Shibboleth and perejil

« previous post | next post »

A recent NYT editorial described the immigration/citizenship/deportation crisis on the divided island of Hispaniola ("Stateless in the Dominican Republic", 7/11/2015):

In 2013, the Dominican Republic’s highest court issued an unconscionable ruling that rendered tens of thousands of Dominican-born people of Haitian descent effectively stateless. Last year, the Dominican government, responding to international criticism, established a process that ostensibly offered them a path to be recognized as citizens. But because the application process was so onerous and poorly administered, tens of thousands of people remain in limbo, shunned in their homeland and unwelcome in neighboring Haiti.

Almost 80 years ago, the Dominicans were more overtly brutal — Abby Phillip, "The bloody origins of the Dominican Republic’s ethnic ‘cleansing’ of Haitians", WaPo 6/17/2015:

[T]he 1937 Parsley Massacre is widely regarded as a turning point in Haitian-Dominican relations. The slaughter, carried out by Dominican dictator Rafael Trujillo, targeted Haitians along with Dominicans who looked dark enough to be Haitian — or whose inability to roll the "r" in perejil, the Spanish word for parsley, gave them away. 

The "pronounce parsley and die" story is modern example of a Shibboleth, named after the Hebrew word shibbólet (שִׁבֹּלֶת) which was once used in a similar way, according to Judges 12:

4   Jephthah then called together the men of Gilead and fought against Ephraim. The Gileadites struck them down because the Ephraimites had said, "You Gileadites are renegades from Ephraim and Manasseh."
5    The Gileadites captured the fords of the Jordan leading to Ephraim, and whenever a survivor of Ephraim said, "Let me cross over," the men of Gilead asked him, "Are you an Ephraimite?" If he replied, "No,"
6   they said, "All right, say `Shibboleth.'" If he said, "Sibboleth," because he could not pronounce the word correctly, they seized him and killed him at the fords of the Jordan. Forty-two thousand Ephraimites were killed at that time.

Here's what I wrote about the 1937 massacre in the lecture notes for LING001, a few years ago:

In early October of 1937, between 15,000 and 35,000 Haitians were massacred by order of Dominican dictator Rafael Trujillo. Some Haitians had become "foreign" when the border between the two countries was established in 1929, while others had crossed into the Dominican Republic a decade earlier to work on the sugar plantations. With the low commodity prices of the depression, they had become unwelcome guests — and the unlucky objects of a century and a half of antihaitianismo.

According to Michele Wucker's book Why the Cocks Fight: Dominicans, Haitians and the Struggle for Hispaniola (New York: Hill & Wang, 1999. Copyright Michele Wucker, used by permission):

Trujillo's men searched the houses and estates of the region one by one, rounding up Haitians and initiating deportation proceedings against them; once the paperwork was done, the Dominican government had `proof' that the Haitians had been sent back to Haiti. The Haitians then were transported like cattle to isolated killing grounds. The soldiers slaughtered them at night, then carried the corpses to the Atlantic Ocean, at the customs port in Montecristi, and threw the bodies to the sharks. For days, the waves carried uneaten body parts until they washed up on the beaches.

Often, the soldiers did not even bother with the charade of covering up their crimes. Entire families were mutilated in their homes. For Haitians away from their homes, in the streets or in the fields, the soldiers applied a simple test. Since Haitians are considerably darker than most Dominicans, soldiers would accost a man or woman with dark skin. Holding up sprigs of parsley, Trujillo's men queried their prospective victims: "Como se llama ésta?" "What is this thing called?" The terrified victim's fate lay in his pronunciation of the answer. For Haitians, whose Kreyol language uses a flat "R", it is difficult to pronounce the trilled "R" in the Spanish word for parsley, "perejil." If the word came out as the Haitian "pe'sil," or a bastardized Spanish "pewehi", the victim was condemned to die.

A fuller account, kindly provided by the author in the form of excerpts from the chapter in her book dealing with the massacre, is available here.

The massacre forms the background of Edwidge Danticat's novel The Farming of Bones.

Wucker's phonetic description of the pronunciation difference seems to be wrong, since Dominicans pronounce the intervocalic single 'r' in Perehil as a flap rather than a trill. But the Haitian Kreyol /r/ is pronounced as a voiced velar fricative [γ] or a velar approximant [ɰ], so the story remains plausible even if the description is not quite right.

The apparent mistake about r-rolling is widespread — thus , Mark Memmott, "Remembering To Never Forget: Dominican Republic's 'Parsley Massacre'", the two-way (NPR), 10/1/2012:

The method his soldiers used in 1937 to try to identify those who would be killed was cruelly unique. When confronting someone in the lands along the border with Haiti, they would hold up a sprig of parsley and ask what it was. If the person responded by trilling the "r" in perejil (Spanish for parsley), he would be free to go. Anyone who didn't trill the "r" was thought to be a Haitian Creole speaker — and was likely to be killed.

The NPR episode has some actual examples:

Here's Edwidge Dandicat's Haitian version:

I have to say, that one sounds rather like a weak dental fricative to me, more like [pe.ðe.hil] than [pe.ɰe.hil] — and not very different from a tap. And it looks that way in a spectrogram as well —

You can see the F2 locus around 1800 Hz, as expected, and the lack of F2-F3 convergence that would be expected for velar place.

And here's Julia Alvarez's Dominican version:

Her /r/ is, as expected, clearly NOT a trill but rather a tap:

[And maybe the biggest differences between the two pronunciations are Dandicat's voiceless /h/ vs. Alvarez's voiced /h/, and Dandicat's aspirated initial /p/ vs. Alvarez's unaspirated initial /p/ — though I don't know whether those features are stable indicators of ethnicity on Hispaniola.]

This minor phonetic confusion (between rolling and tapping) is unfortunate, because it has led some people (in conversation with me) to doubt the whole story.


  1. Chips mackinolty said,

    July 13, 2015 @ 9:50 am

    I hadn't realised the original use of the word Shibboleth in the bible. It has another place in history. A revolt against the Angevin French in Sicily began Easter Monday (30 March 1282) in Palermo after a French soldier insulted a Sicilian woman. As many as 8,000 French were massacred: the test to determine whether a suspect was French or Sicilian was the way in which they pronounced "Cicero". The non-Sicilian/French pronunciation was an abrupt and deadly giveaway. The incident began the successful revolt against the Angevin known as the war of the Sicilian Vespers.

  2. Mean Something said,

    July 13, 2015 @ 12:42 pm

    Rita Dove's poem "Parsley" from her 1983 collection, MUSEUM, addresses this story as well. It's the first I heard of the massacre. It ends: "He will/order many, this time, to be killed//for a single, beautiful word."

  3. Coby Lubliner said,

    July 13, 2015 @ 12:44 pm

    It's interesting (to me) that the word 'shibboleth' is usually stressed on the first syllable in English (it's paroxytone in Hebrew). It seems to me that once upon a time the default stress on exotic words was proparoxytone (except that Latin words, however mangled in pronunciation, kept their original stress). This has changed in recent (?) times, especially with Italian or Italian-seeming words. I remember finding it strange that in the 1996 film Big Night the supposedly Italian cooks (one of them played by the Italian-American Stanly Tucci, who co-wrote and co-directed it) pronounced the Italian word timpano as timPAno (didn't it occur to them that it's the singular of timpani?), but by the time the supposedly Italian characters in The Borgias (2011) referred to Pesaro as PeSAro I was no longer surprised, any more than I am by the recent movie Ex Machina being called Ex MaSHEEna, or by the fact that I am one of the few who pronounce patina as PATina. But then I also say shibBOleth.

  4. Coby Lubliner said,

    July 13, 2015 @ 12:54 pm

    As regards the two recordings, I find more of a difference in the vowels than in the consonants. Edwige Danticat's e's sound like French é and the last syllable like the French word île.

  5. Narmitaj said,

    July 13, 2015 @ 1:28 pm

    @ Cory Lubliner – re Ex_Machina; here in the UK I think the film has generally been pronounced "Ex Makkinna", as in the BBC Radio 4 Film Programme episode featuring it (first at 1:18):

    I also pronounce it PATina myself. Not sure about shibboleth (though probably the first syllable)… it's one of those words I read but rarely generate; like pusillanimous, I keep forgetting exactly what it means (pusillanimous I persist in thinking should mean aggressive, like pugilistic, when it means about the opposite; shibboleth I think of as something like "false god").

  6. rubrick said,

    July 13, 2015 @ 4:15 pm

    Someday our robot overlords will speak in similar dark tones of our onetime tool of repression, the CAPTCHA….

  7. Jon said,

    July 13, 2015 @ 5:25 pm

    Years ago a Dutchman told me that during the second world war occupation of the Netherlands, there were German spies who tried to pass themselves off as Dutch. The crucial test was to ask them to name the Dutch town of Scheveningen: no German could pronounce it correctly. He didn't say what they did to those who failed the test.

  8. David Morris said,

    July 13, 2015 @ 7:40 pm

    I have a large passive vocabulary of words I've read and know (or mostly know) the meaning of, but have never heard and never spoken. Understanding the meaning in written context doesn't require guessing at the pronunciation, so I just skim over it. Just occasionally I hear a word for the first time I can remember and think 'Oh, is that how it's pronounced?'.

  9. Ken said,

    July 13, 2015 @ 9:51 pm

    @David Morris, I have the same problem. I also mentally "mispronounce" some words when reading, due to having learned them from reading before hearing them spoken. Most of these are technical or foreign words, but there is an irritating simple one – I read "sew" to rhyme with "few" every time! I then feel myself (or actually something a little lower down in my mind than my self, if that makes sense) backing up and correcting it.

  10. Bart said,

    July 14, 2015 @ 3:39 am

    I too have heard the story about the Scheveningen test. It may have happened once, I suppose, that somebody noticed that Scheveningen was wrongly pronounced, but I can't think that it was literally a standard test similar to shibboleth in the the Bible.
    It's not difficult for a German or English speaker to pronounce the 'sch' – just an [s] followed by a [x]. There can't have been many German spies who spoke near-perfect Dutch in all respects except that they hadn't mastered the [sx] sound.

  11. Bart said,

    July 14, 2015 @ 4:02 am

    A much more demanding shibboleth word for genuine Dutchness would test vowel sounds that are not present in German or English: IJmuiden, for example.

  12. Peter Erwin said,

    July 14, 2015 @ 5:38 am

    Interesting, this article [PDF] notes that prior to the actual massacre, local Dominican officials and soldiers were in the habit of using pronunciation to distinguish immigrant Haitians from native Dominicans for purposes such as identifying who had to pay "migration taxes". Perejil is mentioned, as is tijera ("scissors") and claro ("clear; sure"). There's also an argument to the effect that during the massacre, this was done by soldiers more as a kind of mockery than a serious test, though this seems to be based on a single person's testimony, so it's hard to know how seriously to take this.

    Ercilia Guerrier, who lived in Restauración, recalled being stopped prior to the massacre by Dominican soldiers checking to see if immigrants had paid their tax: “You were going to the market or to Loma de Cabrera, you run into the guards, they say to you, ‘Stop right there!’ And so you do that, you stop. ‘Say “perejil!’ And so you say, ‘Perejil, perejil, perejil!’ ‘Say claro, ‘¡Claro, claro, claro!’.” Asked if it was ever necessary to produce a birth certificate or baptismal record to avoid the migration tax, Guerrier replied, “No, no. As soon as you could say that [“perejil” or “claro”], you didn’t have any problems with them.”

  13. AA Bender said,

    July 14, 2015 @ 6:51 am

    Michelle Wucker's description of the Dominican Shibboleth question, "Como se pronuncia ésta?" seems oddly ungrammatical.

    Typically Spanish speakers would have said, "Como se pronuncia ésto?" ["How would one pronounce this thing?"]. That is the speaker would have used the masculine form of the demonstrative. A Spanish speaker certainly would have used the masculine form since "perejil" is masculine, and the exercise was in pronunciation and not in identification of the object in question.

    {Alternatively, the speaker may have meant, "What [kind of plant] is this?" as plant [planta] is feminine. But that is an odd construction.)

    Another anomaly of the story is that the rolled "r" sound in Spanish is generally reserved for double "r"s like "burro" and the second syllable of "ferrocaril" and not for single "r"s as in "perejil".

    [(myl) Yes, as it says in the post, "Wucker's phonetic description of the pronunciation difference seems to be wrong, since Dominicans pronounce the intervocalic single 'r' in Perehil as a flap rather than a trill", and "The apparent mistake about r-rolling is widespread ". Not to speak of the sound clip and spectrogram.]

  14. jaap said,

    July 14, 2015 @ 9:32 am

    The Dutch "Scheveningen" shibboleth story has been popularised by the film "Soldaat van Oranje" (Soldier of Orange), as can be seen in this clip with Rutger Hauer:

    It is based on an autobiographical book, so could very well have really happened, though it probably was not a widely used test.

  15. Adrian Bailey said,

    July 14, 2015 @ 9:34 am

    Bart said: "It's not difficult for a German or English speaker to pronounce the 'sch' – just an [s] followed by a [x]."

    But it is difficult, unless you've practised it. Years ago I was staying in Holland chez my friend's grandparents, and since we couldn't speak each other's language we had some fun practising words; Scheveningen was one of the ones they got me to try. Not only is [sx] quite tricky, but to my ear, which is unused to distinguishing this phoneme, it seemed to have a particular character that couldn't be simply summed up by the IPA symbols.

  16. Guy said,

    July 14, 2015 @ 11:19 am

    AA Bender,

    But "palabra" is feminine. And they are, after all, asking them to pronounce a word, as it doesn't make any literal sense to pronounce a plant. So it's not immediately obvious to me which gender would be natural. I'll have to take an informal survey of native Spanish speakers for their impression when I get the chance.

  17. Bastian said,

    July 14, 2015 @ 11:51 am

    Actually 'ésto' is one of the des cases where Spanisch has maintained thr neuter Fender (actually, in all three series of demonstratives) and is the one you would use here. The male Form is 'éste'. When there is no antecedent in discourse (like hierba 'spice') you would use the neuter, unless you are talking about a person (a bit simplified but that's the basic rule).

  18. Bastian said,

    July 14, 2015 @ 11:52 am

    sorry for the typos: "few cases", " neuter gender"

  19. AA Bender said,

    July 14, 2015 @ 4:02 pm

    You are correct; that is the basic rule.

    I was trying gently to point out that the use of "ésta" in the passage above grates on the Spanish ear and rings untrue. The rolling of a single "r" also rings a bit off to my ear, but it is slightly more believable, I suppose. Whether that informs us one way or the other about the accuracy of Wucker's assertions, I can't say.

    [(myl) I'm trying gently to suggest that you read the whole post, not just the first couple of paragraphs. As explained at length, with audio clips and spectrograms, there is no rolling of /r/ in Dominican (or as far as I know other Spanish) pronunciations of "perejil" — /r/ in such word-medial cases is a tap, not a trill. The use of phrases like "trilled 'R'" by Wucker and various journalists is a mistake.]

  20. Y said,

    July 14, 2015 @ 7:27 pm

    I don't know that the /r/ of Danticat, an educated late 20th century Haitian, is a good proxy for that of Dominican Haitians in the 1930s, many of which, I presume, lived in laborer camps and had never had much exposure to Spanish.

    The description of Spanish tapped /r/ as a 'roll' or a 'trill' is indeed sloppy. I wish there was a technical term, however, that would encompass coronal taps, flaps, and trills, all of which rely on momentary stopping of the airstream.

  21. AA Bender said,

    July 14, 2015 @ 9:11 pm

    Yup. I did read (and listen to) the entire post. The audio is pretty conclusive.

    Yours exactly was the point that I, ever, ever more gently, even more so gently, was trying to emphasize.

    Great post.

  22. Guy said,

    July 15, 2015 @ 4:46 am

    Another Shibboleth story dates back to around 1300 in Flanders. The towns of Bruges and Ghent rebelled against the French domination. It was the dawn of the guilds, associations of artisans and merchants (butchers, bakers, weavers,…) in the towns that wanted to get rid of the restrictions and taxes imposed by the feudal system. The French weren't able to pronounce certain consonants in Flemish, so when they were pressed to pronounce the rallying cry of the townspeople 'Schild en Vriend', they said something like 'Skild en Frind'. In Bruges, on the morning of May 18, 1302, the massacre being named after the early morning prayer, 'Brugse Metten', many French speakers had their throats cut because they couldn't pronounce 'Schild en Vriend'.
    It is less clear if the 'test' was also used some months later, when the 'armies' of the Ghent and Bruges guilds defeated the cavalerie of the French king at the Groeninge kouter (a marshy field near Kortrijk) on July 11, 1302 – the Battle of the Golden Spurs (the Flemish national day today). The name refers to the spurs lost by the many casualties among French noblemen who got stuck in the mud of the Kouter and were easy prey for the fierce guilds' men.
    If the people from Kortrijk had to pass the test also, they would be in dire straits, because they can't pronounce the 'sch' sound either.

  23. Bernhard said,

    July 15, 2015 @ 3:30 pm

    Interesting (and sad); thanks!

    I wonder: Doesn't Julia Alvarez say: ‘we [Dominicans] trill the /r/ although Edwige's is pretty good' (0:36–0:37 of the NPR clip)? I.e., is it not suggested that this phrase (which is the one analysed above) is really not representative for Haitian pronunciation by Alvarez's standards, and thus the phonetic analysis seems to supports Alvarez's assessment (except for the trilling part, as you discuss)? Edwige Dandicat's parsley (one false start, both occurrences relatively low, but /r/ sounds a bit like an approximant??) around 0:44 is supposed to be more representative, if I understand correctly.

RSS feed for comments on this post