Shemot: An exciting yet possibly nonexistent variant in the spelling of "amatah"

Summary: Gur Aryeh come up with a variant masorah in order to defend the tradition, and so perhaps confuses Minchas Shai. I investigate, and take a tour of Rav Saadia Gaon's translation, Dunash Ibn Labrat's attack, Ibn Ezra's defense as well as commentary, and Mizrachi and Gur Aryeh's suggestion. Then an analysis, as to why I think Gur Aryeh's suggestion is rather unlikely.

Post: When reading through Minchas Shai on the parsha, I saw a short yet extremely exciting note about a variant spelling of vatishlach et amatah. That is, the pasuk in Shemot reads:

ה  וַתֵּרֶד בַּת-פַּרְעֹה לִרְחֹץ עַל-הַיְאֹר, וְנַעֲרֹתֶיהָ הֹלְכֹת עַל-יַד הַיְאֹר; וַתֵּרֶא אֶת-הַתֵּבָה בְּתוֹךְ הַסּוּף, וַתִּשְׁלַח אֶת-אֲמָתָהּ וַתִּקָּחֶהָ.

5 And the daughter of Pharaoh came down to bathe in the river; and her maidens walked along by the river-side; and she saw the ark among the flags, and sent her handmaid to fetch it.

Amatah: In the first perek of Sota, Rabbi Yehuda and Rabbi Nechemia argue. "One said 'her hand' and one said 'her maidservant', etc." And so it is in Shemot Rabba. And according to the one who darshens it as the language of 'hand', the aleph has to be with a patach  and the mem has to have a dagesh, in accordance with the girsa of Rabbenu Saadia, and so {must have been} the girsa of the Targum. And according to the one who darshens it as a language of 'maidservant', the aleph is with a chataf-patach and the mem is without dagesh, just as it is in our own sefarim. And so wrote Rashi, Ibn Ezra, and Chizkuni {who discuss how the chataf-patach compels this interpretation}. And see Mizrachi and Shorashim.

I was familiar with the grammatical point, and have discussed it in the past. But this is the first I heard of an alternative girsa.
This would be extremely exciting, because it would resolve how there could be a dispute in something resolved by the nikkud. (The alternative is that the particular nikkud between the two quite similar pronunciations is later commentary, choosing sides in the existing Tannaitic machlokes; or that one is midrash and the other peshat, and midrash need not conform to nikkud, though this is something slightly difficult to say in a machlokes.)

But Rashi declared the nikkud dispositive in determining peshat, and made no mention of the existence of this variant girsa. He writes:

her maidservant: Heb. אֲמָתָהּ, her maidservant. Our Sages (Sotah 12b), however, interpreted it as an expression meaning a hand. [The joint from the elbow to the tip of the middle finger is known as אַמָּה, hence the cubit measure bearing the name, אַמָּה, which is the length of the arm from the elbow to the tip of the middle finger.] Following [the rules of] Hebrew grammar, however, it should have been vowelized אַמָּתָהּ, with a dagesh in the mem. They, however, interpreted אֶתאֲמָתָהּ to mean her hand, [that she stretched out her hand,] and her arm grew many cubits (אַמוֹת) [so that she could reach the basket]. [From Sotah 12b, Exod. Rabbah 1:23]

את אמתה: את שפחתה. ורבותינו דרשו לשון יד. אבל לפי דקדוק לשון הקודש היה לו להנקד אמתה מ"ם דגושה. והם דרשו את אמתה את ידה, שנשתרבבה אמתה אמות הרבה:

By contrasting the dikduk leshon hakodesh to the derasha, Rashi effectively states that on a peshat level, it must be precisely as he gave it, "her maidservant". He would not say this if he knew of an alternate girsa.

But according to Minchas Shai, the one with the alternative girsa is Rabbenu Saadia Gaon. I do not see that Saadia Gaon says this anywhere. He certainly does translate it as hand. From his Tafsir:

The Judeo-Arabic word ידהא refers to her hand. This is as much as I have seen in Rav Saadia Gaon's various writings. I have not seen him state that he is basing himself on an alternative girsa, or speak about the nikkud on these words.

Indeed, Saadia Gaon is roundly criticized for this translation by Dunash Ibn Labrat, in his response to Saadia Gaon about errors in the latter's translation. And this is the very first siman!

That is, roughly, he notes that Saadia translated וַתִּשְׁלַח אֶת-אֲמָתָהּ as that she stretched forth her ammah, arm. And he says that this is an error. Firstly, amatah has a chataf-patach and no dagesh in the mem. And if it were as Saadia Gaon translated, it should have had a patach dagesh; and he gives other examples. Rather, it is like in Bereishit 30:3, הִנֵּה אֲמָתִי בִלְהָה. And this works out with the beginning of the parasha that her maidservants came with her walking on the riverbank (Shemot 2:5), and these were אמהותיה; and when she saw the basket, she sent one of her servants to take it. And it is not possible to explain it without this. And you cannot say that אמתה is her actual אמה, because an amma is a measurement equal to that of an armspan, like a zeret or tefach. And we do not find in the entire Scriptures in this manner that someone sends forth a zeret or a tefach. Rather, one sends forth a hand. And he gives examples. And this is a destruction of the Hebrew language, forfend for us to make a breach in the explanation of a single one of the Living God's words to vanity, without judgment, analysis {dikduk} and weighing of every word, such that we are not punished. The opposite as those who are zealous, who are careful with their words, and they are found righteous.

This ends my rough summary of Dunash ben Lebrat's critique. Thus, the three points he makes are that the nikkud does not support it, that the context of maidservants accompanying her does support it, and that amma is a unit of measurement based on the arm, but we would not expect this lexical item to be used in place of the word hand or arm.

Note that in this attack, he does not even entertain the possibility that Saadia Gaon is operating on a different masoretic text, which has nikkud which supports the translation as "hand".

Ibn Ezra wrote a work, Sefat Yeter, defending Saadia Gaon against Dunash ben Labrat's various assaults. He writes:

The Gaon wrote that it is like amma arko. And Rabbenu Dunash wrote that he made three errors in grammar. And I, Avraham {Ibn Ezra} say: Who would give that I and Dunash Ibn Labrat would be in the same generation, such that I would reprove him on such matters. For any thing which a student repeats from his teacher and his teacher {in turn} from his teacher, it is only fitting ting to respond to the first teacher. And I know that Rav Saadia was the lightest of the talmidim from the Sages of the Talmud, and the greatest of them was lighter than the Sages of the Mishna. If so, how did the Gaon err? And why do you have contentions with him? Arise and contend with the mountains, the foundations of the world -- they are the Sages of the Mishna who explained it such. And furthermore, we find that the Gaon writes in many places in his sefarim, "And even though the logic of grammar is not in accordance with the words of the 'Rishonim', we rely upon them, and abandon our own opinion, for it is contemptible {or worth little} against their opinion." And so he did in the words ונקה לא ינקה; that only after the majority one leans; and many like this.

And while Ibn Ezra is quite familiar with Saadia Gaon's writings, he does not suggest that indeed the Gaon had a different girsa in the pasuk. Rather, his position is that who says that our limited understanding of dikduk, or our application of dikduk, is correct. And that he is relying on the chachmei HaMishna, who also seem to be against the dikduk. If there were a variant girsa, and had Saadia Gaon mentioned this, I would have expected Ibn Ezra to have mentioned it.

I would digress to mention that despite this defense of Saadia Gaon, in accordance with Saadia Gaon's own principles, in his own commentary Ibn Ezra sides with the other position in Sota, and with Rashi, and thus with the explanation put forth by Ben Labrat, declaring that the nikkud more or less forces an explanation of "maidservant".

Thus,

לרחוץ - מנהג המצריות היה.

והזכיר דבר נערותיה שהולכות על מקום היאור. כמו: ויד תהיה לך. בעבור ששלחה אמתה והיא אחת מהנערות לקחת התיבה. כי בת המלך לא תכנס אל מקום הסוף. כי במקום רחוק מן היבשה הושמה התיבה, שלא יגיע אליה כל עובר.

ועוד: כי אין מדרך הדקדוק להיות זרוע כי המ"ם רפה ומ"ם אמה ארכו דגוש.

ועוד: כי אמה ארכו מדה לא זרוע.

That is, he understands it to mean "maidservant" as we see existed in context. Furthermore it is not grammatical because of the nikkud. And finally, it is a measurement, rather than the arm itself. He thus echoes Dunash Ben Labrat entirely.

And once again, had he been aware of a variant girsa mentioned by Saadia Gaon, I do not believe he would not have so readily echoed Dunash Ben Labrat. Rather, just as elsewhere he refers to different nikkud meaning different things, he would do so here.

This indicates to me that Saadia Gaon likely never said anything to indicate another girsa. So where did Minchas Shai get this idea? I would guess from his relative contemporary, the Maharal of Prague. But maybe from Mizrachi. Indeed, in a likut of Rabbenu Saadia Gaon's perushim applied to the parsha, the idea that there is a variant girsa in this pasuk appears, but it is attributed to Maharal of Prague, in Gur Aryeh, rather to R' Saadia Gaon himself.

It seems Mizrachi suggested this:

He ends with "and this is according to the position of Rashi za"l. But others read it with a dagesh. And perhaps the one who darshened it as "her hand" read it so."

I haven't seen such a variant girsa, but perhaps he is aware of one, such that "others" read it so. Or else he means

Looking it up now in Gur Aryeh, inside, we see:

{Citing Rashi}"amatah -- the mem has a dagesh" {end citation -- meaning the mem would have to have a dagesh, but it does not.}.

But according to man de`amar who darshens it as "her hand", certainly he would read the mem as with a dagesh. For there is no thought, chas veshalom, that he did not know grammar! And so did Rav Saadia Gaon explain that that man de`amar would read the mem with a dagesh. And Rashi is explaining it only according to our reading, that since we read it אמתה without a dagesh, one cannot explain it as "her hand". And such is its explanation. But according to the rules of grammar, it is fitting that it would be read with a dagesh, and there is no dagesh according to our own reading.


Alternatively, even the one who darshens such maintains that it is written without a dagesh, in order to say that her hand changed and extended many cubits. For nouns with dagesh do not change because of semichut and kinui, and here, where it says amatah with a kinnui, and the vowel under the aleph changed to a sheva, where it should have been a patach. Rather, because her hand changed and extended many cubits, the word changed as well. And even though nouns with dagesh do not generally change {with elision of dagesh and changes of vowel} in kinnui, but rather stay just as they are, this one did change. And therefore the aleph is with a sheva and the mem is with a kametz. And thus it works out nicely. For Rashi should have written his commentary, citing a different Chazal, namely that which they explained that it was actually a hand {rather than the bit about extending many cubits}. Rather, he comes to answer the question as to why the mem has no dagesh.

Now that we have surveyed some of the relevant sources, here are my thoughts on the matter:

1. There might be a textual variant out there with amatah spelled with a dagesh. This would be rather nice, but I would suspect that it would have been influenced by the derasha and the purported grammatical "difficulty", rather than the other way around.
   
   
2. Perhaps Rav Saadia Gaon someplace does explicitly say that he is relying on an alternative girsa, as Gur Aryeh suggests. But I haven't seen it, and indeed think it likely that he did not -- those much closer to his time should have been aware of it, but they make other excuses for his translation.
   
   
3. I believe that both Mizrachi and Gur Aryeh are misguided in their approach here, which is the same as their approach when it came to "defending" Chazal's derasha on Arami Oved Avi on grammatical grounds. If something is midrash, then it need not conform to the rules of dikduk. Admittedly, this is more difficult here where a dispute, and where we want to say ain mikra yotzei midei peshuto, but this is still resolvable.
   
   
4. In terms of Gur Aryeh's idea that Rashi is reading the grammatical difficulty into the derasha, such that the pattern changes just as her hand changed, this strikes me as rather fanciful. Though it is fairly common in his commentary, IMHO, to see such fanciful reinterpretations of Rashi and of midrashim, then attributed to Rashi true intent.
   
   The reason for mentioning the full midrash rather than just that it means yadah mamash is that the gemara there in Sotah 12b indeed elaborates on why amatah was used rather than the more common choice of yadah. It thus means both hand and cubit. And Rashi wants to cite the midrash in full, with its full import and intent.
   
   And indeed, this extra also makes clear that this is midrash rather than an attempt at peshat. Because unlike Gur Aryeh and Mizrachi, Rashi does not feel that midrash must work out with dikduk. Indeed, that seems to be Rashi's very point here, that based on dikduk, the peshat must be X, but Chazal were still free to interpret Y, though operating on the level called derash.
   
   
5. At the end of the day, how can we understand the explanation of it meaning hand, when it is against the nikkud? Saadia Gaon was just following Chazal. But as for Chazal, I would note three things.
   
   First, who says that nikkud was dispositive for Chazal? Maybe they felt free to darshen against the nikkud, and claim that this is true meaning.
   
   Secondly, who says that they had fixed nikkud? It could be that there was a dispute as to the nikkud (and this would be along the lines of Gur Aryeh and Mizrachi). Or it could be that there was a dispute as to the meaning, and the established nikkud chose sides in that Tannaitic dispute. Consider the discussion Chazal have in Shir HaShirim, whether it should be כִּי-טוֹבִים דֹּדֶיךָ מִיָּיִן or ki tovim dodayich. And this is resolved not by appeal to tradition, but from sevara from other instances.
   
   Thirdly, who says that the fine points of nikkud were so established back then? It could be that the massoretes encoded what they heard, as an oral tradition, but they did not talk in earlier generations specifically about chataf-patach or dagesh geminating certain letters. And if so, amatah can be pronounced more or less identically. Sure, there are grammatical and phonological rules at play, such that perhaps, but perhaps not, one would pronounce the a sound longer or shorter or the mem slightly more stressed, much as the "u" sound in English has slightly different pronunciations in the word "cute" and "flute" (insertion of "y"). But they might not have formally noted that. If so, the word was amatah. And both Rabbi Yehuda and Rabbi Nechemia have no problem with the traditional pronunciation of the word. It is only when orthography was invented for this that different encodings for the two different derivations came into play. If so, then perhaps Rashi, Ibn Ezra, Dunash ben Labrat, et al. are all incorrect to dismiss an interpretation (from Chazal or elsewhere) for not being in line with the very fine points of nikkud, such as chataf-patach vs. patach, or dagesh vs. rafeh, or patach vs. kametz.