Some notes collected after 2 weeks of Readlang usage.
- Translation Options: Sometimes translations seem to be done using translation services rather than explanations. Wouldn’t it be better if I could choose which translation suits the context best? I know I can type ‘edit’ to make changes, but it would be helpful to automate this process by presenting a list of options from Google, DeepL, LLM, etc. I could then select the most fitting translation by clicking on it, and the system would automatically integrate my choices.
- Connected Word Identification: A helpful feature for many languages is identifying connected words. For example, in Dutch, I might come across the sentence “Wie het hier ook voor het zeggen krijgt” and not understand its full meaning, even though I know the individual words. If I click on “krijgt,” I expect related words like “voor het zeggen” to be highlighted, since “voor het zeggen krijgen” means “to take control.” This would help me realize I need to learn the phrase, not just the word “krijgen.”
- Handling Separable Verbs: Another useful feature would be highlighting separable verbs, like in the Dutch sentence “Ik houd mijn uitgaven bij.” If I select “houd,” the word “bij” at the end of the sentence should also be highlighted, as the correct verb is “bijhouden.” While I can figure this out manually, the platform could improve convenience by recognizing such patterns.
- Phrase Linking in Flashcards: Currently, the Readland dictionary struggles to link phrases or words that don’t repeat the exact sentence in context. This means when I review flashcards, it doesn’t always correctly highlight or cloze the relevant parts of the phrase.
- Dutch Dictionary Integration: The Dutch dictionary feature isn’t working smoothly—I have to open a separate page for translations. I noticed the Van Dale dictionary can be easily parsed based on its page code, so this seems like a fixable issue. I’d even be willing to help with the parsing process. I appreciate dictionaries because they provide essential information about word forms, gender, articles, verb tenses, and examples, all in one place. LLMs can offer this too, but the dictionary is more reliable, and I’d prefer to access that info without switching tabs. Ideally, I could add dictionary details to my flashcards with one click.
- Customizable AI Prompts: I use LLMs for contextual translations, which saves time, but I’d like to be able to improve the experience. I use the “Chat with AI” feature to request explanations or check idioms, and I’ve even pinned a prompt to get the infinitive form of verbs and check if it’s part of a stable expression. However, I’d like to customize the hidden prompt directly, so I can get the exact response I need without asking follow-up questions.
- LLM Options: The current LLM used sometimes doesn’t provide the best results. Could you offer the option to use a more powerful (and potentially more expensive) LLM, perhaps with a ‘plus’ subscription or by allowing users to input their own API key?
- Word Limit on Phrases: The current limit of 12 words per phrase seems restrictive. Some sentences are longer, and I don’t see the value in limiting it to 12 words. While I don’t necessarily want to memorize longer sentences, I do often need to translate them.
- Manual Dictionary Additions: I want to check the translation of a word, phrase, or sentence before deciding if I want to add it to my dictionary. At the moment, any selection automatically creates a dictionary entry, which I find inconvenient.
- Selection Glitches: When selecting multiple words across different lines, the text is often glued together, and the selection jumps to a new line, which is frustrating. The text position shouldn’t shift due to selection—it disrupts the process and causes me to lose focus.
- User Feedback Tool: While the current forum setup is nice, it’s difficult to follow what’s trending and what features are being worked on. Have you considered adding a more structured feedback tool, like https://noorahq.com/? This could help with prioritization and transparency.