Rainbow Latin Reader

[κάττα]

Merry Christmas and Saturnalia! I've just finished this Latin learning tool and now sharing it as a gift. This video explains it:

The main webpage is here:

Rainbow Latin Reader - Index

This project is still in an experimental phase so any inputs from teachers are appreciated.
Like: Are the most important properties of the words highlighted, or would would there be others? Etc.

I will also share the software in a couple of weeks which generated it. It just needs to get into a better shape first.
Many thanks for any feedback!

 

[cinefactus]

I think it would be helpful to have a function to just expose one word at once, to give people practice at reading naturally rather than using the puzzle solving approach.

 

[κάττα]

@cinefactus You mean the color? So like: Only show the color if hovering it? That would be easy.

Or if it's a single word, then it could also display all information from the lemmatization.

 

[cinefactus]

No I mean covering all the words and revealing them one at a time when the user asks for it, such as recommended by William Gardner Hale. Displaying the colour and lemmatisation would aid this technique.

 

[κάττα]

Thank you! That looks like a great book. I'll read it.
Then I think there would be around 3 modes of reading, and something on the user interface which would indicate the mode. When a word is clicked, it would be selected.
Modes:
1. Show all words with colors. Word selection only triggers the dictionary window.
2. Show only the selected word (with color). For the other words only their framing rectangles are shown, without colors. Questions:

• - Should it show punctuation characters?
• - How to step through the words? By clicking on them or by a keyboard button, like "space" or the arrow keys?

3. Show all words without colors. (Or: the selected one could be colored.) Otherwise the same as 1.

I think I'll add the remaining lemmatization information in the dictionary window. The colors only show the fraction of that data, and it might be useful. (Though I'm not sure how dependable it is.)

 

[Godmy]

Very beautiful and thanks for the efforts! This is always awesome when one does something like this for Latin (and for free) and there too few people in the world to fully appreciate it!

If I had any objections, it would be pretty much along the lines as I wrote yesterday in a sister thread:

Quickly checking the first video, I feel just a slight skepticism in whether rephrasing the difficult parts of the classical work (albeit in many various tiers) helps in the long run. That is, having gone through all the tiers, it is likely the reader will finally get why the classical author worded it the way he did for that one particular part of the poem, but what will the reader do when he stumbles upon another classical part (in poetry) which is not covered by the book? (e.g., some entirely different work)?

....
... it was you, not the reader, who figured out how the syntax and semantics worked in the classical piece and it was you who rephrased it and you do not necessarily pass the know-how of how you did it onto the reader. The rephrasing makes the classical piece available indeed, but does it teach the student how to do it? To teach the student how to think analatically when reading difficult classical works as to figure the answer themselves WITHOUT resorting to a "cheat sheet", is much more difficult (yes, it is doable, but it is difficult), yes, it most often requires using a vernacular language and will not provide the final answer or a translation... but at the end, the reader must figure it out on their own. Without a translation (ideally), without rephrasing, without too many hints. (ideally with zero).

What I mean to say, your goal is to make the reader being able to rephrase difficult stuff at the end themselves. I suppose you do hope that the reader will uncover the method on their own *somehow* inductively, but I'm not sure about that. I'm afraid they will just get used to fall onto the simpler tiers anytime the syntax and semantics proves too difficult, because, why not, they paid for it! Otherwise they could have just read the original text for free (although they can't). This is the old: give a poor man a fish or teach them how to fish...

So, that's just my skepticism when it comes to the method. But hey, it looks really awesome, it must have taken many many pains to be released in such a beautiful & well illustrated form and I think it will find its readers! The various tiers thing is a nice touch!


============================

On a similar note, I remember when I coded up the Syntax Graphs tool at http://linguax.com (it produces graphs like this e.g.) 12 years ago, I was so amazed by it. I immediately thought I must write a book where I will go every every verse in Ovid's metamorphoses and provide a graph for it, thus give the reader the solution. But then I realized that was the whole problem: I'm giving the reader the solution. If the reader wants a solution, they can just open the vernacular translation and read the semantics without ever fully enjoying figuring out the original language. It would have gone against what I consider a good didactic method...

So, while this is helpful for a single text, it only enables the reader to ... not learn how to parse the language themselves. Teaching the method to analyze the text correctly on the go as you FLUENTLY read it from left to right is much more difficult than either 1) giving them a cheat sheet 2) let them try to tackle the sentence as a puzzle. (=to read first without any understanding and then hunting individual elements of the sentence ex post)

A common misconception of many Latin readers and teachers is that you read a sentence and then start the hunt for subjects, objects, attributes, adverbials.... and you try to connect it together. But a native reader or anyone achieving a passive fluency in Latin would never read a natural-language text like that (no matter the word order!). You always read from left to right and you understand the thing on the go. Yes, you can teach that and you can learn how to do it. You read a word, you go in your mind through various options of what case it could be, what the probabilities of a given case are given its position in the sentence (or a gender*) and you go on and then gradually eliminate the options until it all clicks in at the end once you reach the final word: no puzzle but a method to learn to read fluently. More you do it, more instinctive it becomes until you read these texts almost as fluently as if you were reading English.

*e.g. a neuter noun in the canonical case is much more likely to be an accusative, since in active clauses, the verbs often require an agent and a neuter, by its nature, rarely can do that, thus you assume accusative first, only then, by elimination, you come back to assuming it was nominative; another similar tip like that is that first declension words ending on "a" (in a text where vowel lengths are wanting), you assume ablative first and only then by elimination nominative. It is not because ablative would be more frequent, but because it is less intuitive to the mind and harder to figure out if you do not assume it beforehand, unlike nominative singular.

Anyway, still, given your objective for this task: good job!

 

[κάττα]

@Godmy The intention for the colors is not analysis of any kind, but that they narrow down the possible meanings from the dictionary, thus reducing the possible combinations to keep in mind. Software like Whitaker's Words can only analyze a single word, thus it cannot provide the actual case of the words, but only a list of possible cases and meaning. The colors are coming from the work of Thibault Clérice, who worked on an AI model that uses the context to tell the actual cases of the words.

 

[κάττα]

@cinefactus From that book or "pamphlet" now I can see the future of Latin learning! It's here:

We could create an AI model that can do exactly this for any sentence in any book.
I would call this: "Zero abstraction language learning."
One would not even need to know what a "noun" or what "accusative" is.
Though I would rather finish this project first as I have too many unfinished projects

 

[κάττα]

Actually ChatGPT can already do that. It only needs the context in addition to the sentence:

From the Latin text below please analyse the sentence "Sed ut tum ad senem senex de senectute, sic hoc libro ad amicum amicissimus scripsi de amicitia.".
Please break it into smaller parts and describe each part in simple English terms.

"Cum enim saepe mecum ageres, ut de amicitia scriberem aliquid, digna mihi res cum omnium cognitione tum nostra familiaritate visa est; itaque feci non invitus ut prodessem multis rogatu tuo. Sed ut in Catone maiore, qui est scriptus ad te de senectute, Catonem induxi senem disputantem, quia nulla videbatur aptior persona quae de illa aetate loqueretur, quam eius, qui et diutissime senex fuisset et in ipsa senectute praeter ceteros floruisset; Sic, cum accepissemus a patribus maxime memorabilem C. Laeli et P. Scipionis familiaritatem fuisse, idonea mihi Laeli persona visa est quae de amicitia ea ipsa dissereret, quae disputata ab eo meminisset Scaevola. Genus autem hoc sermonum positum in hominum veterum auctoritate et eorum illustrium plus nescio quo pacto videtur habere gravitatis: itaque ipse mea legens sic afficior interdum, ut Catonem, non me, loqui existimem.

Sed ut tum ad senem senex de senectute, sic hoc libro ad amicum amicissimus scripsi de amicitia. Tum est Cato locutus, quo erat nemo fere senior temporibus illis, nemo prudentior; nunc Laelius et sapiens, sic enim est habitus, et amicitiae gloria excellens de amicitia loquetur. Tu velim a me animum parumper avertas, Laelium loqui ipsum putes. C. Fannius et Q. Mucius ad socerum veniunt post mortem Africani; ab his sermo oritur, respondet Laelius, cuius tota disputatio est de amicitia, quam legens te ipse cognosces.

Sunt ista, Laeli; nec enim melior vir fuit Africano quisquam nec clarior. Sed existimare debes omnium oculos in te esse coniectos unum; te sapientem et appellant et existimant Tribuebatur hoc modo M. Catoni, scimus L. Acilium apud patres nostros appellatum esse sapientem, sed uterque alio quodam modo: Acilius quia prudens esse in iure civili putabatur, Cato quia multarum rerum usum habebat et multa eius et in senatu et in foro vel provisa prudenter vel acta constanter vel responsa acute ferebantur; propterea quasi cognomen iam habebat in senectute sapientis."

Response:

1. "Sed ut tum ad senem senex de senectute"

• Sed: A conjunction meaning "but" or "however," introducing a comparison with the preceding discussion.
• ut tum: "Just as at that time" or "As then." Refers back to a past situation, drawing a parallel.
• ad senem senex: "To an old man, as an old man." Refers to the author (Cicero) writing about old age to another elder (Cato in Cato Maior de Senectute).
• de senectute: "About old age." Indicates the topic of the earlier work (Cato Maior).

Translation: "But just as then, an old man wrote to an old man about old age..."

2. "sic hoc libro ad amicum amicissimus scripsi de amicitia"

• sic: "So" or "in the same way," continuing the parallel being drawn.
• hoc libro: "In this book." Refers to the current work (Laelius de Amicitia).
• ad amicum: "To a friend." Refers to the intended recipient of this work, who is a close friend of Cicero.
• amicissimus: "As a very close friend" or "as a most devoted friend." Describes Cicero's relationship to the recipient, emphasizing the intimacy.
• scripsi: "I wrote." Refers to the act of composing this treatise.
• de amicitia: "About friendship." Indicates the topic of the current work.

Translation: "...so, in this book, as a very close friend, I wrote to a friend about friendship."

Though it would be nice to have something where you only need to click on parts of the text to get these explanations, even offline.
(People don't realize that everything that is online can be gone in a couple of years, as this happened already many times.)

 

[κάττα]

I made the source code public in the meantime:
- https://github.com/bolner/rainbow-latin-generator

But now I'm more motivated to create a new project that would collect English explanations for parts of the sentences with API calls to ChatGPT
(Though I have some other projects to finish before that )

 

[Godmy]

@Godmy The intention for the colors is not analysis of any kind, but that they narrow down the possible meanings from the dictionary, thus reducing the possible combinations to keep in mind. Software like Whitaker's Words can only analyze a single word, thus it cannot provide the actual case of the words, but only a list of possible cases and meaning. The colors are coming from the work of Thibault Clérice, who worked on an AI model that uses the context to tell the actual cases of the words.

My problem indeed is not Whitaker's words, nor did I imply it was. I actually encourage readers to use it, because they sometimes do not realize the options they have and thus limit their understanding by under-analysing some word.

My problem is, as I said in length, the coloring itself (while I understand the coloring itself is the main benefit you brought to the text... and then I go in length about my opinion of it, I would only repeat myself). I probably recommend that you read my critique again, more carefully, since you may have misunderstood.

I would also be careful with ChatGPT when applied to Latin. It sometimes authoritatively tells you something about Latin text and then, if you really are an expert in it, you read it and find its slightly but fundamentally wrong. I stopped using ChatGPT altogether, because the amount of bull*itting it does is just too dangerously high. This is a good thread.