ואמר רבי אלעזר חנה הטיחה דברים כלפי מעלה שנאמר ותתפלל על ה' מלמד שהטיחה דברים כלפי מעלה
Or, in English:
R. Eleazar also said: Hannah spoke insolently31 toward heaven, as it says, And Hannah prayed unto32 the Lord.33 This teaches that she spoke insolently toward heaven.The idea is that al Hashem rather than the expected el Hashem in the pasuk implies that the spoke insolently upon the Lord.
Rashi rejects a girsa he saw in the gemara before him:
ותתפלל על ה' - כך כתוב הוא במקרא ולא גרסינן הכא אל תקרי:"And she prayed al Hashem" -- such is what is written in the verse, and so we are not gores here al tikrei [read not X but Y].
It seems that there was some girsa which cited the pasuk as אל, el, with an aleph, and then suggested rereading it as על. Yet our masoret text has על, with no need for a rereading.
To support Rashi, I'll not that there is an al tikrei just a bit later in the gemara, on 32a, with an explanation of just why we can swap the ayin and aleph:
ואמר רבי אלעזר משה הטיח דברים כלפי מעלה שנאמר(במדבר יא, ב) ויתפלל משה אל ה' אל תקרי אל ה' אלא על ה' שכן דבי רבי אליעזר בן יעקב קורין לאלפין עיינין ולעיינין אלפיןThat this whole explanation of swapping aleph and ayin is not made here (as we shall see) but only later is evidence that this al tikrei here is artificial and the result of scribal error, as a duplication of the al tikrei which appears in a short while, or as a corruption of a diyuk of al as opposed to el.
We can actually find a gemara that has what Rashi saw before him. Thus, in this manuscript:
Firenze, Biblioteca Nazionale Centrale , II.1.7
ברכות לא ע"ב - ברכות לא ע"ב
That is, al tikrei el ela al. However, if we look at the Munich manuscript, we see (two lines down in the image, in the middle of the image):
ואמר רבי אלעזר חנה הטיחה דברים כלפי מעלה שנאמר ותתפלל על ה' אל ה' לא כתיב מלמד שהטיחה דברים כלפי מעלהThe words אל ה' לא כתיב are likely original, and that was corrupted into the al tikrei, and our gemara text was basically the text before Rashi but emended to remove any phrase there in its entirety. The alternative is that our gemara, without any explanatory phrase, was original, and all these parallel explanations appeared in various manuscripts.
And finally, this, four lines down, starting in the middle of the line:
Paris, Bibliotheque Nationale , Heb. 671
ברכות לא ע"ב - ברכות לב ע"א
ואמר רבי אלעזר חנה הטיחה דברים כלפי מעלה שנאמר ותתפלל על ה' אל ה' לא נאמר אלא על ה' מלמד שהטיחה דברים כלפי מעלהwith the phrase in אל ה' לא נאמר אלא על ה'in Hebrew rather than Aramaic, and spelling it out in a bit clearer language.
We should not pass from this sugya without checking out if there are indeed variant texts to the standard masoretic text in sefer Shmuel, perek 1, pasuk 10. After all, maybe the text of the gemara was accurate, in terms of Chazal's writing, and it is our current printed Neviim which are faulty!
For this, we can turn to Vetus Testamentum, where we discover the following. In the main text, of course, he has על. On the bottom, listing the variants, he has:
That is, in one text, #150, instead of ותתפלל על ה, it has ותתפלל חנה אל ה. Thus, Chana is explicitly added and al turns to el. And in texts #30, 70, 96, 136, 149, 158, 180, and 250, al turns to el.
What are these texts? Are they super-accurate texts? He has an explanation of what these numbers mean after his entire presentation of sefer Shmuel. And so he writes:
At the end of the day, my assessment is that the error is in the Talmudic text, and not in our texts of sefer Shmuel. And Rashi is correct that the al tikrei is incorrect; but perhaps we should substitute the words of the gemara as found in other manuscripts, rather than eliminating the words entirely.