On this page
- The findings at a glance
- Why particles never stick (and why it's not your fault)
- What Japanese particles actually do, in 60 seconds
- Methodology and limitations
- Every Japanese particle, ranked by real usage
- The Big 5, with lines from unscripted videos
- は vs が: what 374,858 subtitle lines show
- Common in speech, introduced later in many textbooks
- When speakers drop particles entirely
- Practice: the Particle Drop game
- How to actually make particles stick
- FAQ
- The Bottom Line
The findings at a glance
The short version
- 736,715 particle uses across 374,858 unscripted subtitle lines
- 12 particles covered 77% of every particle use in the sample
- が narrowly exceeded は, by 62,575 uses to 61,856
- 25.2% of lines contained no particle at all
Frequency is not the only reason to learn something, but it is a useful way to decide where to spend limited study time. The ranking below is a priority guide, not a claim that lower-frequency particles do not matter.
Why particles never stick (and why it's not your fault)
You've read the particle chart. Twice. You still freeze between に and で when it's time to speak.
Here's the thing: it was never going to work. Japanese particles are the most listed, least practiced part of the language. Every guide hands you the same thing: a dictionary-style table, one stiff example sentence each, good luck out there.
Reading a definition feels like learning. It isn't. Memory research keeps landing on the same result: pulling knowledge out of your head (testing yourself) beats putting it in again (rereading), and it's not close. In one classic study, students who reread a passage four times recalled 40% of it a week later. Students who read it once and then tested themselves recalled 61%.1
This post combines two things that are usually separated: a frequency ranking from unscripted Japanese and retrieval practice built from sentences in the same corpus.
No invented examples. No "Naomi is American." Let's go.
What Japanese particles actually do, in 60 seconds
English tells you who did what with word order. "The dog bit the man" and "the man bit the dog" are different stories.
Japanese doesn't care about order nearly as much. It sticks a little tag after each word instead. Those tags are particles. Think of them as luggage labels: the word is the suitcase, the particle says where it's going.
Here's a toy example (ours, not from the corpus). Same words, same order, two swapped tags:
- 私が魚を食べた = I ate the fish.
- 私を魚が食べた = The fish ate me.
Two tiny characters just decided whether you had dinner or were dinner. That's the whole job description. Particles mark who's doing the action (が), what it's done to (を), where it happens (で), where it's headed (に), what we're talking about (は), and how the speaker feels about it (ね, よ).
That last group is the one textbooks barely mention. The data says it should be near the top. Keep reading.
Quick check before the ranking
Try one real sentence now. You will see the same destination-versus-venue distinction throughout the article.
Methodology and limitations
This is an original analysis of HayaiLearn's caption corpus. The goal was to measure particle frequency in spontaneous, Japanese-language video rather than in textbook examples or scripted entertainment.
- Corpus: HayaiLearn's library contained 9,640,020 Japanese subtitle lines from 24,151 videos when this analysis was run.
- Sample: We drew a random 480,000-line sample. After applying the content filters below, 374,858 lines from 13,049 videos remained, representing about 2.75 million analyzed words.
- Included: vlogs, gaming streams, podcasts, street interviews, variety shows, cooking and travel channels, and teachers speaking naturally.
- Excluded: anime, films, dramas, music videos, song covers, comedy skits, and Netflix content. Those sources are scripted, performed, or sung, so they were not included in the main ranking.
- Comparison group: 88,029 scripted anime, film, and drama lines from the same sample were analyzed separately.
- Counting: The totals use the particle tokens in HayaiLearn's analyzed subtitle corpus. The resulting unscripted sample contained 736,715 particle uses.
How to interpret the results: this is a frequency study of one large subtitle corpus, not a census of every Japanese speaker. Subtitle errors, ambiguous tokenization, repeated speakers, channel mix, and HayaiLearn's content categories can all affect the totals. Frequency also does not measure how difficult, versatile, or pedagogically important a particle is.
The aggregate counts are reported throughout the article so the comparisons are inspectable.
Every Japanese particle, ranked by real usage
With that scope established, three patterns stand out:
- Particles are everywhere. Roughly 3 out of every 10 analyzed words were particles: 736,715 particle uses across 374,858 lines, almost 2 per line.
- A tiny club does almost all the work. Japanese has well over a hundred particles on paper. The top 12 covered 77% of every particle use we counted. The top 20 covered 87%.
- Observed frequency differs from teaching order. が slightly out-appeared は, while ね, which is often introduced later, appeared more often than で.
The quick text version, for the skimmers:
| Rank | Particle | What it marks | Uses in 374,858 lines |
|---|---|---|---|
| 1 | の | linking, possession | 98,299 |
| 2 | に | destination, time | 69,232 |
| 3 | が | subject | 62,575 |
| 4 | は | topic | 61,856 |
| 5 | を | object | 47,401 |
| 6 | ね | agreement, "right?" | 42,939 |
| 7 | で | place, means | 41,563 |
| 8 | と | and, with, quotes | 38,962 |
| 9 | も | too, also | 30,211 |
| 10 | か | questions | 28,716 |
| 11 | って | casual quotes | 22,110 |
| 12 | よ | assertion, "I'm telling you" | 19,807 |
Study recommendation: learn these 12 first, in roughly this order. Recognize lower-frequency particles as they appear, but spend most of your deliberate practice on the forms you will meet constantly.
Why anime got cut (with data)
Anime dialogue is written Japanese performed out loud, so it earns its own bucket rather than a place in the unscripted ranking. To check the difference, we ran the 88,029 anime, film, and drama lines from the same sample separately. Per word, the unscripted group used ね 2.4 times more often than the scripted group, とか 4 times more, and って 1.6 times more. The scripted group used the dramatic enders ぞ 2.3 times more and ぜ 2.7 times more.
Anime remains useful listening practice, but copying its particle habits too literally can make everyday speech sound more theatrical than intended.
The Big 5, with lines from unscripted videos
Textbooks call は, が, を, に, で the core five, and the data mostly agrees: all five made the top seven. The observed order differs from the usual teaching sequence, and ね sits between を and で at #6. Every example below comes from an unscripted video in the corpus, rather than being written to illustrate the rule.
に: the arrow (69,232 uses)
に points at a destination or a point in time. Going somewhere? に. Doing something at 9 o'clock? に. It's the #2 particle overall, ahead of every textbook favorite.
中学校の時の友達の結婚式に行って
For example, I went to my junior high school friend's wedding.
The wedding is where the speaker headed, so it gets tagged with に. Bonus: count the three の's linking the nouns in that sentence. This kind of stacking helps explain its frequency.
が: the pointer (62,575 uses)
が picks out who or what does something, or what exists. When something shows up, exists, or is possible, Japanese reaches for が. And in unscripted speech it edged out は for the #3 spot.
めっちゃ達成感がありました
I felt a great sense of accomplishment.
達成感がある, "there is a feeling of accomplishment." Existence runs on が, not は. And yes, that's めっちゃ ("super") glued onto a polite ました ending. Real speech mixes registers like that constantly.
は: the spotlight (61,856 uses)
は doesn't mark who does the action. It marks what the sentence is about. Whatever comes before は is the thing on stage; everything after is commentary about it.
普通のポケモンの色違いは
A normal shiny Pokémon is...
A real streamer, mid-thought. Notice they set the topic with は and then just... kept going in the next line. Natives think in topics first. That's は doing its job.
を: the target (47,401 uses)
を tags the thing the verb happens to. Eat what? Watch what? Tell us about what? The answer wears を.
みんなのご飯を教えてくださいね
Please tell us about everyone's breakfast.
A host asking viewers to comment. The meal is the thing being told about, so it takes を. Check the tail of that sentence too: ください plus ね, a request with warmth stacked on top. Particles do teamwork.
で: the venue (41,563 uses)
で marks where an action happens, or what tool it happens with. The action takes place at the location; the location just hosts it.
エレベーターの前で会った
Met in front of the elevator.
The meeting happened at the elevator, so で. Also notice what's missing: nobody said "I" or "we". Real Japanese drops everything it can get away with. More on that in a minute.
Listening cue: に vs で trips everyone. Ask one question: is the place a destination or a venue? Arrows take に (結婚式に行く, go TO the wedding). Venues take で (エレベーターの前で会う, meet AT the elevator).
は vs が: what 374,858 subtitle lines show
Here's the contrast everyone came to see. Textbooks may spend a page on it while learners spend years noticing it. The corpus gives us a measurable answer.
Finding 1: it's a dead heat, and が is on top. が appeared 62,575 times, は appeared 61,856. A 1% gap, in が's favor. Textbooks introduce は in lesson one and treat が like an advanced topic. Unscripted speech treats them as equals, with が slightly ahead.
Finding 2: the mix flips with difficulty. In the easiest videos of our sample (difficulty under 3 on HayaiLearn's 1 to 10 scale), は beat が by a factor of 1.3. In the hardest videos, が beat は by a factor of 1.2. Casual chatting runs on topics ("me, I'm..."), while denser content narrates events and points at subjects.
So when does each one appear? Three patterns cover many everyday examples:
- Old news takes は, new news takes が. Once something is already on stage, it's a topic: は. The first time something enters the scene, or when which one is the whole point, it's が.
- Existence and feelings take が. ある, いる, できる, わかる, ほしい: all of these pull が. That's why the corpus line above was 達成感がありました.
- Negatives love は. In negative constructions, speakers routinely mark the denied thing with は for contrast, like this corpus line:
あんまり重い感じは全然しないですね
It doesn't feel heavy at all.
Study recommendation: stop trying to "solve" は vs が from rules alone. Notice the contrast in every sentence you watch this week. Repeated examples will build an instinct that a rule chart cannot provide by itself.
Common in speech, introduced later in many textbooks
The most surprising part of the ranking is how often several conversational particles appear despite receiving relatively little early textbook attention.
の: the most frequent in this sample (98,299 uses)
Roughly 1 in every 7 particle tokens was の. Textbooks often introduce it as "possession" ('s), but の also links nouns more broadly, and it stacks like Lego: 中学校の時の友達の結婚式, "my junior-high-days friend's wedding." Three の's, one breath, zero drama. It also hides inside casual explanation endings (んだ, its shrunken form ん, another 11,885 uses on its own).
ね: a top-6 conversational particle (42,939 uses)
Read this twice: ね out-ranked で. It beat も, か and から too. In the easiest, most conversational videos, ね was the #3 particle overall, breathing down は's neck and ahead of に, が and を. It softens a statement and invites agreement: "right?", "isn't it?", "eh?"
Here's one in the wild, from a hololive stream clip, one streamer teasing her collab partner:
あービビさん可愛いね
Ah, Bibi-chan, you're so cute~
No ね, and the sentence turns into a flat observation. With ね, it's an invitation: aren't they cute? That tiny sound is doing social work.
よ: the "trust me" particle (19,807 uses)
よ asserts: I'm telling you something you don't know yet. A real line from the corpus, one gamer reassuring another after a rough round:
いやいや上手かったよ
No, no, you were good.
The scale puts the phrase "optional sentence-ending particles" in perspective: ね, よ and their combination よね add up to 68,534 uses. That's more than は, the first particle many textbooks teach. These particles may be grammatically optional, but they carry important social meaning.
って: how people actually quote (22,110 uses)
Textbooks teach と for quoting speech. Real conversation overwhelmingly grabs って, the casual quote marker: "he said", "it says", "they call it". It out-appeared both よ and から. Here's one of Japan's biggest YouTubers reading a sign during a haunted-house challenge:
桃の木って書いてある
It says 'peach tree.'
Lower-frequency particles to recognize, not prioritize
Since we're ranking: へ ("toward") appeared 884 times. に appeared 69,232 times, about 78 times more often. Genki teaches へ in its very first chapters. And や ("and, among others") got beaten roughly 7 to 1 by its casual cousin とか. If your study time is limited, you now know exactly where not to spend it.
When speakers drop particles entirely
One more thing no chart will tell you: 25.2% of the lines in our sample contained no particle at all. A quarter of real speech just... skips them.
Casual Japanese drops particles the way English drops "are" in "you okay?" Two real lines from the corpus:
嬉しい。ハートマーク作ってくれてる。
Happy. They're making a heart mark.
Full grammar would demand ハートマークを作ってくれてる. Nobody said を. Nobody missed it.
うん。俺あれがいいな
Yeah. I like that one.
That should be 俺は. The topic marker vanished, the meaning didn't.
The pattern is consistent: は, を and が get dropped when context makes them obvious. The arrows and venues (に, で, から) usually stay, because they carry information you can't guess. And the sentence-enders (ね, よ) almost never drop, because they carry the feeling, and feeling is the point of casual talk.
Listening cue: don't copy this on day one. Learn the full versions first, then let repeated exposure teach you when omission sounds natural. Expect to hear it, though, or a large share of conversational lines may sound incomplete when they are not.
Practice: the Particle Drop game
Time to stop reading and start retrieving. Below are lines from unscripted Japanese videos with one particle blanked out. Tap the particle that belongs in the gap, and you'll get the answer, translation, and a one-line breakdown. Miss one? Good. That's the testing effect doing its job.
Estimated Time: 5 minutes
How to actually make particles stick
You now know which particles matter. Here's the system for making them automatic, in three moves.
1. Respect the ranking. The top 12 particles covered 77% of particle use in the sample. Master の, に, が, は, を, ね, で, と, も, か, って, よ before you spend much drilling time on こそ or さえ. Effort should reflect both frequency and communicative value, not only chapter order.
2. Meet them in content you can mostly understand. Particles don't stick from definitions because a definition gives your brain one meeting. You need hundreds of meetings, in sentences you mostly follow. Linguists call this comprehensible input: material just slightly above your level, what Stephen Krashen famously labeled "i+1", your current level plus one small step.2
The catch is finding content at that sweet spot. This is exactly why HayaiLearn grades every video in its library with a difficulty score from 1 to 10, and can rank videos by how many of the words you already know. Beginners live happily in the under-3 zone, which, as you saw above, is also where ね and よ run the show. If you're new to learning through video, our Japanese immersion guide walks through the whole setup.
3. Blank the particle, every time. When you save a sentence, quiz yourself by hiding the particle and guessing it back, exactly like the game above. Retrieval beats rereading. Ten sentences a day is plenty.
Estimated Time: 10 minutes a day
Try it: next time you watch anything in Japanese, pick one particle from the top 12 and count how many times you hear it in five minutes. You'll never unhear ね again.
FAQ
How many Japanese particles are there?
A lot on paper, very few in practice. A standard reference dictionary of Japanese particles covers over a hundred distinct ones.3
In our 374,858-line sample of unscripted speech, the top 12 particles did 77% of the work and the top 20 did 87%. Learn the short list; meet the rest in the wild.
Which Japanese particles should I learn first for JLPT N5?
Start with the Big 5 (は, が, を, に, で), then add の, と, も, か. That's nine particles, all in the sample's top 10, and they cover the core N5 patterns. Add ね and よ early too: they may receive less attention on the test, but they are common in conversation.
Is は or が more common in real Japanese?
They're nearly tied, and が actually wins. In our sample of 374,858 unscripted subtitle lines, が appeared 62,575 times and は appeared 61,856 times, a 1% difference. は leads in casual, easy content; が leads in harder content. You need both, so learn them as a contrast pair, not as "main particle plus exception."
Do anime characters use particles differently from unscripted speakers?
Yes, measurably. We compared 88,029 anime, film, and drama lines against 374,858 unscripted lines from the same library. Per word, the unscripted group used ね 2.4 times more and とか 4 times more, while the scripted group leaned on dramatic enders like ぞ (2.3 times more) and ぜ (2.7 times more). Anime is useful listening practice, but calibrate your own particle habits with unscripted speech too.
Why do Japanese speakers drop particles when talking?
Because context already does the job. In our unscripted sample, 25.2% of lines contained no particle at all. は, を and が vanish first when the topic and object are obvious. Particles that carry harder-to-guess information (に, で, から) tend to remain, as do sentence-ending particles such as ね and よ.
Do I really need to learn へ?
Recognize it early, prioritize it later. へ showed up 884 times in our sample; に showed up 69,232 times, about 78 times more often, and に can replace へ in almost every beginner sentence. Recognize へ when you see it (train announcements love it), but spend your drilling time on に.
The Bottom Line
Japanese particles feel impossible when you study them as a dictionary. They're very possible when you study them as a frequency list.
The 374,858-line unscripted sample gives a practical priority list: 12 particles account for 77% of observed particle use. ね ranks sixth, が slightly exceeds は, って is common in casual quotation, and a quarter of the lines contain no particle at all.
Learn the top 12. Meet them daily in videos you mostly understand. Blank them out and guess them back. That's the entire strategy.
Now go watch something unscripted. In this sample, the particles were waiting at a rate of almost 2 per subtitle line.
References
-
Roediger, H. L., & Karpicke, J. D. (2006). "Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention." Psychological Science, 17(3), 249-255. In Experiment 2, one week after study, repeated re-reading yielded 40% recall versus 61% for repeated self-testing. Study abstract. ↩
-
Krashen, S. D. Principles and Practice in Second Language Acquisition (1982), Chapter II. The Input Hypothesis describes comprehensible input using the shorthand "i+1." Original book PDF. ↩
-
Kawashima, Sue A. A Dictionary of Japanese Particles (Kodansha), which covers over 100 particles in alphabetical order. Publisher page. ↩
