parnasa

Yesterday I was discussing the word parnasa פרנסה on the new Balashon Page on Facebook (please join!). I quoted my post on "pras and prize", where I wrote:

The Hebrew root פרס means "to split, divide, break", and a pras therefore originally meant "a half of something", "a portion". Klein points out that there are scholars who derive the word פרנס - -"to maintain, support" and parnasa פרנסה - "livelihood" as coming from פרס, with an added letter nun.

A reader wrote:

You think "parnasa" comes from "pras"? It seems rather doubtful. There are no other words in Hebrew derived in this manner. The other way around, maybe, if both words were borrowed from another language.

I have to admit that I can't think now of any other Hebrew words where the letter nun was added to make a four letter word (resh, however, often is added). But it is what Klein claims. Let's look at what he wrote on the entry for the verb פרנס:

Aramaic פרנס ( = he maintained, sustained, supported), Syriac פרנס (= he cared for, provided; he appointed, distributed,; he took the oversight, managed, ruled). Of uncertain origin. Some scholars derive base פרנס from base פרס.

He then offers the derivatives parnas פרנס - "supporter, manager" (from where Arabic gets firnas - "governor", maybe the origin of the name Abbas Ibn Firnas, for whom the lunar crater was named) and parnasa פרנסה - "livelihood".

It is interesting that Klein only provides one possible etymology. He certainly saw Ben Yehuda's entry, who mentions the "pras" theory, as suggested by Hoffman and Brockelmann, based on the Syriac cognate (although this book quotes Brockelmann as giving it a Greek origin). But then Ben Yehuda says that etymology doesn't seem likely to him. He then says it seems to him more likely that this four letter root derives from Greek or Latin, particularly due to samech being the last letter (as in such words as טכס tekes).

From Greek he suggests pronoos - "careful". As mentioned here, related words are offered by the Musaf HaAruch: πρόνηος -  proneos and πρόνοιος -  pronoios, and by the Aruch HaShalem:
πύρνος -  purnos (however the Aruch Hashalem rejects the Greek origin, and prefers the Syriac theory.) These words all appear to be related to the Greek pronoeo "to consider in advance, look out for beforehand, plan before", associated with the Greek words for "provide" and "providence". It is made up of two Greek words, "pro" (before) and "noeo" (think, perceive). From here we get the term pronoia - the system of land grants in the Byzantine Empire.

An alternate theory offered by Ben Yehuda is a derivation from the Greek phronis - thought, prudence, related to the Greek word for mind, phren, as found in "schizophrenia."

In the end, however, Ben Yehuda thinks the most likely origin is from the Latin pensum, "payment" (and related to the English word "pension"), with a resh added in later. This is the also the only etymology suggested by Even Shoshan.

I agree with this book that the Latin origin doesn't seem very likely. We've noted before that there was very little borrowing from Latin to Hebrew, and here we've seen cognates in Aramaic and Syriac as well. And even Jastrow, who usually tries to avoid giving Greek origins to Hebrew words, says parnas is an "enlargement of paran פרן" - "dowry". Klein says paran comes from the Greek phrene, related to pherein "to carry". Jastrow says that this only a "phonetic coincidence", and that paran is actually related to paras. But in the end, Jastrow likely knew the Syriac theory, or could have come up with it on his own, and still thought it was more likely that the root of parnas was somehow p-r-n.

So I'm not sure which root to accept, but it's all Greek to me...

tipul and tafel

In the post about pashtida, we mentioned how in the Talmud, the verb טפל meant "to paste, to plaster" (dough in the example we brought). But today that root is found in what appears to be unrelated words - for example tipul טיפול - "care, treatment" and tafel טפל - "of secondary importance" (as found, for example, in this halachic concept). How did the root end up with so many meanings?

While it doesn't appear often, the root טפל does appear in Biblical Hebrew: Tehilim 119:69, Iyov 13:4 and 14:17. The Even-Shoshan concordance says they all mean "to attach", but only in the metaphorical sense. Iyov 14:17 has וַתִּטְפֹּל, עַל-עֲו‍ֹנִי - "you would coat over my iniquity" (JPS), which seems to carry the meaning of "paste". Iyov 13:4 has טֹפְלֵי-שָׁקֶר, which the JPS translates as "invent lies", and other translations try to express the sense of "attach" or "paste" by writing "smear with lies" or "forgers of lies". However, the Daat Mikra writes that the pasting itself is not metaphorical, but the phrase is an allegory to builders who use fake plaster when building walls or paste plaster to cover defective walls or to doctors who spread phony medicines on their patients. In Tehilim the phrase טָפְלוּ עָלַי שֶׁקֶר זֵדִים is used metaphorically, and means "the arrogant have smeared me with lies".  This combination - טפל שקר - is found in the vidui (confession) prayer as tafalnu sheker טפלנו שקר. Most translations ignore the sense of attach or paste. Koren Sacks has "we have framed lies", and Artscroll uses "we have accused falsely." However, in their "Expanded Viduy Service" in the Yom Kippur Machzor, they point out that "literally the phrase means 'we have attached falsehood'."

But even with all that, we're still not closer to seeing the modern uses of the root. Klein writes that the original (Biblical) meaning was "to smear, paste plaster" and the Post-Biblical meaning was "to add, join, attach." From here he writes that tafel means "additional, subsidiary, of secondary importance." He also writes that the pi'el form of the verb progressed from "he plastered" to "he cared for, attended to" (maybe in the medical sense that we mentioned above.) From here we get tipul, metapel מטפל and metapelet מטפלת - "caregiver", particularly for children.

One confusing factor here is that in Aramaic, taflaya טפליא means "children", with tefel טפל meaning "child". However, Klein (and others) say that these words don't derive from our root טפל, but rather are related to the Hebrew taf טף, meaning "children". This leads to the strange scenario where metupal מטופל means both "having many children" (based on tefel) and the child being taken care of (related to tipul). According to Avineri (Yad Halashon p. 330) the former was the more popular understanding of the word, based its usage in that sense in Yiddish. He wrote that piece in 1960, but something likely changed, for I've never heard metupal with that meaning in conversational Hebrew.

Another related word is tapil טפיל - "parasite." Ben Yehudah writes that it was coined by the physician and linguist Aharon Meir Mazia on the basis of the Arabic tufayli - "unwanted guest". Even Shoshan and Klein connect the Arabic word back to the root טפל, but don't explain why (the parasite is attached to the host? the parasite is of secondary importance to the host?). But Arabic dictionaries such as this one connect it to the Arabic cognate of tefel (child), and this book, discussing pre-Islamic Arab culture writes that children were viewed as "nonproductive" and that:

the etymological connection in Arabic between tifl, child, and tufayli, parasite, party crasher, "sponge", is illuminating in this regard

And if all that wasn't confusing enough, there's another homophone out there. Klein writes that טפל is related to the root תפל also meaning "to smear, paste, plaster". However, there's an unrelated root, also spelled תפל, meaning "to be tasteless, be unseasoned" (this is the source of hatpala התפלה - "desalination"). Klein connects this second root with the Aramaic תפל ( = he talked silly things) and the Arabic tafala ( = he spat). A related Hebrew word is tifla תפלה - "folly" (this is entirely unrelated to tefila תפילה - "prayer".)

As you can perhaps imagine, תפלה as "folly" and טפלה as "of secondary importance" could be mixed up on their own, without even adding the fact that sometimes the tet and tav were switched in either direction. (For an example of the word switching meaning and spelling, see the play on words in Targum Yonatan, Sifrei and Rashi on Devarim 1:1). This page is meant to help Hebrew speakers tell the two roots apart, which isn't easy, particularly in the term emunot tefelot אמונות תפלות - "superstitions" (apparently coined by Krochmal).

Avineri wrote in Yad Halashon (page 216) that he suggested tefela טפלה as the Hebrew word for "by-product", but it was never adopted, probably due to the negative connotation of טפל (the word used today is totzar livay תוצר לואי). This probably also answers my question (in the pashtida post) as to why tefela was never adopted as the Hebrew term for pastry...

pashtida

In the discussion of shibolet, we mentioned that the Radak described how the Eprhaimites pronounced the word improperly. Here's part of his explanation as to why:

אולי היה אויר ארצם גורם להם זה כמו אנשי צרפת שאינן מבינים לקרא השי"ן וקוראין אותה כמו תי"ו רפה

Perhaps it was due to the climate of their land, just as the people [Jews?] of France, who can't properly say shin, and instead read it as a soft tav ("th")

Avshalom Kor, on his radio program, claims that this French speech issue could explain the Hebrew word pashtida פשטידה - "pie, casserole". He says that pashtida derives from pasta, the "-da" suffix is like the suffix of "limonada" (lemonade), what's left is pashti = pasta. The Jews used "shin" to spell "s" - like Shaloniki שלוניקי for Saloniki. He understands the Radak as saying that the soft tav was "s" - like Shabbos. So he says that in general the French mixed up "sh" and "s", and that explains Rashi (in France) writing "pasta" as "pashti(da)".

I'm certainly a big fan of Avshalom Kor (I wake up to his program almost every morning), so I'll give him the benefit of the doubt. But there are a number of things I don't really understand about his theory:

• Why does he assume that "da" is a separate suffix? According to this site,  "-ade", or the Latin "ata" means "product of an action" (lemonade from lemons). If pashtida means pasta, then what action is being described?
• Why does he assume that שלוניקי was pronounced Shaloniki instead of Saloniki (with a "sin")? That's how it's written in the Hebrew Wikipedia page.
• Is he saying that the switch from pasta to pashti happened because the Jews wrote "s" with a shin deliberately, or because the French Jews mixed up "s" and "sh"?
• The Radak says that in France they pronounced "sh" as "th" (or "s"). How does that explain a mispronunciation of "s" as "sh"?

As far as the etymology goes, there are certainly other opinions. Klein gives the following:

Probably from Middle Dutch pastede (=pie), from Late Latin pasta (whence also Italian, Spanish, etc. pasta), from Greek paste (=barleymash), which is related to pastos, paston (=sprinkled with salt), verbal adjective of passein (= to sprinkle), which is of uncertain etymology.

Other theories are that pashtida derives from the German (pastete) and Italian (pestette), from the Italian pasticcio (the source of pastitsio) or pastetta (according to Stahl) / pasteta (according to Even Shoshan) or even from the Polish word pashtet (or pasztet).

But I think to find the most reasonable etymology, we need to find the earliest source. Let's look at the Rashi that Avshalom Kor mentioned. The source is Pesachim 74b, which is discussing birds (pigeons, ducks) covered in dough (made of semida - solet ) and then roasted. Rashi (s.v. טפליה) says this is called pashtida פשטיד"א. Tosafot (s.v. טפלו ליה בר אווזא) disagrees with Rashi, saying the verb טפל means טיחה - "spreading", and therefore the gemara is describing something much lighter - they use a foreign word which Steinsaltz identifies as "crepes" (Stahl says perhaps related to the Yiddish kreplach). However, they are familiar with pashtida - they just claim pashtida is too thick to be the item described, and is baked in an oven, unlike the birds here, which were roasted on a skewer ("shipud"). So both Rashi and Tosafot see the pashtida to be fairly similar to its use today - a type of meat pastry, and in that light Steinsaltz says the word derives from the old French pastede, with the same meaning.¹

However, Rashi on Shmuel I 25:18 seems to give a different explanation to the word. The verse is describing the food that Avigail prepared for David, and mentions five "prepared" sheep - וְחָמֵשׁ צֹאן עֲשׂוּיוֹת. How were they prepared? Rashi writes:

תרגם יונתן תכברא ממולאות בשר דק ובצים פשטי"ץ בלע"ז

Targum Yonatan translates this as tachbera ["basket" - Jastrow]. They were filled with ground meat and eggs - pashtitz in Laaz [foreign language].

He then goes on to quote the gemara in Pesachim - one page earlier, 74a, where it describes the Pesach sacrifice being stuffed with meat (and according to some manuscripts use the word תכברא.) While the word pashtitz² is likely synonymous (or at least cognate) to pashtida, clearly this is not the pashtida described on the following page. Perhaps this why someone (who?) added the parenthetical comment to Rashi:

ע"ש היטב ברש"י ד"ה רבי ישמעאל קורהו וכו' ומצוה ליישב

Look closely at Rashi s.v.  רבי ישמעאל קורהו, there is an obligation to resolve it

Both this Rashi and the Rashi on Shmuel say that this explanation is by Rabbi Kalonymus ben Shabbethai, a rabbi who moved from Rome to Worms, where Rashi lived. (The Radak, who calls the stuffed meat פשטיצ"ו pastitsio, also mentions Rabbi Kalonymus as the source.)

It's possible that Rabbi Kalonymus is very significant to understanding the real story here. For he was responsible for introducing the Aruch dictionary (or parts of it) of Rabbi Nathan ben Yechiel of Rome to Rashi. And the Aruch, explaining the word mul'yeta מולייתא in Pesachim 74a offers the following definition:

כל דבר שממלא מבשר וצולהו היינו מולייתא כגון פשטידא
Anything filled with meat and roasted is mul'yeta, such as pashtida

Now while mul'yeta is referring to a lamb stuffed with meat, the Aruch's definition is wide enough to include the dough stuffed with meat found on the following page. However, the etymology of pashtida / pashtitz shouldn't have anything to do with pasta, dough, etc., but should somehow be associated with stuffed meat. Does anything fit that definition? I have a possible suggestion - the German pastete - ground meat, meaning (and cognate to) pate (which is still related to "pasta", but only via an earlier sense of "paste"). I admit that this is not fully researched, but it does seem important to disconnect pashtida from "pasta", at least regarding the Aruch's definition, and possibly Rashi's as well. Did Rabbi Kalonymus bring this meaning from Rome to Worms? If so, it would be nice to find an Italian word with the same meaning. In any case, clearly by the time of Tosafot, pashtida had found its place as "pie, casserole", since they use the word without needing to explain its meaning.

So today we find in the Israeli kitchen both pashtida and pasta פסטה, whether or not they are directly related. Another related food term is the Ladino pastel פסטל, which originally meant a meat pie ("Pastel de Carne"), but today usually refers to fried dumplings stuffed with potatoes, meat, etc.

One more word that has a similar origin is "pastiche", meaning "a medley of various ingredients; a hotchpotch, farrago, jumble." The word pastiche is the French version of the Greco-Roman dish pastitsio or pasticcio, which designated a kind of pie made of many different ingredients. All of the unanswered questions here make me feel like this post is a bit of a jumble. If you have any answers - please let me know!

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

1. In Modern Hebrew, pashtida does not only refer to meat pies, but perhaps even more likely to dairy ones. However, in Rabbinic literature, as Raffi Sirkis points out in his appendix to Pashtidot Olamiot, the term for a dairy pie was fladen or fladon פלאדן / פלאדון, cognate to the English "flan".

Also interesting is that pashtida is a legitimate Hebrew word today (I would have guessed that the Hebrew version would be tefelah טפילה). Sirkis points out that Ben-Yehuda did not include it in his dictionary, but the Vaad HaLashon included pashtida פשטידא in their 1912 dictionary of cooking terms as the Hebrew translation of the Yiddish "kugel". By 1938, they were spelling it with a heh at the end: פשטידה, and became the official translation for the English "pie" and "pastry".

2.  This spelling can also be found in Tosafot Beitza 16b, s.v. קמ"ל קמחא עיקר, and Sefer Kolbo 145 s.v. דין.

Update: Shortly after publishing this post, I saw the Mikraot Gedolot HaKeter edition of Shmuel. I was very surprised to see that they had a rather different version of Rashi and Radak on our verse, with Rashi using the foreign word “פרשידש” and Radak providing “פרשירש”. These were both sufficiently different from any form of the word pashtida, that I thought it was important to research the point further.

From correspondence with Prof. Jordan Penkower and Daniel Fano, I discovered something very interesting. As Penkower mentions in his article, “גלגולי נוסח תרגום יונתן ופירוש רד"ק ליחזקאל כג, כ; לד, יח”, pgs 254-5 (in שנתון לחקר המקרא והמזרח הקדום יג), the Haketer version of Rashi (he has Radak using the same word) seems to be the accurate one. Other older manuscripts have similar spellings. What is the origin of this word? According to Fano, it is from the Old French farcid (farcides in plural). It means “to stuff” and is related to the English word “farce”. 
According to Penkower, the Venice edition of Mikraot Gedolot (which I had used for my post) had no justification to use a variation of the word pashtida. He doesn’t say why they did – but my guess is that Rashi having used pashtida in Pesachim, and with pashtida and farcides both referring to stuffed food, they simply became confused.

So all of this, together with the important comment on my post by reader Mike Gerver that in the Arukh “mul'yeta is defined as any kind of food stuffed with meat, and gives pashtida as an example”, leads me to believe that perhaps Avshalom Kor’s position has more merit than I assumed. Pashtida may really refer to specifically a kind of pie or pastry, and not just anything stuffed with meat.

shibboleth

As I've mentioned before, William Safire was the first language writer I ever read, and a major inspiration for Balashon. Since his death earlier this year, I've been trying to think of a post that would be a worthy tribute to him. Hopefully, this one will fit the bill.

In his 1982 book, What's the Good Word?, Safire discusses "shibboleths" (page 246), which he defines as "those passwords that signal who is a native and who an outlander." Most of the entry contains examples of various American shibboleths, but he does provide the following background:

"Shibboleth," the Hebrew word for "stream," was used by the soldiers of Gilead to separate their neighbors from the Ephraimites, who pronounced the sh as an s; because they couldn't get it right, according to Judges 12:6, 42,000 Ephraimites were slain. In those days, pronunciation meant something.

However, this etymology was challenged in these two letters he received:

The Concise Oxford Dictionary states that "shibboleth" derives from the Hebrew word for "an ear of corn," not "stream," as you report. Perhaps you base your information on rabbinical authority that would eclipse Oxford.

~~~~~~~~~~

You state that "shibboleth" is the Hebrew word for "stream." That is not quite correct. It means the current of a stream, not the stream itself. It is used with this meaning in verse 2 of Psalm 69. The word, which is pronounced "shibolet" in modern Hebrew, also means "an ear of corn or a spike." It is also the Hebrew name for the star Spica.

He then prefaces the next letter, from David B. Guralnik (Dictionary Division, Simon & Schuster) with the note, "The current etymology is in the following letter from an eminent lexicographer":

The Hebrew word שבלת, from an Ugaritic root, has several meanings: flowing stream (as in Isaiah 27:12 or Psalms 69:2), ear of corn (as in Ruth 2:2 or Isaiah 17:5), or twigs or branches (as in Zechariah 4:12). The first meaning makes most sense in the Judges narrative precisely because the incident took place at a stream. Seems like a logical password.

So who's right?

First of all, let's look at the verses themselves (Shoftim 12:5-6):

וַיִּלְכֹּד גִּלְעָד אֶת-מַעְבְּרוֹת הַיַּרְדֵּן, לְאֶפְרָיִם; וְהָיָה כִּי יֹאמְרוּ פְּלִיטֵי אֶפְרַיִם, אֶעֱבֹרָה, וַיֹּאמְרוּ לוֹ אַנְשֵׁי-גִלְעָד הַאֶפְרָתִי אַתָּה, וַיֹּאמֶר לֹא.  ו וַיֹּאמְרוּ לוֹ אֱמָר-נָא שִׁבֹּלֶת וַיֹּאמֶר סִבֹּלֶת, וְלֹא יָכִין לְדַבֵּר כֵּן, וַיֹּאחֲזוּ אוֹתוֹ, וַיִּשְׁחָטוּהוּ אֶל-מַעְבְּרוֹת הַיַּרְדֵּן; וַיִּפֹּל בָּעֵת הַהִיא, מֵאֶפְרַיִם, אַרְבָּעִים וּשְׁנַיִם, אָלֶף

The Gileadites held the fords of the Jordan against the Ephraimites. And when any fugitive from Ephraim said, "Let me cross," the men of Gilead would ask him, "Are you an Ephraimite?"; if he said, "No," they would say to him, "Then say shibolet"; but he would say "sibolet," not being able to pronounce it correctly. Thereupon they would seize him and slay him by the fords of the Jordan. Forty-two thousand Ephraimites fell at that time.

(An interesting discussion of how the words shibolet and sibolet were transliterated in various non-Hebrew Bibles can be found in this Philologos column and in this book.)

All of the classical Jewish commentaries identify the shibolet in the story as the "current of a stream", although I was surprised to see relatively little discussion of the linguistic nature of this episode (other than by the Radak, who we'll come back to later.) However, a number of scholarly articles in the past 70 years have done much to clarify the issue further.

One of the first was E. A. Speiser, in his 1942 article, "The Shibboleth Incident". Speiser doesn't put much significance in the meaning of the word, but does explain:

The meaning of the word is of minor importance. Elsewhere in the Old Testament it has the sense of "ear of corn" (Gen 41:5 ff.; Ruth 2:2; Zach 4:12) or, less commonly, "flood, torrent" (Psalms 69:3,16; Isa. 27:12). In out passage it is taken in the former sense by such versions as the Greek Codex Vaticanus and Aquila, as well as some modern scholars (Cf. Liebmann ZAW 25 (1905) 161). On the other hand, reference to flowing water is assumed by the medieval Hebrew commentators and a majority of the moderns, evidently because such an allusion would be more appropriate to the occasion.

Instead, he prefers to focus on the phonetic aspect of the issue. While we've been translating the words as "shibolet" and "sibolet" - are either of those accurate representations of how the word was pronounced in Gilad and Efraim? And why couldn't Efraim pronounce it "properly"? Who had the non-standard pronunciation? How were the various pronunciations represented with the limited letters available (and no dots available to indicate the difference between shin and sin)?A full discussion of these questions is beyond the scope of this post, but it turns out that the meaning of the word remains significant, even for Speiser.

He mentions a theory, foreshadowed by the Radak on Shoftim 12:6, and later developed by Joseph Marquart and Zellig Harris. According to them, the two meanings of shibolet have two separate etymologies: "ear of corn" comes from the root sh-b-l  שבל, whereas the meaning "flood, torrent" comes from a distinct root:  th-b-l. Therefore, the Ephraimites pronounced the word as "thibolet", which gave them away. Beeston (1979) thinks that the original root was sbl סבל (as pronounced by the Ephraimites, but agrees the two meanings of shibolet have two different origins:

If we assume that "ear of corn" was indeed *sblt in Proto-Semitic, then the Ephraimite, Arabic and MSA forms are adequately accounted for, and the problem assumes the shape, "Why did non-Ephraimite Hebrew show the shift to shblt?" The latter is a homonym of the word for "watercourse", but it is unlikely that the two words have a common origin. "Watercourse" is closely linked with the very common shebil "road, way", and both should probably be assigned to a PS root *shbl; but "ear of corn" is isolated in the lexicon, with no associated terms. This fact, together with the very close phonetic similarity between it and "watercourse" may have led to a confusion of the two words and the creation of a homonym.

He goes on to say that shevil שביל and shovel שובל (in Yishayahu 47:2, which he translates as "well, fountain") have an "underlying feeling of semantic association".

While Speiser agrees that the anomaly began on the eastern side of the Jordan, with the Gileadites, he does not accept either theory regarding the separate etymology, and instead writes that both of meanings of the word shibolet, along with its various Semitic cognates, derive from one root:

In reality, however, there is no reason for deriving the established homonyms for "ear of corn" and "flood" from two distinct roots. Arabic sbl may underline both "hang down" (whence we get sunbulat-, sabalat- "ear of corn") and "rain, flow." The two ranges are thus easily linked semantically, which accords fully with their apparent etymological identity.

(Yet another view can be found in Rendsburg, 1986, who agrees that the two meanings of shibolet come from distinct roots, but that the Gileadites, influenced by their neighbors the Ammonites, pronounced the meaning "torrent, stream" as thibbolet, which the fleeing Ephraimites rendered as "sibbolet".)

Klein agrees with Speiser, and connects the various words to one root. For example in the entry for shibolet ("ear of corn") he writes:

Formed from שבלת shubolet through dissimilation of the vowels u-o to i-o, related to Aramaic שבלתא, Syriac שבלתא, Arabic sunbulah, sabalah, Ethiopian sabl, Akkadian shubultu (= ear of corn). These words derive from base שבל (= to hang down, draw along, move along)

And for the meaning "flowing stream, current of a river", he adds:

Derived from שבל (= to move along), hence of the same origin as shibolet ["ear of corn"]

As far as shovel, he gives the more common translation as "flowing skirt, train of a robe", and says that it too derives from שבל, literally meaning "that which hangs down". Shevil - "path, way" - has the same origin, and literally means "that which runs along". (From here we get beshvil בשביל - "for", and literally means "in the path of".)

In the end, I find aspects of all the arguments convincing. Luckily for me, I don't need to decide who's right and wrong. But I am curious - shouldn't the meaning of the Biblical text be more obvious? Shouldn't the reader simply know what is meant without all of these debates and discussions? Perhaps it was left deliberately ambiguous, either to encourage study (and future Safire columns), or to allow the word play found with homonyms. That certainly seems to be the case in Yishayahu 27:12, as Hendel points out. The verse reads:

וְהָיָה בַּיּוֹם הַהוּא, יַחְבֹּט ה' מִשִּׁבֹּלֶת הַנָּהָר עַד-נַחַל מִצְרָיִם; וְאַתֶּם תְּלֻקְּטוּ לְאַחַד אֶחָד, בְּנֵי יִשְׂרָאֵל.

And in that day, the Lord will beat out [the peoples like grain] from the flowing stream (shibolet) of the River [Euphrates] to the Wadi of Egypt; and you shall be picked up one by one, O Children of Israel!

Here we have a wonderful play on words. On the one hand, we have the beating of the shibolet of grain (see Ruth 2:2,17) and on the other the shibolet of a river. Maybe understanding this is a sign that when it comes to Hebrew, we're natives, and not outlanders...