I’ve seen Analayo say this in another less-well-argued paper, but here again, he either is unaware or avoiding the definition of the angas found in the Yogâcārabhūmi (418c7-9) that defines the vyākaraṇa:
云何記別。謂於是中記別弟子命過已後當生等事。或復宣說已了義經。是名記別。
What is vyākaraṇa? It refers to predictions of events in the later births of disciples after they die or to sutra that explain the meaning of discourses that have been given. This is called vyākaraṇa.
Notice that the Theravada tradition doesn’t recognize the first meaning, from what I’ve gathered, which perhaps inconveniently was taken up by bodhisattva theorists to create a genre of texts in which Buddhas give predictions to bodhisattvas about their future buddhahood, one of which is included in the Madhyama Agama. However, that isn’t the primary meaning originally, in my opinion; there are sutras in which the Buddha tells about a disciple’s destiny after dying (such as whether they attained parinirvāṇa, etc).
What is meant by the second definition? I’ve gathered that it means sutras in which someone comes to the Buddha and asks for a explanation about a given term or topic. These sutras usually name the person who comes forward, and in the case of notable disciples they were collected together into samyuktas. This is the meaning that appears to be followed in SA and SN, so it may well be the older one.
Turning back to the passage Analayo is quoting, it reads like this:
當知如是一切相應略由三相。何等為三。一是能說。二是所說。三是所為說。若如來若如來弟子是能說。如弟子所說佛所說分。若所了知若能了知。是所說。如五取蘊六處因緣相應分。及道品分。若諸苾芻天魔等眾。是所為說。如結集品。如是一切粗略標舉能說所說及所為說。即彼一切事相應教間廁鳩集。是故說名雜阿笈摩。即彼相應教。
It should be known that all of the samyuktas essentially come from three characteristics. What are the three? 1. The teacher, 2. what’s taught, and 3. to whom it’s taught. Whether it’s the Tathagata or his disciples, they are the teachers, as in the “Teachings by Disciples” (SA 28-33) and “Teachings by the Buddha” (SA 34-51) divisions. Whether it’s what’s understood or understanding (itself), this is what is taught, as in the “Five Acquired Aggregates,” “Six Sense Fields,” and “Dependent Origination” samyukta divisions (SA 1-6) as well as “The Path” division (SA 7-16). If the assemblies are monks, gods, Mara, etc., they are who are taught, as in the chapters that collect them together (SA 17-27).
I’ve added the samyukta numbers for convenience. Essentially, though, this passage divides SA into three broad divisions. Analayo is correct that Ybs doesn’t actually say each corresponds to one of the three angas. That isn’t the point. Even if it did, it would be a late era Mahayana commentator telling us his theory.
Now, in the Theravada SN, we have a very close correspondence between SA 17-27 and SN 1-11, which is literally called the “Sagatha” division in the Theravada tradition. So, we have a clear division of verse-centric sutras in both traditions; that’s one aṅga.
If we consider sutras to be straightforward lectures on doctrine without any special features to cause them to fall into other genres, then we can see that SA 1-16 falls easily into that criteria, and these samyuktas appear to be grouped together consciously as doctrinal subjects. For SN, however, it’s a messier affair. The Nidana Vagga mixes together doctrine with collections on disciples. The same is true of Khandha Vagga, which begins with the aggregates samyutta but includes a samyutta on Sariputta. So, SN doesn’t separate the doctrinal samyuttas from those that center around disciples the way SA does.
So, let’s look at these disciple-topic samyuktas. Take SN 23 the Radha Samyutta. Suttas 1-22 all take the format of Radha asking the Buddha about a topic. It’s only at the end, Suttas 23-34, that we see lectures given to Radha personally. SA 34 is the parallel to SN 23, and it has 133 sutras (counting alot of variants at the end that are like SN 23.24-34). The first nine sutras are also Radha approaching the Buddha and asking questions, then afterward the sutras are personal teachings given to Radha. So, we see a pattern. Generally speaking, if we consider the first sections as older and later sections as gradual accretions, we could say that these samyutkas were originally cases of someone wanting an explanation of a prior teaching, and that fits Ybs’s definition of vyākaraṇa. Then, they were later expanded to include teachings to that particular disciple.
In the Sarvastivada tradition, at least, these disciple samyuttas were consciously grouped together, whereas in the Theravada canon that’s no longer the case if it was at a earlier time. We’d want to look at all of these disciple-centered samyuktas to see if there’s this underlying pattern, but these issues are left aside by Analayo in his arguments, which are essentially throwing shade on a teacher from another tradition by tossing out many objections without investigating the matter further. It’s easy to undermine a theory with skepticism in this way.