My daily reading speed log after six months of reading visual novels

Summary
Over a span of six months, I dedicated my time to immersing myself by reading Japanese visual novels, meticulously tracking various metrics related to my reading experience, including time spent reading and my reading speed. These findings have been comprehensively organized within a Google spreadsheet for easy reference.

The primary objective behind this endeavor is to provide individuals with a tangible insight into the realm of achievable progress in reading speed when undertaking a Japanese language learning approach focused on reading. It’s also worth noting that I started reading visual novels early around Wanikani level 4, with subsequent levels achieved within an average span of 7 to 8 days.

*For those curious about the strategy I employed, I provide a more elaborate exposition below. Additionally, I delve into my perspective regarding the role of Wanikani on the outcomes yielded by this experiment.

Method

I started reading Visual Novels (VNs) as a learning medium when I was level 4 on Wanikani. VNs possess a distinct advantage due to their popularity in Japan and frequent incorporation of voiceovers. The combination of text and audio that they offer proves invaluable. Line delivery assists in understanding the meaning of a sentence, and also assists in associating words with their pronunciation. I mostly read VNs that mirror contemporary Japanese daily life, ensuring I learn common vocabulary faster.

Prior to embarking on my VN journey, I started by quickly reading Tae Kim’s guide and learning a few hundred common words in Anki. My approach to learning with VNs is described as follows: whenever I encounter an unfamiliar word in a sentence, I looked it up. For more intricate word patterns, I turned to my search engine to determine if they were grammar constructs. If so, I read several example sentences with their translation to comprehend the grammar rule’s impact on meaning. This method facilitated learning common vocabulary and grammar organically, resembling a spaced repetition system sans external tools—especially since it takes quite a few tries for things to stick. A Texthooker tool expedited this process, enabling instant word lookup within the VN. I did not spend time extensively analyzing sentences; the optimal learning zone lies where the material slightly challenges comprehension. I allotted 1 to 2 seconds per sentence for comprehension if I did not understand them after reading them once.

This regimen persisted for half a year, which I documented within a Google spreadsheet tracking my reading time and characters read each day.

In tandem with reading, I did my Wanikani lessons and reviews each day. Additionally, I did 3 Bunpro lessons and 5 new Anki cards daily while also finishing their respective reviews every day. I used Anki with the 6k core deck and sometimes added words which were not sticking while reading.

Throughout this process, I employed a unique strategy with Wanikani. For efficiency, especially in higher levels, I cheated Wanikani’s internal SRS system. I’d close the app upon answering incorrectly and repeat this process until I answered correctly. This accelerated my Wanikani progress allowing me to get new lessons every three to four days. However, I did not cheat on enlightened-stage items to keep my burned-stage items pure. Despite my divergence from Wanikani’s norm, I think this approach complements daily reading practice. My main goal is to become familiar with new vocabulary and kanji, rather than learn them in Wanikani.

True vocabulary absorption stems from reading. A rough grasp through Wanikani enhances vocabulary understanding from encounters in context. Thus, I think a steady influx of 30 new kanji weekly proves more advantageous than perfect recall in Wanikani. Wanikani enhances the pace at which I learn new vocabulary as I encounter them, but genuine vocabulary mastery culminates from dedicated reading and finding the kanji and vocabulary in context.

Results

The culmination of my six-month endeavor was marked by significant enhancements in my reading speed. It’s important to note that the reading speed data exhibits strong variations, attributed in part to things my texthooker captured which did constitute actual significant text. Some VNs incorporate stylistic kana strings, contributing to fluctuations in reading speed counts. Examples are a person screaming “あああああああああ” and the likes. Also, Internal code being dumped between scene changes and character names being counted with each sentence also contribute to some counts being inflated. For instance, “Little Busters Ecstasy!” has a consistent reading speed above 20,000 character per hour, while “Air” has a speed dip to 14,000. Still, a consistent upward trend within each individual VN is evident.

Furthermore, the difficulty of vocabulary of certain VNs also plays a role. VNs like “Planetarian,” rich in sci-fi elements, and “House of Morgana,” which is rich in medieval elements, introduced large amounts of unfamiliar vocabulary. This also caused reading speed differences between VNs.

“Kanon” and “Air” are captured in a “pure” way by my texthooker, meaning the characters measured are characters representing actual meaningful text. By this measure, a conservative estimate of my reading speed increase is ~5,000 to ~15,000 characters per hour, achieved through an average of 2-hour sessions per day over six months. The time spent to complete different VNs also showcases a strong increase in reading speed. I spend 140 hours on “Clannad” (1.3 million characters) while my completion time of “Little Busters Ecstasy!” (1.6 million characters) is 95 hours.

While these findings offer a quantitative lens, my subjective sense of growth is equally large. I think I can now navigate most VNs centered around everyday Japanese life without a texthooker, comprehending plots and enjoying the experience with ease. It is uncommon for sentences to demand more than 1 to 2 word look-ups at this point. In contrast, when I started even menu navigation was a hurdle. Revisiting VNs like “Kanon” and “Air” is an entirely new experience, as nuanced content and sentences are far clearer now. My present level of comprehension stands as a significant departure from my initial starting point. I now find myself able to fluidly read slice-of-life Japanese cartoon subtitles, understanding them with ease sometimes. I am confident that if presented with an N3 test, I would likely pass at this time. Furthermore, I am optimistic about achieving N2 proficiency by the end of the year.

Discussion
Expanding upon the synergy of my reading journey with Wanikani, I’d like to state that while Wanikani is a good tool, it thrives in tandem with reading. The vocabulary that Wanikani covers, although often important, doesn’t encompass all common vocabulary. Also, Wanikani barely offers hiragana words by design, which constitute a significant portion of everyday language.

Frequent encounters with common kanji words while reading naturally helps in internalizing them, rendering later levels in Wanikani more manageable. This dynamic also extends itself to reading, where familiarity with Wanikani-learned kanji facilitates comprehension and faster learning of vocabulary encountered through reading. Among all the tools I employed, except those designed to facilitate VN reading like the texthooker, Wanikani has helped the most.

The commitment of 1 to 2 hours daily to reading is crucial for substantial improvement in my opinion. During my own journey, days with reduced reading time yielded noticeable decreases in progress. Language acquisition is a time-intensive endeavor, necessitating thousands of hours for proficient Japanese skills.

For those inspired to integrate VN reading into their learning regime, it’s imperative to acknowledge that there are big initial hurdles. The first visual novel I read “Kanon” is known to be easy, but it can still introduce complex literary sentences such as “もう忘れていたとばかり思っていた、子供の頃に 見た雪の景色を重ね合わせながら…”. Almost all vocabulary will be new, and knowing which hiragana sequence is a word or grammatical inflection will be entirely unclear. The early stages demand the celebration of small victories: grasping single sentences, comprehending new vocabulary, and understanding major plot elements. Every triumph accumulates, contributing to a smoother learning curve over time, and increasingly larger achievements. It is important to recognize that this journey requires thousands of hours.

I can definitely recommend resources like TheMoeWay or Animecards for those wanting to start on this journey, as they provide invaluable starting points. Although I lack formal training in linguistics or pedagogy, a cursory review of related literature highlights I did showed a certain alignment with the methodology employed for language acquisition shown in these resources—particularly when incorporating Visual Novels (VNs). I omitted a brief literature review comparing their approach to both implicit and explicit learning, along with some relevant papers on the subject, for the sake of brevity. However, if there is interest, I’m more than willing to share these papers while they remain accessible to me. Lastly, I am still a beginner, but I think the data I gathered could be interesting for fellow beginners, and I think that most of the things I mentioned align with the views of more experienced learners.

I believe many beginners tend to overlook the power of reading, yet it holds significant potential for both enjoyment and learning, even right from the start, particularly when a substantial amount of time is dedicated to it.

EDIT: removed a learning resources which contained links to pirated content.
EDIT #2: Added the name of the learning resource but I did not link to it since it is has a good starting guide.

24 Likes

As someone learning mainly through VNs from a while ago, this is pretty cool, thanks for sharing :slight_smile: . It also seems like a serious improvement, congrats! I never considered a spreadsheet tbh but I like how everything displays and is recorded over time for reference.

You mentioned that you were doing Anki 6k core and sometimes mined vocabulary from the VNs but not usually, so I take it that you mainly looked words up and kept going, and the reading speed recorded is not significantly affected by creating cards and so on, right? Do you feel like just reading and moving on has been enough to really learn and absorb the words that you didn’t do SRS for? I’ve tried both methods (usually when I’m tired of the SRS occasionally), and when I was just reading I always felt a little bit “uneasy” trusting the process, so I eventually went back to mining unknown vocabulary. I never used SRS to learn English, but Japanese somehow feels a bit different to me. It gives me the feeling of having something “tangible”, though in the end as you say real mastery of those words comes from reading them hundreds of times in different contexts and sentences. As you learn more and more words, do you feel like just reading and moving on is enough to keep sustaining your learning in the long term, especially as the words you learn become rarer to find in the wild?

Lastly I’m just curious, do you use monolingual dictionaries, is it something you’re interested in using, or not really?

4 Likes

SEE PEOPLE, THIS IS THE POWER OF VNS.

And you can mention The Moe Way still and just not link to them.

Great progress. I was a little (very) doubtful when I saw the 27kch/hr but your explanation cleared it up. Seems like an incredibly productive and fun 6 months you had. Godspeed bro

4 Likes

I enjoyed reading your analysis, thanks for the report; very inspiring :slight_smile:

3 Likes

Yes, reading speed is unaffected by card creation. I would also pause my texthooker when there are cutscenes or other times when I was not reading. In fact, for most words that I wanted to mine I just went to my 6k anki deck and set the review date to be that day, since it contains most of the common words I struggled remembering. The statistics I collected are really just the time it took me to read and quickly lookup unknown words.

The words I mine are usually the kanji I remember having seen quite a few times before, but also remember not sticking. For example 謎 is one of the words that was just infrequent enough for me to struggle with each time it came up. I am sure that I would have remembered them with time without mining, it is just that some of those words got annoying to not know ;).

I think most words that an average Japanese person knows can be learned organically without mining. Natives also learn vocabulary by reading and moving on in most cases. Mining words and putting them into a dedicated SRS system works well if the vocabulary you are learning falls far beyond something you would see every day or week. This is mostly going to be obscure technical terms or antiquated terms which are important for something like Kanji Kentei 1, but I think this exceeds what most Japanese learners want to learn by a long shot. Still, I think in a lot of those cases changing the things you read will help a lot, since vocabulary that is rare in one context is common in an other context.

It is the same in English, if I write something like: “Full waveform inversion proved suitable at imaging the Silurian aged faults, however due to the large geophone spacing there is aliasing in the frequency-wavenumber domain”, I am sure most native English speakers are going to have no clue as to what that half the words in the sentence mean or imply. However, if you read geophysics papers for a living all those words are going to be so common they become second nature.

Dedicated SRS systems are great at expediting the process of vocabulary acquisition though, but I think a lot of the benefit also comes from spending a bit more time to try to recall a flashcard. But recalling a word in a SRS system does not imply a true “mastery”, since in a real life context multiple words are spoken by second, which is a lot less time than what it takes most people to recall something in a SRS context. SRS should aid reading, not the other way around. Lastly, there is also this paper from Karpicke and Bauernschmidt (2011) where they found that the spacing between recalls does not give any significant differences. Only no spacing was found to not have any significant effects on long term comprehension.

I use a bilingual dictionary. I’ll switch to monolingual when I know all common vocabulary. I think the main benefit from monolingual dictionaries comes from them having more vocabulary, and being able to explain certain nuanced vocabulary better.

2 Likes

Thanks for sharing your thoughts, I appreciate the extensive reply :wink: .

2 Likes

Wow, very cool that you reread Kanon and that the second time around was almost twice faster

2 Likes

I have done my own research on SRS timings and had come to a similar conclusion, except that what I found seemed to indicate that the number of repetitions over some static interval was more impactful than expedient recall over a multiplicative interval for committing items to long-term memory, and that traditional flashcards severely trump SRS for commitment to short-term memory. It could be that a combination of zero-interval and short static interval repetition may achieve the same results in less time.

Also, great job btw. I’ve been slacking lately due to being laid off and feeling morose, but I think I’m ready to get back on the horse.

Ha, I was just last night thinking “I’ve been reading manga for an hour a day for 5 months now, surely I’ve gotten faster, but I have no way of being able to tell”.

The first time I did WaniKani I did all 60 levels in 2 years and then forgot half of it, but at least I was able to tangibly measure my progress. Now that my focus is on reading manga, I just have to trust I’m getting better.

“I’ve been reading manga for an hour a day for 5 months now, surely I’ve gotten faster, but I have no way of being able to tell".

This seems off. I feel like you should easily be able to tell if you have gotten better in a month, let alone five months. Unless you are already reading Japanese at a very high level and only encounter new words every week or month, I think something is going wrong. I personally don’t read comics at all, and I am unsure how they compare to what I read, but I still would not expect such a drastic plateau.

For me, the difference between my reading ability five months ago and now is drastic, even my listening ability has grown significantly as a result from more intuitively understanding things from reading. Even if I reread things from a month ago the improvements are glaring.

You might want to consider how many of the words you are currently reading were unknown to you even a month ago, same for word recall when you encounter words which you know but take a while to remember. I am not sure if comics allows for complicated run on sentences with multiple clauses, but also consider if you are able to more intuitively understand those with each month as well if you do encounter them.

If you simply do not encounter many cases where any of the aforementioned is a problem, increase the difficulty of the comics you are reading. If you still struggle reading with what you are currently reading, but at the same do not improve, I would suggest getting a more information dense media, like light novels or actual novels.

Well the problem is that I used to do a certain number of chapters each night, but then I changed it to one hour each night, because the amount of dialogue in each chapter varies greatly.

Also some parts are just simpler dialogue, and some are more complex explanations of things, so it’s hard for me to just say that I’ve gotten faster, because it depends greatly on the difficulty of what I’m reading.

Also I’ve just never kept track of it. I don’t remember what my experience was like when I started consistently reading every day 5 months ago. It’s too much of a gradual process for me to be able to tell.

1 Like