A post to the graded-reader mailing list from March 23, 2008.
[this will be a bit of an experiment as to whether I can cut and paste formatted Greek and have it pass through Google Groups. I apologize in advance if it doesn’t work]
One aspect of the reader that seems to have received a lot of interest is the embedding of the target language (in my case Greek) in English.
It is important to note that this is entirely independent of the 95% of the code and data which has to do which choosing the order in which to learn things.
I wanted to explain a little about how it’s produced and what the variables are that could be tweaked or changed all together.
First of all, consider the very first block of text introduced:
John 3.26: So they came to John and said to him, “Rabbi, the one who was with you on the other side of the Jordan River, about whom you testified – see, he is baptizing, and everyone is flocking to him!” John 3.27: John replied καὶ εἶπεν, “No one can receive anything unless it has been given to him from heaven.
For those of you who don’t know Greek, καὶ εἶπεν means “and (he) said”.
This was generated because the ordering component of the software said that the first thing to be introduced is clause
NT.John.3_c117. That’s a clause reference from OpenText.org’s clause analysis of the New Testament. Part of my database is a listing of all the clauses, as identified by OpenText.org along with this unique identifier and what chapter/verse the clause comes from:
So my code knows that the clause to show is from John 3.27. I decided to always include the previous verse for context as well. So I retrieve John 3.26 and John 3.27 from a database containing the NET translation but annotated with the OpenText.org clause boundaries:
3.26 [c108 So they came to John ] [c109 and said to him, ] “Rabbi, the one who was with you on the other side of the Jordan River, [c112 about whom you testified – ] [c113 see, ] [c114 he is baptizing, ] [c115 and everyone is flocking to him!” ] 3.27 [c116 John replied ] [c117 and said, ] “No one can receive anything unless it has been given to him from heaven.
Notice that I haven’t annotated everything yet. It’s a slow and laborious process so I tend to just mark clauses as they are needed.
In some cases, I slightly alter the NET translation so there is something to annotate. This becomes challenging when NET has altered clause order and even more so when the Greek breaks apart words from the one clause that have to be together in the English. I still want to do more work in this area as the key thing to note is I never use the actual translation of the clause when introducing the clause; rather I use everything except the translation of the clause and that might make the problem easier if thought about in those terms (rather than what my annotation above focuses on which is annotating what English text corresponds to what Greek clause).
But this annotated NET is used to then produce what you see in the example-reader.html extra shown at the start. If other clauses were known at this point, they would be replaced by the Greek as well. Any clauses already known are show at normal weight and the new clause being introduced is shown in bold. Hence later on in example-reader.html (at step 13.)
John 4.49: The official said to him, “Sir, come down before my child dies.” John 4.50: λέγει αὐτῷ ὁ Ἰησοῦς, “Go home; your son will live.” The man believed the word ὃν εἶπεν αὐτῷ ὁ Ἰησοῦς and set off for home.
So, to summarize: the input to this part of the process is:
- what clause to introduce (by reference number)
- what verse this clause is in
- what other clauses are already known (by reference numbers)
- what the English text of the verse (from 2) and the one before are, annotated by clause references that can be replaced by Greek if known
The variables to this particular step are:
- the unit of text being introduced (in this example, a clause)
- the unit of text to show (in this example, the verse containing the clause and the verse before it)
There is no reason why the unit of text being introduced in Greek could not be smaller (a phrase or even a word) and the unit of text being shown in English larger (a paragraph, for example).
Note that the clauses I am currently dealing with included embedded clauses such as relative clauses and so in the John 4.50 example, we have the relative clause ὃν εἶπεν αὐτῷ ὁ Ἰησοῦς (“that Jesus said to him”) even though it might have been better to wait until the containing noun phrase were readable (which would, of course, have required knowledge of phrase boundaries)