Quite a set of questions! I think I will respond with an essay instead of taking them one by one.
These issues are complicated because translation is an act of communication, and what’s considered best depends on who we ask. Translators and readers who know the original language will have different answers than readers who only know the target language. Practitioners will often have different answers that philologists and scholars. I’ll give my own personal thoughts about this, though.
Chinese Buddhist translations went through different periods of methodology that caused them to change quite a bit in vocabulary and grammar when we pass from the pre-Tang Dynasty era (200-600 CE) into the Tang and Sung eras (600-1200 CE). That’s because during the Tang Dynasty native Chinese Buddhists began learning the Indic languages and translating them with large teams of editors, sometimes under Imperial auspices. The result was quite different from the translations made by Indian missionaries in the earlier period. Once these changes were in place, missionary translators adopted the new vocabulary.
The translation method that was used in the 4th century CE usually worked like this:
- A Central Asian or Indian monk with a memorized text recited it
- A Central Asian or Indian monk who was fluent in a spoken Chinese dialect translated the recited text
- A native Chinese assistant fluent in writing wrote it down
That’s the bare bones version. Sometimes there was a team of editors and assistants who functioned to check for accuracy, but the process involved at least two people because missionaries found it difficult to learn to write Chinese. This method worked well enough, but it introduces opportunities for miscommunication. The translator may not realize when the text is misspoken or mis-remembered. He probably isn’t able to personally check the written document that results, either. What we think is a translator error could be a reciter or copyist error.
Another issue is that it was a common belief that Daoism and Buddhism had the same basis philosophically. There was even a theory that Buddhism was a foreign version of Daoism that had returned after many centuries of change outside of China. Early texts translated Buddhist concepts by borrowing Daoist terminology. It works once you understand the original language, but it causes some grief when Buddhist ideas are conflated with Daoist ones, or the nuances are lost in translation. For native readers, it probably reinforced the idea that Buddhism was really a type of Daoism.
Another issue with the earlier translations is that we simply don’t know what some obscure, archaic Chinese terms mean precisely. Sometimes they are characters that fell out of use; sometimes they changed from one use to another over the centuries. The other related problem is that the transliterations used were not standardized yet in the 2nd-4th century, and so it’s impossible sometimes to decipher them today.
Faxian became the first famous pilgrim-translator when he decided to go with a mission of Chinese monks to India to retrieve a proper vinaya in the 390s. This was after the four Agamas were translated, but before Kumarajiva begins work. Faxian began the move to a new era of Chinese Buddhist translation. Faxian, Xuanzang, and Yijing all undertook Marco Polo-esque travel adventures through Central Asia, India, and Indonesia. They often spent a couple decades in these countries, studying under Indian teachers, learning to read and write Indic languages, etc. Xuanzang, for example, eventually settled in at Nalanda for a number of years to study under Buddhist scholars there.
So, you can imagine what a difference this made to the quality of their translations. They had both the language skills and the Indian Buddhist education to translate these texts to Chinese as native speakers.
Xuanzang revolutionized Chinese Buddhist translation, sweeping away the practice of borrowing native terminology and creating a method of more literal translation. He translated vast amounts of Indian texts like the Sarvastivada Abhidharma, Madhyamika and Yogacara texts, and even the entire Satasahasrika Prajnaparamita Sutra in 400 fascicles (which I think may be the longest scripture ever created). Many of these texts would disappear from India a few centuries later because of the Islamic invasions. The fact that he decided to bring an entire library back to China preserved a great deal of Buddhist literature.
Yijing was not quite as prolific, but he did translate an entire vinaya corpus. The part of the canon that was missing from these new translations were the Agamas, unfortunately. We get only retranslations of some texts considered important to the Abhidharma or Mahayana literature.
What makes these later Chinese translations “better” to us is that they are more transparent representations of the original Indic texts, usually BHS. They weren’t necessarily better to native Chinese readers because they were more cryptic. Like today, you had to pick up a working knowledge of Sanskrit to understand them. Sometimes Sanskrit terms are broken down into parts and translated literally in a way that doesn’t really capture the functional meaning in a passage. The grammar is also not very Chinese. It’s more a kind of hybrid Buddhist Chinese.
Another problem for scholars who want to find the earliest versions of Buddhist texts is that these highly accurate translations are later versions. We have good translations of later texts and not-so-understandable translations of the early texts in the Chinese canon. It’s frustrating to scholars studying early Mahayana texts, for example. I think this is what drives the search for Gandhari remnants. They want to find a way to lift the “fog” of the early Chinese translations. I think it’s a pot of gold at the end of a rainbow, personally. We’re just too far removed in time to get a really detailed picture.
Finally, to your questions about the Agamas themselves.
I don’t think there’s a better or worse translation between the four major collections. They were made by translators around the same time who worked with each other on different projects. Chu Fonian was an assistant to Gautama Sanghadeva [edit: Oops, confused Sanghadeva with Sanghabhadra. Chu Fonian actually had translated the Madhyama Agama separately, but it was lost], for example. Gunabhadra has a different style, but his translations are more fluent like Kumarajiva in my personal opinion.
The bigger issue with the four Chinese Agamas is the source texts rather than the translators. Chu Fonian was translating from a canon that incorporated Maitreya and early Mahayana thought. It can be seen in the Ekottarika Agama and in his translation of the Udana collection, Taisho 212. The Madhyama and Dirgha seem to be from similar canons, and perhaps the Samyukta fits with them more-or-less as well, but the Ekottarika is an outlier. It has many of the texts the Madhyama does, which is actually similar to the Pali Canon that places them in the Anguttara. Yet, there’s also a bunch of early Mahayana material, too. It’s quite a category buster.
Anyway, those are my thoughts on the topic in the time I have to jot them down. Hope it helps some.