Friday, 17 July 2026

What should Japanese learners read? Following interests or building variety?

When choosing reading material for language learning, should we follow our interests or deliberately choose variety? It's a question I've found myself thinking about since launching Context Japanese.

Newspaper concept

One of the main ideas behind Context Japanese is that learners will enjoy language learning more—and stay motivated for longer—if they can read content that genuinely interests them at an appropriate level, rather than being limited to a fixed set of textbook topics. To make that possible, I wanted the site to offer a wide variety of topics. To achieve this, the concept of a newspaper with different article categories seemed a logical choice. A newspaper does not assume everyone wants to read the same thing. Different readers naturally gravitate towards different sections. It's still early and the data set is small, but I have been looking at user behaviour in relation to topic choice and some interesting insights are already visible.

Category selection

The first thing I have noticed is that users do not try to cover every category. Instead, they are selective and repeatedly choose the same categories. This is similar to how we might approach reading the news in our native tongue. We rarely approach news consumption in our first language by saying, "Today I should read one article from politics, one from science, one from sport and one from culture." We usually follow our curiosity. I often read the sports section and the top stories and perhaps some politics or business-related news, but rarely go into the health and lifestyle sections or arts and culture. One might expect that if you are using a site to learn a language, it would make sense to cover every category in order to expose yourself to the widest possible range of language. So far, that doesn't seem to be happening. 

Favoured topics

The behaviour of registered users already hints at a couple of interesting patterns. Users tend to gravitate towards one or two categories—we might assume these reflect their interests. One user, for example, has read eight articles, six of them in Arts & Culture and two in Lifestyle & Health, but none from the other seven categories. Another user has returned four times and on each occasion chosen a Science & Technology article without reading from another category. One more user has wider tastes, spreading their reading across five categories, but still ignoring the remaining four. Similarly, another user has chosen Lifestyle & Health for more than half of the articles they have read.

Possible implications for language learning

This data sets the scene well for the sort of investigations that could follow once there are more users on the site. I am interested in how topic choice affects learning, and whether there is a better approach: staying within the same topic area or deliberately varying it more widely. One possibility is that repeatedly reading the same type of content makes key nouns and phrases increasingly familiar. That removes one cognitive barrier and may allow learners to devote more attention to new grammar and unfamiliar ways of expressing ideas. Equally, it may be that different topics naturally expose learners to different vocabulary and grammatical patterns, making variety the more effective approach. That is something I would love to investigate further.

Let's take sport as an example. If you are a football fan, does it make sense to read football articles every day? You would probably become very familiar with words like match, player, controversy and dispute. That familiarity might free up enough attention to notice new grammar or unfamiliar turns of phrase. But what might you miss? Would you encounter words like research, which might appear regularly in Science & Technology articles, or bankrupt, which you might see more often in Business & Economics? Perhaps repeated exposure to one topic also means repeated exposure to particular grammar patterns and ways of expressing ideas.

Looking forward

As Context Japanese grows, I'm looking forward to revisiting these observations in six months' time to see whether they still hold. For now, I'm left with one question: should learners follow their curiosity, or deliberately push themselves towards unfamiliar topics? My instinct is that both have a role. Familiar topics may help learners build confidence and process language more easily, while variety may expose them to a wider range of vocabulary and structures. As Context Japanese grows, this is one of the questions I hope to explore further.

Thursday, 9 July 2026

Six Months Building Context Japanese: Five Things I've Learnt

It's about six months since Context Japanese went live, so seems a good moment to pause and give an update. Here are five things I have learnt.

1. My concept has been validated

I'd define the concept as the provision of varied and fresh content that is adjusted to a range of learner levels that can maintain a regular reading habit will appeal to learners. Users have found the site organically online, signed up for accounts, and returned to use the site. The most encouraging sign isn't the number of users—it's that some of them came back. I had set myself the goal of having returning users within one year. Now I can add a number to that to adjust the goal. I would like to have a regular user base of 100 by the one year mark. I believe this is an achievable goal with some more proactive promotion and distribution of the site.

2. Topic sourcing is more challenging than article generation

One area of the app that is key to making the premise work is topic sourcing. I have found that using news APIs to source content is not the best method. News APIs are built for a different purpose. I can imagine they are excellent tools for sourcing information and articles on a particular topic. For example, if you want to track all the latest information in the agriculture industry in the United States, then a news API would be a good tool. For my purpose, sourcing a variety of content on a range of topics, they prove less useful. For this reason I have moved towards using rss feeds for the topic sourcing ahead of APIs. Doing this I am able to be more selective about the sites being used to source topics. The range is more limited, but this can be broadened over time with some investment in researching good feeds.

3. Lower level simplification seems harder (but is it?)

Grading lower level articles seem harder, however, it is possible that it is just a case of being able to see the areas that need to be improved more easily. I believe Context Japanese is a good tool for beginners. However, it is the intermediate plateau (which a motivated beginner can enter quite quickly) where I believe the main target audience for this product are. The main job of Context Japanese is to help learners move beyond the intermediate plateau and towards confident, independent reading. I believe that more work will be needed on the higher levels to make this a reality. I am not ready for that yet but am looking forward to the challenge still to come.

4. Hiragana v Kanji at the lower levels is a challenge

One unique challenge with Japanese is the alphabets. As a Japanese learner myself, I am aware of how intimidating kanji can be, but I am also aware of how confusing a wall of hiragana is. Japanese is quite unique in this sense (are there other languages that mix alphabets like this!?). Kanji are the most difficult of the three alphabets used, but if the text only uses one of the simpler phonetic alphabets, it can be really challenging to understand the meaning because it is difficult to parse where one word ends and another one begins. I believe beginners must get exposed to simple Kanji early and begin reading with these as they will do at higher levels or they will likely become disillusioned quickly. I have had differing experiences with the three AI providers I am using, with some more likely than others to throw out the hiragana wall. There are some articles still creeping through at N5 level like this and I have a few ideas about how to fix this.

5. Renaming the project changed how I think about it

Context Japanese is now the third and I think final name for the product. Yomu Yomu was cute, but I don't think it really embodied what the product is aiming to do. I prefer Context because it also guides my thinking about the future product roadmap while capturing the essence of how language is acquired through the app. The next function release I have in mind will really feed into this idea and compound the learning possible using the product.

Looking forward

So here we are. There are some interesting challenges ahead. The levelling, topic selection and ensuring a variety of content for all tastes, optimisation for traversing the plateau to name a few. I will explore all of these further in coming months. If I finish with one question from the areas I have outlined that I will look at next, it would be on the subject of topic v level? Which is more important for a learner trying to learn a language? Would you rather have a topic you weren't interested in, perfectly levelled for you to learn from? Or a topic you really were interested in that was either too difficult or too easy for you? Which would you prefer? Something to think about.