment, as I point out just after that.) All of this indicates that the phenomenon is more socially than geographically stratified, as also noted by two commenters on my original post (the second of which self-identifies as a “Native Chilean”) and as this Slate Explainer article from Dec. 12 notes (thanks to Ben Zimmer for the link):

The confusion starts with the ch sound, which can serve as a marker of social class in Chilean Spanish. In educated speech, the Spanish ch is similar to the English pronunciation, as in the word chess. But popular dialect turns the ch into something more like sh. A high-class Chilean would probably pronounce the country’s name as “chee-lay,” while someone with less status might say “shee-lay.” Likewise, the same two people might describe the ex-dictator as “pee-no-chay” and “pee-no-shay.” (Pinochet himself was known for speaking in a rough, working-class style. Listen to him pronounce Chile with an sh, about 24 seconds into this video.)

This brings us back to the NPR story that Geoff Pullum originally told me about around the water cooler at Language Log Plaza. Steve Inskeep’s question to Nathan Crooks was specifically about how Pinochet himself pronounced his own name. (Pace this commenter, who claims to know what “Inskeep’s intended question” was; see Note 2 below.) Like Canfield’s radio announcer, Pinochet is from Valparaíso (according to his Wikipedia entry), and the paragraph quoted just above implies that he’d use the fricative [š]. But the same Explainer article continues:

How did Pinochet himself say it? Three different sources told the Explainer they knew or remembered how the general or his family pronounced the name. And they gave three conflicting answers. You can hear Pinochet utter his own name two seconds into this video clip from 1980–it sounds a lot like “pee-no-chay.” If you’ve come across another audio or video clip in which Pinochet or a member of his family pronounces the name, please send it to the Explainer.

To me, the audio quality of the linked video clip is too poor — and Pinochet says his name too fast — to tell definitively whether he says [č] or [š]. (Listen for yourself.) There even seems to be some paper-shuffling sound, much closer to the microphone, right around the time he says “Pinochet”. (On the other hand, it seems fairly clear that he didn’t pronounce the final “t”; if he had, it would be distinguishable in this context, right before a vowel-initial word — “Ugarte”, Pinochet’s second surname.)

The audio quality of the previously linked video clip is much better. (This is the clip in which Pinochet utters “Chile” with what seems like an [š]; you can hear just that part of the clip here.) This clip is an excerpt from an interview, and the topic of the excerpt is about alleged human rights violations in Pinochet’s detention centers. The word for “rights” in Spanish is “derechos”, and Pinochet says it twice in this clip: the first time with a clear [č], and the second time with a clear [š]. (If you’re interested you can see waveforms and spectrograms for these two sound files here and here.)

Anyway, the Explainer article agrees with me: there’s no single way any given Chilean is expected to pronounce “Pinochet” — but that’s only after having tried to select one pronunciation over the others not just once but twice in the past (thanks also to this commenter for pointing me to the first of these):

Slate has twice tried to get to the bottom of this question: Eight years ago we went with “pee-no-CHAY” but then reversed course last year with “pin-oh-CHET.” So, which is it?

All of the above. There’s no single correct pronunciation for the name in Chile. The first two syllables don’t change too much, and should be something between “pin-oh” and “pee-no.” But the last syllable is up for grabs: Some Chileans go with “shay,” others “chay,” and still others “chet.”

And, in the aggregate, the folks who commented on my Language Log post agree. (For the sake of completeness, however, I would follow the lead of Nathan Crooks, and add “shet” to the Explainer’s list of three pronunciations just above.)

I also have to add to the Explainer’s story about the final “t” issue:

It gets more complicated with the final t. As a general rule, the whole syllable–“chet”–should be spoken aloud. But in casual conversation, Chileans tend to drop the final sound. Someone who pronounced Pinochet as “pee-no-chet” would be correct, but he’d also be speaking in a formal (and perhaps a bit uppity) tone. On the other hand, some Chileans are inclined to use the French pronunciation of Pinochet, since the name is of French Basque origin. In that case, they’d drop the t and go back to “pee-no-shay” or “pee-no-chay.”

I don’t know about the “French Basque” business; Pinochet’s Wikipedia entry says his father was a “descendant of Breton immigrants who arrived in Chile during the 18th century”, and Brittany’s quite a way from the Basque country. (Then again, my parents are both from Bolivia but my surname’s Croatian, so I suppose it could happen.) But anyway, as I explained in my Language Log post, not pronouncing a word-final “t” is expected of most Spanish speakers, not just Chileans; there’s no need to invoke the French pronunciation bit.

One final note on this: it may be tough for a typical English speaker to tell whether or not there’s a final [t] in a given pronunciation of “Pinochet” by a Spanish speaker, for reasons I mentioned in passing in my Language Log post. The monophthongal [e] of Spanish sounds more like the [ɛ] of English bed than the diphthong [eɪ] of English bay, and [ɛ] can’t end a word in English (this comment notwithstanding). Unless the cues you’re listening for are the release of the [t] — which you typically wouldn’t get unless there’s a following vowel — you may get fooled by the vowel quality: if the vowel sounds like [ɛ], and you know (subconsciously) that words don’t end in [ɛ], then you may (again, subconsciously) fill in a consonant after that vowel.

---

And now to correct myself, and to add a little bit more, about [š] in other varieties of Spanish.

First, orthographic (non-final) “y” and “ll” pronounced as [š] is indeed found in the speech of some Uruguayans, but I was wrong to restrict this to Montevideo, and as at least three commenters have so far pointed out, I was wrong to contrast this with Argentinian speech. On Uruguayan Spanish, Canfield (p. 88) writes:

As in a good part of eastern Argentina and in sections of the Northwest, both /lˬ/ and /y/ of the traditional sound system have become [ž], with a tendency toward unvoicing [that is, toward [š]] in recent times, especially among women.

Note that the ‘ˬ’ diacritic is supposed to be under the ‘l’ there; this is Canfield’s voiced palatal lateral symbol (= IPA [ʎ]). Elaborating on Argentinian Spanish, Canfield (p. 24) writes:

In the porteño area and south through Patagonia and east of a line running roughly from Córdoba to Bariloche on the border of Chile, there is leveling [of both /ʎ/ and /y/] to one phoneme whose phonetic form is [ž] and occasionally [š], the latter manifestation becoming more common recently, especially in the speech of women.

(According to Wikipedia, “Porteño is the Spanish demonym for those born in the Argentine city of Buenos Aires“.) One of Canfield’s maps of Argentina — the one on the right — shows the extent of this phenomenon; compare also his broader map of the Americas.

The variety of Spanish that Canfield delimits here (more or less) is known as Rioplatense, which also includes most of Uruguay. (Thanks to Miskwito for the Wikipedia link, where it is also clarified that “either voiceless [ʃ] (this phenomenon is called sheísmo) or voiced [ʒ] (called zheísmo)” are found (in place of /ʎ/ and /y/) in this variety.

Canfield identifies three countries other than Chile with speakers who realize /č/ as [š] (p. 12): “in the younger generation of Panama’s urban centers and sporadically in Cuba, Puerto Rico, and the cities of norther Chile it [that is, /č/] is [š]”.

• Regarding Panama (p. 67): “Among the youth of Panama, especially in the cities, /č/ is rendered as [š]”.
• Regarding Cuba (p. 42): “Many observers have noted that the /č/ is weakening to [š] among Cubans, and here again it seems to be more common among women than men.”
• Regarding Puerto Rico (p. 76): “In the Southwest it [that is, /č/] is [č], and one notes today, as in Cuba, a tendency toward [š] at times.”

Using Lipski’s index I found that he specifically mentions the pronunciation of /č/ as [š] in seven out of his nineteen country-by-country descriptions. Note in particular Lipski’s detailed descriptions of the situations in Panama City (5) and Puerto Rico (7) below, which confirm what I said in Note 1 of my Language Log post … except I said it about Chile. (These particular descriptions also add support to observations made by Penny Eckert and others that younger generations — in particular, young women — “are the movers and shakers in linguistic change”.)

1. Regarding Argentina generally (p. 169): “The affricate /č/ almost never loses its occlusive elements; if this were to occur, it could potentially create a large number of homonyms, given the prevailing pronunciation of /y/ as [š].”
      But, more specifically with respect to the “Northeast/Guaraní-influenced zone” of the country (p. 171): “In Misiones, some older speakers give a fricative pronunciation to /č/, but in this dialect neither /y/ nor /ʎ/ receives a pronunciation which might result in merger with /č/”.
2. Regarding Chile (p. 200): “Chilean /č/ is routinely cited as a distinguishing feature of this dialect, in view of its frequent prepalatal articulation, approximating [t^s]. This was first described by Lenz (1940: 150). Oroz (1966: 113) disagrees, noting instead a frequent fricative /č/, predominantly in the northern regions. Bernales (1986) confirms the fricative pronunciation of /č/ in southern Chile, while also observing a more forwardly-articulated affricate similar to the one described by Lenz. Wigdorsky (1983) and Diaz Campos (1986) also note the fricative pronunciation.
3. Regarding Cuba (p. 231): “The affricate /č/ only rarely deaffricates; Canfield (1981: 42) notes incipient deaffrication for Cuba, but it is much less frequent than in Puerto Rico, Panama or western Andalusia.”
4. Regarding the “Cañar and Azuay (including Cuenca)” areas of Ecuador (p. 249): “The affricate /č/ is sometimes realized as a fricative (Candau 1987).”
5. Regarding Panama (p. 299): “In Panama City, the affricate /č/ frequently receives a fricative pronunciation (Cedergren 1973). Reduction is more common in word-internal intervocalic position than word-initially. Women apply deaffrication more frequently than men, and the change is of recent origin, spreading among Panama City’s youngest residents. The fricative varaint is used most frequently among middle-class speakers, with frequency of deaffrication dropping off among upper-class speakers and among the lowest classes.”
6. Regarding the “Andean highlands” of Peru (p. 319): “The affricate /č/ often emerges as a fricative (Escobar 1978: 46).”
      But, in “Lima/central coast” (p. 321): “The affricate /č/ does not normally lose its occlusive element (Escobar 1978: 46).”
7. Regarding Puerto Rico (p. 332): “The affricate /č/ in Puerto Rican Spanish (especially in the eastern part of the island) was described by Navarro Tomás (1948) as adherente, meaning that the occlusive onset predominates over the fricative continuation. Quilis and Vaquero (1973) and Vaquero (1978) discovered fewer instances of this type of pronunciation, finding an increasing tendency for fricative realization of intervocalic /č/. López Morales (1983a: 147-156) determined that the fricative pronunciation is relatively recent in San Juan, is preferred among women, and is being reversed in the youngest generations. The fricative pronunciation is more frequent in the lower social classes, but only in the urban environment.

Additionally, in the Dominican Republic (p. 238) “the occlusive element [of /č/] generally remains”, and in Venezuela (p. 350) “/č/ rarely loses its occlusive element” (emphasis added in both cases).

Finally, Lipski comments on the [ž] and [š] realizations of /y/ (and /ʎ/) in Argentina and Uruguay:

• Regarding the “Buenos Aires/southern littoral” area of Argentina (p. 170): “The phoneme /ʎ/ does not exist, and /y/ receives a groove fricative pronunciation known as žeísmo or rehilamiento. Although the original sound was voiced [ž], most younger residents of Buenos Aires now pronounce a voiceless [š], and the devoicing is spreading throughout Argentina.”
• Regarding Uruguay (p. 340): “The phonemes /y/ and /ʎ/ have merged, giving a groove fricative pronunciation [ž]. Devoicing to [š] is not as all-encompassing in Montevideo as in the Argentine capital [Buenos Aires], but is rapidly gaining ground.”

Rather than quote similar passages from Hualde’s book, I instead have some brief sound clips from the accompanying CD. According to the read-me file included on the CD: “it should be noted that the Buenos Aires speaker [a relatively young-sounding female] consistently employs a voiceless fricative [that is, [š]] in words with orthographic y, ll, instead of the more conservative voiced variant [that is, [ž]] that appears in the Argentinian transcriptions in the book.”

ellos ‘they’	brilló ‘shone’	toalla ‘towel’	playa ‘beach’
yeso ‘plaster’	enyesa ‘he/she plasters’	enyesar ‘to plaster’	el yeso ‘the plaster’
deshielo ‘thaw’	desierto ‘desert’	hierbas ‘grasses’	contrast

The first two examples are selected from a text reading, so they sound cut off; the remaining items were elicited more or less individually. The first row shows a few different words with orthographic “ll” or “y” between vowels that is pronounced as [š]. The second row shows a word that begins with orthographic “y” that is pronounced as [š], followed by variants of the same root morpheme in different contexts. (These happen to be contexts — after a nasal or lateral — where Argentinian speakers who otherwise use voiced [ž] between vowels are reported to produce an affricate [ʤ] (a voiced postalveolar affricate, as in jump or germ in English). The last row contains a few words with a [y] sound (represented orthographically with hi- or i-), demonstrating that Argentinian speakers distinguish [y] from [š]/[ž] — in other words, it’s not that they “can’t pronounce a [y]-sound” or anything like that. The very last cell of the table is a longer sound file specifically showing this contrast; it’s a reading of the examples in Table 9.9 of Hualde’s book (p. 169):

yeso ‘plaster’	hielo ‘ice’
llena ‘full, fem.‘	hiena ‘hyena’
tramoya ‘artifice’, cebolla ‘onion’	paranoia ‘paranoia’
yerba ‘mate leaves’	hierba ‘grass’

That’s about all I’m willing/able to muster on this topic, at least for now. (I have done my best to answer comments on other aspects of the Language Log post in the <a href=”comments area of that post. Feel free to add more comments there or, preferably, here.)

---

Notes

1. Concerning “country-by-country classifications”, Lipski (pp. 4-5) writes:
   

   No serious observer of Latin America would propose that contemporary national boundaries should form the primary variable in determining dialect zones, but there is some value attached to organizing a purely descriptive presentation as a catalogue of national traits […]. The larger [primarily Spanish-speaking] nations in Latin America (e.g. Mexico, Colombia, Argentina and Chile) fully circumscribe entire dialect zones, and even small nations like Ecuador and Costa Rica envelop complete dialect capsules within their borders. Most dialect zones straddle national boundaries, and the only variables which show a close correlation with national boundaries are vocabulary items intimately related to the idiosyncracies of national culture, such as colloquial terms for national political parties or residents of a particular region. Despite the unlikelihood that models based solely on national identity will yield any theoretical insights, most descriptive studies have focused on single nations […], or on cities or regions within a single country.

2. The main point of the relevant comment is that “Spanish usage is not the determinant of [how to pronounce a name in English]” — something with which I heartily agree, as I’ve written about before here, here, and here. (Also of relevance are the comments here, especially the links provided in Bob Kennedy‘s comment.) Note that this commenter also does not provide an answer to what he claims to be “Inskeep’s intended question”, presumably because all the options are equally permissible (though perhaps not equally frequent/familiar) sequences of close-enough-to-English sounds.