|ב וַיְדַבֵּר מֹשֶׁה אֶל-רָאשֵׁי הַמַּטּוֹת, לִבְנֵי יִשְׂרָאֵל לֵאמֹר: זֶה הַדָּבָר, אֲשֶׁר צִוָּה ה.||2 And Moses spoke unto the heads of the tribes of the children of Israel, saying: This is the thing which the LORD hath commanded.|
Rashi writes that there were two separate transmissions, once from Moshe to the roshei hamatos, and once from Moshe to all of Bnei Yisrael. Thus:
Ibn Ezra writes that it was a communication to the heads of the tribes, and then, via them, to the children of Israel. Thus:
Aharon ben Yosef the Karaite is certainly no friend to midrash, and yet he also seems something awkward in this pasuk. He first cites Ibn Ezra as described above, but is unhappy with it. It is more correct that everyone should hear this directly from the prophet himself. Therefore, אֶל-רָאשֵׁי הַמַּטּוֹת לִבְנֵי יִשְׂרָאֵל is written awkwardly, and really should be read with an extra vav. It is el roshei hamatot and livnei Yisrael.
Meanwhile, Rav Saadia Gaon argues with both Ibn Ezra and Rashi, considering it to be "the heads of the tribes of the children of Israel."
Mendelssohn agrees with Rav Saadia Gaon, that it means the roshei hamatot of the bnei Yisrael. He explains the presence of the lamed, which otherwise could be interpreted as there being two targets. Without the heh of the definite article, it could have put everything in the construct form and had roshei matot benei Yisrael. The heh of Hamatot ruined that, and they showed the construct, "of", by means of the lamed.
What is motivating Rashi to say what he said? Well, they are the words of Chazal, from a gemara in Nedarim. It is not always the case that Rashi intends peshat. But here, if this is what Chazal say, and could be read into the pesukim in terms of the order of informing, then he should cite it.
Mendelssohn offers a reason for Rashi beside this, based on the trup. He notes that there is a zakef katon on the word Hamatot and a tipcha on the word Yisrael. And a zakef is a greater pause than the tipcha.
To bring it into Wickes terminology and theory, both zakef and tipcha are pauses of equal value, in that they both subdivide a clause ending in etnacha. But the zakef must have subdivided it first, or else we would not the zakef (for it would be subdividing a clause ending in tipcha). Therefore, we begin with:
וַיְדַבֵּר מֹשֶׁה אֶל-רָאשֵׁי הַמַּטּוֹת לִבְנֵי יִשְׂרָאֵל לֵאמֹר
This is divided into:
וַיְדַבֵּר מֹשֶׁה אֶל-רָאשֵׁי הַמַּטּוֹת
לִבְנֵי יִשְׂרָאֵל לֵאמֹר
which appears to denote two different targets. Otherwise, if we would only have the tipcha dividing first, we would have
וַיְדַבֵּר מֹשֶׁה אֶל-רָאשֵׁי הַמַּטּוֹת לִבְנֵי יִשְׂרָאֵל
and then perhaps
אֶל-רָאשֵׁי הַמַּטּוֹת לִבְנֵי יִשְׂרָאֵל
but dividing livnei yisrael off from el roshei hamatot so early seems to be suggesting that it is not one object.
Perhaps an argument could be made for the trup not intending that; or if it did, being derived from the midrash presented in the gemara rather than being the basis for it. At any rate, it is an interesting connection between peshat and trup.