In the Arabic Version, Rabbeinu Avraham always quotes his sources in full. In the Hebrew, by contrast, these references are generally shortened, sometimes by tens or hundreds of words. This makes sense if the translator’s goal was not to preserve a faithful copy of the “Ma’amar” [Discourse] for posterity... -- Rabbi Moshe Meiselman TCS pg. 99
The Accuracy of the Translation
The full Hebrew version of Discourse that we have today is a translation from the original Arabic. Is there any way for us to verify that the translation was competent or accurate? Happily, the answer is "yes". While it is not possible to validate the entire translation, an Arabic version exists for approximately one third of the Discourse. Rabbi Meiselman stipulates that "[a] comparison of the published text with the surviving Arabic segment reveals that the translation is generally competent." (TCS pg 97). 
- There are a number of places where the Hebrew translation differs from the Arabic including one case where the meaning of the text is completely reversed.
- The translator has “used poetic license to spice up the text”.
- The translator has added words to either clarify the text or for “innocent embellishment”
- The translator in one case includes a translation which “bears no relation to the extant Arabic version at all”.
We'll base our analysis on a side by side comparison of the Hebrew translation of the Discourse and the new Hebrew translation from the surviving Arabic fragments which can be found in the Appendix of TCS.
This bit of the Discourse discusses the following Midrash (Taanis 31a):
In the future, the Holy One, praised be He, will take a dance for the righteous in the garden of Eden placing Himself in the center, and everyone will point at Him with his finger and say, (Is. 25, 9) 'Behold, this is our God; we have waited for him; we will be glad and rejoice in His salvation.’ (Soncino Translation)Rabbeinu Avraham unsurprisingly rejects the plain meaning of the Midrash as a gross anthropomorphism. The phrase "in the future" indicates that the Midrash is actually describing an element of the non-physical world to come. Souls in the world to come, now detached from the corrupting influence of their physical bodies, can come to a higher understanding of God than they could while attached to a body in this world. The happiness of a dance is an allegory for the happiness that the soul experiences at its new deeper understanding of God, represented by the pointing of the finger.
The “old” translation of the Discourse has the following (the bold section of the translation is where the Rabbi Meiselman claims that the meaning has been reversed):
The reward for the righteous who are remembered for life in the world to come, is an understanding of God, the elevated one, that they were not able to understand in this world in any manner. This is the ultimate good such that there is none higher than it. And he allegorized this happiness of this understanding as the happiness of a dance. And he also compared the understanding of each individual, that which he was not able to understand initially [while in this world], [with the allegory] "and each and every one gestures towards Him with a finger.”The new translation has it this way:
The reward for the righteous who are remembered for life in the world to come is their understanding of Him, the elevelated One, what was impossible for them to understand in this world. This is the ultimate reward and the pinnacle of happiness, and he allegorized it to the happiness of a dance. And he allegorized what each of them reached in understanding of Him, the elevated One, in saying, each and every one of them gestures at Him with his finger.
Here is another section that Rabbi Meiselman takes issue with. Rabbeinu Avraham continues in his explanation of the Midrash. In the Midrash, the participants in the dance recite a Pasuk which refers to God's salvation. This indicates the fact that the soul survives the destruction of the body in death.
The old translation:
And he brought a proof of the escape of the intellective soul from the wrath and the fury, with God’s help, in that which the pasuk says “and he will save us”.The new translation:
And he brought proof of the privilege and the escape from the true destruction in the next world in that which the pasuk says “and he will save us”.Rabbi Meiselman’s first criticism of this translation is the addition of the words “intellective soul” in the translation. In my humble opinion, the intent of the translator was clear. He understood that Rabbeinu Avraham did not believe that one lives on intact as a physical being in the world to come; rather the body is destroyed while the intellective soul lives on. (See Mishneh Torah, Yesodei Hatorah 4:14-16 where the Rambam make this explicit). Thus the “added” words do not alter the meaning in any way, but simply serve to clarify.
Next, Rabbi Meiselman characterizes the use of the phrase “the wrath and the fury” in place of the “true destruction” as an attempt by the translator to use “poetic license to spice up” the text. He also characterizes the words “with God’s help” as an “innocent embellishment”. However, the intent of the translator with both of these seems clear. It is traditional to substitute a euphemism or otherwise avoid direct reference to very negative occurrences and to soften them. The words “the wrath and the fury” were not chosen at random. They are a reference to the same words in Deuteronomy 9:19, where they mean precisely destruction.
כִּי יָגֹרְתִּי, מִפְּנֵי הָאַף וְהַחֵמָה, אֲשֶׁר קָצַף יְהוָה עֲלֵיכֶם, לְהַשְׁמִיד אֶתְכֶם
For I was in dread of the anger and hot displeasure [the wrath and the fury], wherewith the LORD was wroth against you to destroy you.The addition of the words “with God’s help” is not an embellishment, but a expression of hope that we are not afflicted with such a punishment.
[Edit: Commenter "Magiha" points out an alternate explanation here: the Roshei Teivos (abbreviation) for "with God's help" and "the world to come" ("B'Ezras Hashem"/"B'Olam Haba") are identical. Thus, the Hebrew translator could have used the abbreviation, which was then expanded incorrectly by a later copyist].
Let's examine Rabbi Meiselman's example “where the translation bears no relation the extant Arabic at all”. (TCS pg. 98). The context is Rabbeinu Avraham’s explanation of the following Aggadah (Berachos 5a)
R. Levi b. Hama says in the name of R. Simeon b. Lakish: A man should always incite the good impulse [in his soul] to fight against the evil impulse. For it is written: Tremble and sin not. If he subdues it, well and good. If not, let him study the Torah. For it is written: 'Commune with your own heart'. If he subdues it, well and good. If not, let him recite the Shema'. For it is written: 'Upon your bed'. If he subdues it, well and good. If not, let him remind himself of the day of death. For it is written: 'And be still, Selah'.
Here is the old translation on the right and the new one on the left:
The old translation explains that the aggadah is providing advice on the method to subjugate his passions and desires to his intellect. The advice to “let him study the Torah” works as follows: one should both concentrate on and enunciate pesukim that involve the subjugation of human desires.
Rabbi Meiselman points out that the new translation merely mentions that the recitation of pesukim helps because speech is the “vessel” of thought and so reinforces proper thought, but says nothing about specifically mentioning pesukim that involve the subjugation human desires. Did the translator make that part up and insert a “translation [that] bears no relation the extant Arabic at all”?
It is easy to see that the answer to this question is “no”, if you read the passage in context. Here is the preceding paragraph:
Here, Rabbeinu Avraham explains that the first step is to subjugate human desire through proper thought. Then, if this is not sufficient, the advice is to recite pesukim, since speech reinforces thought. The translator interpreted this to mean that the topic of the pesukim themselves would the subjugation of human passion.
Moreover, the translator didn’t invent this idea. Rabbeinu Avraham himself supplies this explanation in the following paragraphs:
Here, Rabbeinu Avraham explains why reciting Shema is the next piece of advice that the Gemara has. He says here explicitly that one of the reasons for this advice is that the Shema includes the subjugation of the passions in the commandment “do not stray after your heart and after your eyes”. The translator applies Rabbeinu Avraham's explanation for reciting Shema as the reason for recitation of other pesukim.
We have shown untrue that "the translation bears no relation the extant Arabic at all". But why did the translator introduce Rabbeinu Avraham's explanation one paragraph early if it was not in the original text?
The answer is clear if you compare the text of the Midrash as quoted in the Hebrew version and Arabic versions as we see here:
The Hebrew version omits the clause “If not, let him study the Torah. For it is written: 'Commune with your own heart'. If he subdues it, well and good.” As a result, it is likely the translator considered Rabbeinu Avraham’s discussion of the recitation of pesukim as part of his discussion of the recitation of Shema. Thus is further reinforced by the following:
The Hebrew translation omits the segue “If this is enough, otherwise…” between the discussion of the recitation of pesukim from the recitation of the Shema. The translator understood Rabbeinu Avraham's discussion of pesukim and Shema to be a single unit since his version of the Midrash did not have an independent clause referring to the recitation of pesukim. As a result, there was no transposition at all in the translation.
The Purpose of the Translation
Besides the accuracy of the translation, Rabbi Meiselman notes two aspects of the Discourse which indicate “translator’s goal was not to preserve a faithful copy of the Ma’amar for posterity, but to draw support from it for his own thesis.”
In my humble opinion, discerning the intent of the translator is close to irrelevant for our purposes. We have direct access to the Arabic versions of portions of the Discourse, we can judged directly for ourselves the competence of the translator. Since the translation is accurate for the significant fraction of the Arabic text that we have, we can have confidence in the translation. Nevertheless, we'll address the evidence that Rabbi Meiselman brings and show that they do not bear any relevance to the issue.
His first piece of evidence for this is the fact that the translation of the Discourse is introduced by the following introduction (Rabbi Meiselman’s translation TCS pg 91):
I have found written by HaRav Rabbeinu Avraham Ben HaRav Rabbeinu Moshe ztz”l is a sefer he composed in the Arabic language called al-Kafiyah, the following words which contain great benefit for my needs, and translating them from his language into the Holy Tongue is fitting for me and proper.In the one manuscript of the Discourse, there is an even longer prologue which contains a possible reference to a larger work by the prologue’s author: “And this principle and its ramifications will be explained thoroughly in the second chapter”. (As Rabbi Meiselman notes, this prologue may not have been written by the translator as it is omitted from a different manuscript).
To begin with, if it is true that the translator intended the translation be incorporated into a larger work (and this may not be the case), it would not indicate that the translation is somehow deficient. Rabbi Meiselman includes the Discourse in his own work in order to support his thesis, but it is for this very reason that he treats the Discourse with painstaking care.
Next, Rabbi Meiselman then notes that the sources quoted in the Arabic are abridged in the Hebrew version, sometimes by hundreds of words. He states that “[t]his makes perfect sense if the translator’s goal was not to preserve a faithful copy of the Ma’amar for posterity, but to draw support for it from his own thesis”. In my humble opinion, this is a complete non sequitur.
To begin with, all of the shortened references are in Hebrew and not in Arabic. The inclusion of the full text of these references would not serve to preserve the Arabic for the Hebrew reader in any way. They would be simply an exercise in tedious copying while adding nothing to the store of the world’s knowledge.
Secondly, in the age before the printing press, it is not at all clear that making the translation longer would serve to “preserve a faithful copy of the Ma’amar”. The shorter the work, the more likely it could be copied further and spread to a wider audience.
Thirdly, it is quite traditional to abridge sources in religious works. The Talmud itself abridges pesukim all the time with the assumption that the reader knows what is being referred to. The Vilna Gaon's commentary on the Shulchan Aruch often consists of a string of references to the Talmud or Rishonim with no other explanation. In the religious context, the translator would then have no reason not to abridge sources to save costs as well as to make his translation more compact and easier to copy.
Finally, as Rabbi Meiselman notes, the different manuscripts abridge the sources differently. Thus, it is not even clear that it was the translator who did the abridgment. (Ironically, Rabbi Meiselman himself reports in the preface of TCS (pg. xxii) that he was forced by his publisher to abridge sources in order to make the length of his book practical).
At this point, we've shown that there is no evidence to question the reliability of the Hebrew translation of the Discourse. We'll now turn to placing the Discourse into a modern context.
Comments are both welcome and encouraged. I'll make every effort to address any questions or arguments posted in the comments.
 The full text of the Discourse that we have today is known to us only in Hebrew translation. The Discourse was first published in 1836 in the journal Keren Chemed and eventually included in 1887 Vilna edition of Ein Yaakov based on a manuscript currently housed in the Oxford Bodleian Library (TCS pg. 94). Two other manuscripts also exist, one of which is unreliable as it contains obvious changes, as well as an attempt by its “author” Eliezer Eilburg to attribute part of the Discourse to himself. (TCS pg. 95). However, it is clear that this manuscript was derived from the Oxford manuscript so we can rule out the possibility that Eilburg had inserted the section which references Chazal’s scientific statements. (TCS pg 95 note 264).
In addition to the Hebrew translations, a version of the Discourse in Arabic (approximately the middle third) was found in the Cairo Geniza. In the appendices of TCS, Rabbi Meiselman has published a version of the Hebrew text which utilizes all three manuscripts as well the publication of Discourse in Kovetz Teshuvos HaRambam. TCS also includes a publication of the Arabic text, along with a new translation by Rabbi Yaakov Wincelberg with editing by Rabbi Pinchas Korach and with a comparison to the Hebrew Translation.