• Word Lists: on a list's page, words written with kanji now show their reading underneath, as the dictionary search results there already do, and a long meaning wraps instead of being cut off. • Word Lists: the Saved list's row shows its star once instead of twice. • Corrected 22 example sentences where a number and its counter were read wrong: for example, 四月 is now read しがつ, 九時 くじ, 三千円 さんぜんえん, 十分 (ten minutes) じゅっぷん, and 四人 よにん.

• Word Lists: search the whole dictionary from a list — by kanji, kana, romaji or meaning — and add a word with one tap. • My text: when a paste is over the limit (200 sentences or 20,000 characters), the notice now leads with the counts, and Add asks before saving the shortened text. VoiceOver reads the notice. • Corrected 22 dictionary entries, mostly readings you type in example sentences (何を is now read なにを), and fixed stray apostrophes in 18 English translations of Practice passages. • iPad: when the status bar at the top of a ride runs out of room, it no longer splits a number across two lines. It tightens first, then leaves out the least-needed items, which VoiceOver still reads. • With VoiceOver at the accessibility text sizes, the Verbs drill now reads out the score its status bar hides. Small text on the results screen has a little more contrast.

• The screen at the end of a ride now tells you your streak and how many words and verb forms will be due for review tomorrow. • With VoiceOver at the accessibility text sizes, the ride's progress now also reads out the numbers the status bar hides at those sizes. • A long sentence's romaji hint now wraps onto a second line instead of being cut off. • My text: pasting more than 200 sentences or 20,000 characters now tells you how much is kept and how much is left out, and a text can be deleted from its context menu (right-click on a Mac, where it could not be deleted before; touch and hold on iPhone and iPad). • The small grey text on the About page is easier to read.

• Readable at the largest text sizes. With the accessibility text sizes on, the results screen, the ride and drill displays, the typing coach, Word Lists, the Ride Log, Stats, Practice, Settings and About used to break words mid-letter, wrap numbers onto a second line and push words off the edges of the screen. They now rearrange to fit. • A ride or drill that ends before you type anything no longer congratulates you with a grade and 100% accuracy. It says the ride ended before its first word or sentence, or the drill before its first answer. • Practice's labels now appear in Chinese when the app is in Chinese. • The small grey captions in Settings and the on-device counters in About are easier to read, and the accuracy chart shows your first ride instead of an empty chart.

• A new Stats card: what you keep missing. The app has always known which kana refused your keystrokes during a ride, and always forgot at the end of it. It now keeps a running tally and shows the ones that come back across different days, with the rule behind the worst of them. It stays on your device. • The first-launch intro now names all six modes. Verbs, Sentence and Listen were in the app and mentioned nowhere a new rider would look. • The Ride Log is readable at the large accessibility text sizes. Every column used to squeeze until the values wrapped one letter per line, and the title and Back button ran off the edges of the screen. • A conjugation card the drill could not actually build a prompt for no longer counts toward the review button, the badge, the reminder, the widget or the forecast. It made all six promise work that no review could clear.

• Practice your own Japanese. Paste in a song, a news item, a page of your textbook — the app works out the readings, shows them above the kanji, and you can correct any it gets wrong. Your text stays on your device and is never uploaded. • The conjugation drill now notices how long a form took you. Until now every clean answer scored the same, so the forms you laboured over left the review rotation as fast as the ones you knew cold. They now come back when they should. • Pausing stops the clock on the word you are typing, not just on the ride. A word interrupted by a pause, a sheet, or the app going to the background is no longer graded on the time you were away. • On a Mac or iPad the ride now shows your speed as you go, over ridden time — the same number the Ride Log records afterwards. • The menu's route strip now responds wherever you tap it. Only its text did before.

• The road continues past Kyōto. The Tōkaidō ends there and always did; now there is a second journey beyond it — 8 stretches west along the inland sea, Ōsaka to Nagasaki, each with its own sky, light and road surface. It is opened by one optional, one-time purchase, from a single row in Settings. • The Tōkaidō stays free and complete, all 8 stretches of it, and nothing you already had has moved behind anything. This version takes nothing away from anyone. • The menu's route strip now marks where you actually are on the road, and says so once you have ridden the Tōkaidō to its end. • On a Mac, a short window no longer cuts the menu off at both ends — it scrolls. • About now carries an address to write to, and the app's on-device counters. Those counters stay on your device and are never sent anywhere.

• Dictation now draws on 5,931 sentences, 161 more than before. Those 161 had been held back by a check that judged them indirectly. Each has now been compared against the voice one at a time, and these say exactly what their answer key says. • The sentences still held back are held back on a direct measurement of what the voice actually says, rather than on an estimate — so what Dictation marks you against is what you hear.

• Dictation now draws on 5,770 sentences, 10 more than before. Those 10 had been set aside on an assumption rather than a measurement — that a word the voice misreads in one sentence is misread in every sentence. Each was checked on its own against the voice, and these say exactly what their answer key says. • The App Store listing is now available in Japanese, and the reading-passage count in the English and Chinese listings was out of date.

• After a ride that went well, Nihongo Ride may now ask whether you would like to rate it on the App Store. It waits until you have come back on a second day and finished a few rides, it does not ask after a rough ride, and it will not ask again for months. • That is still nothing to do with the network. The app keeps its own count of when it asked and when it decided not to, on your device, and that count never leaves it.

• 5 example sentences were spelling out their own word the wrong way. These were all verbs, which is why they had gone unchecked: a verb in a sentence is conjugated, so its dictionary form never appears and the check that guards these sentences could not find it. 潜る, to duck under a shop curtain, was written as "moguru" -- to dive underwater, a different verb that happens to share the character. 気に入る was written as "ki ni hairu" instead of "ki ni iru". • The menu now tells you how many words a drill will actually give you. The weak words button announced every word you had ever reviewed and then ran 15; the conjugation review button announced every form that was due and then ran 12. • A sentence or dictation ride now counts sentences everywhere it counts them. The progress display during the ride, and the summary card you can share, both still called them words.

• Forty-seven example sentences were spelling out their own word the wrong way. Every sentence carries the reading you type and the small kana printed above the characters, and in these the two did not match the word the card teaches -- 何を was written as "nan wo" instead of "nani wo", 四つ as "yontsu" instead of "yottsu", 二日 as "futaka" instead of "futsuka". Counters and dates were the worst affected, which are the first irregular readings anyone meets. • Fifty more sentences are available in Dictation. They had been held back because the voice did not say what the answer key said -- and it turned out the voice was right, so fixing the sentences released them. • The results screen after a sentence or dictation ride now says what each number counts. It was reporting sentences under a heading that said words, and listing whole sentences in boxes meant for single words.

• The words that stopped you can now be acted on. After a sentence or dictation ride, tap a word to save it, hold it to add it to one of your lists, or ride just those words in a quick session that leaves your review schedule alone. • A word the app cannot identify with certainty stays on the list and offers nothing, instead of pointing you at the wrong entry. A particle you misheard in dictation, or the tail of a verb, is worth telling you about and is not a word to study -- so it is named and left alone. • A results screen no longer says "to review" about a ride that schedules no review. Sentence rides, dictation and weak-word practice have never fed the review queue, and the tile now says what it actually counts. • Your Stats page no longer counts verb forms you can never practise again. A word withdrawn from the dictionary left conjugation cards behind that no amount of practice could clear, and both "forms practiced" and "tough forms" kept counting them.

• After a sentence or dictation ride, the results screen now names the words you were stopped on, with their readings. It could already tell you how many mistakes you made, and when a mistake followed a typing rule it explained the rule -- but when you simply did not know a word, or did not catch it, it never said which word that was. In dictation, which word is the whole answer. • Fixed a riding streak that disagreed with the dots beside it. A ride that crossed midnight counted for both days in the number but lit only one dot, so a chain you had not broken showed a gap in the middle. • Weak-words practice and the due verb-form review were handing back short sessions. Words withdrawn from the dictionary took up places in the run and were then dropped, so a session that offered fifteen could give you twelve. • Passage practice no longer counts time while the app is in the background. Putting the app down mid-passage used to be written into the Ride Log as riding, which stretched how long the ride looked and slowed down its speed.

• Fixed a review count that was too high. If a word had been withdrawn from the dictionary after you studied it, its card kept counting toward the number on the app icon, in the widget and in the reminder -- and could never be cleared, because a card whose word is gone can never come up for review. Those counts now match the words the app will actually give you. • 64 cards now say which reading of their spelling is the everyday one. Where the app teaches you the less common reading of a kanji -- and the ordinary reading has its own card -- the card says so, instead of leaving you with two cards that look identical. • Dictation can now follow what you are studying: hear the sentences for a saved word list, or for the words due for review today. It tells you how many sentences it can actually play, which is a smaller number than the list's -- some sentences are left out of dictation because the built-in voice reads a kanji in them differently from the way the app writes it.

• Dictation, a new mode. The sentence is spoken, not shown, and you type what you hear. Replaying is free and unlimited -- this is listening practice, not a memory test -- and the results tell you how many replays a run took, because that is the part a score cannot say. • Sentence mode can now follow what you are actually studying: practise the sentences for a saved word list, or for the words due for review today. It tells you how many sentences it found instead of quietly padding the run out with unrelated ones. • Dictation leaves out the sentences where the built-in voice was measured to read a kanji differently from the way the app writes it. Those stay in Sentence mode, where you can see the reading, rather than being played to you and then marked wrong for typing what you heard.

• More example sentences at every level: 6,737 in all, up from 6,458, and every level is now at or above 91% of its words. Nearly all of them can be typed in Sentence mode. • 28 definitions corrected or extended. Two were teaching you to write incorrect Japanese: てっきり was defined as "surely", but it always means you were sure and turned out to be wrong, and 素直 was defined only as "obedient" where it usually means frank or candid. • Words that no honest example can teach -- where every natural sentence reads the kanji a different way than the card does -- are now identified and left without one, rather than given a sentence that would train the wrong reading.

• Example sentences at every level. The corpus grows from 3,722 to 6,458 sentences: N2 goes from 34% to 93% of words covered and N1 from 23% to 86%, so Sentence mode and furigana now have material at the levels where they were nearly empty. • Every new sentence was checked for the thing that matters most in a typing app: that the context forces the reading the card teaches. Sentences that would have you typing the wrong reading of a kanji were rejected rather than shipped -- 21 at N1 alone, including ones whose Japanese was otherwise flawless. • 20 definitions corrected. Two were describing a different word than the card teaches: 強請る is the word for pestering someone for something, but its English read "to extort". • Words that could not be given an honest example -- where no natural sentence produces the reading the card teaches -- were left without one rather than given a misleading one.

• NEW: Sentence mode. Type a whole example sentence instead of a single word -- over 3,700 of them, at every level. Your speed and accuracy are tracked the same way, and the sentence tells you which word it was teaching when you finish. • NEW: Furigana. The reading now sits above the kanji in example sentences, so you can tell which reading a sentence uses -- some kanji have several, and until now there was no way to know. Turn it off in Settings if you would rather work it out. • More example sentences, and corrections to the Chinese translations of 35 of them: tense that did not match the Japanese, a word that read as a riot where the Japanese said commotion, a wrong character in a verb. • Four definitions were describing a different word than the card teaches -- the entry for the reading げ carried した's definition, so two different words showed the same meaning.

• N3 example sentences, nearly complete. 95% of N3 words now have one, up from 40% -- 807 new sentences, each with English and Chinese. • Every sentence was checked for the thing that matters most in a typing app: that the surrounding words actually force the reading the card teaches. Sentences that would have had you typing the wrong reading of a kanji were rejected rather than shipped. • A few word definitions gained a missing sense, so the definition on the card explains the sentence beneath it -- 単位 now says "credit (academic)" and not only "unit", 笛 says "whistle" and not only "flute". • Two entries were showing the same Chinese definition twice.

• One setting for hints, instead of two that did not know about each other. Choose Always (full romaji on screen, as before), When stuck, or Off. "When stuck" shows nothing until you are genuinely stuck -- repeated different attempts at the same character -- and then OFFERS the answer rather than handing it over. Leaning on a key is not being stuck, so the app does not count a held key as trying again. Taking the offer costs what it should: the word scores minimally and comes back for review. Time Attack never offers -- the clock runs either way. Off means off. • Your word lists are safer. A list of more than 500 saved words is no longer silently trimmed, and if the app cannot read your lists at startup it now says so and refuses to overwrite them instead of quietly replacing them with an empty set. • Practice no longer hides the characters you are supposed to be typing on short screens, and its text scrolls at every text size. • Accessibility: at the large text sizes the Time Attack countdown no longer truncates 35 seconds into something that reads as 3, list buttons and journal dates stop clipping, and charts grow with their labels. VoiceOver fixes too -- the Best WPM tile said "em dash" where it meant "no rides yet", the BEST badge in Speed trend was on screen but unreachable, and your best speed is now the same number in the journal and the stats.

• NEW: the Typing Coach. When a mistake keeps happening across different words, the app now names the pattern -- particles typed the way they sound (konnichiwa for こんにちは), a dropped small tsu, kya typed as ki+ya, m before b/p -- shows YOUR keystrokes next to a spelling that works, explains the rule, and offers a targeted drill built from words you already know. All on your device, no AI guesswork: the app knows exactly which key was refused and why. • 779 new example sentences for N5 and N4 words -- generated, machine-checked against the app's own reading and level data, then reviewed before shipping. Nearly every N5/N4 word now has one. • Recorded numbers are honest now: paused time no longer counts as riding time in your WPM, a ride that crosses midnight keeps your streak, and on Mac, stray presses of Tab or the arrow keys no longer count as typos or reschedule words you typed correctly. • Data fixes: eleven headwords that displayed one word while grading another (ジェット, けれど, キロ and friends), and your review history now survives a damaged file -- one bad record costs that record, not everything. • Small honesty fixes: passage practice shows the passage's length instead of calling everything N5, the space bar no longer silently skips your passage, and clearing a word list asks first.

• Fixed: some verbs were taught with the WRONG conjugation. Nouns read -iku (hoiku, hatsuiku, shiiku and friends) were drilled with the irregular ending that belongs to iku, and three words that merely sound like verbs (hirou, yosou, kisou) were conjugated from the wrong pattern entirely. If you practised those, the answer you were graded against was not real Japanese -- sorry. They are correct now. • Fixed: two entries showed the wrong word. One headword was missing its last character, and one N5 card showed the wrong word for the reading it taught. • Fixed: review reminders and the app icon badge ignored verb-conjugation drills, so if you were practising only conjugations you were never reminded and the badge stayed at zero. Both now count words and conjugations together. • Accessibility: at the larger text sizes the results tiles overlapped each other and cut digits off the numbers, the main buttons truncated their labels, and the route strip on the home screen collapsed. Everything now grows with the text.

• NEW: lock-screen widgets. Add Nihongo Ride to your Lock Screen to see how many words and conjugations are due at a glance -- inline, rectangular, or circular. • Fixed: if you cleared the app's iCloud data, this device stopped uploading and still said "Synced". It now rebuilds your iCloud copy and re-uploads everything. • Fixed: the small widget could say "all caught up" while verb-conjugation drills were still due. It now shows those too. • iPhone polish: a horizon landmark no longer overlaps the word you're typing, and lists no longer slide under the status bar clock while scrolling.

• Fixed: on a second device, your lifetime distance and word count could end up roughly doubled. New installs no longer count another device's rides as their own. • Fixed: the widget, Ride Log and daily reminder could say words were due while the app offered none. Both now agree, and a card is due for the whole day it falls on. • Fixed: turning iCloud Sync off now stops it immediately, instead of finishing an upload and still showing "Synced". • Fixed: a ride where nothing was typed is no longer graded as a clean run. • The road now goes somewhere: as your lifetime distance grows you ride eight stretches from Edo to Kyoto, each with its own sky, light and road surface. The results screen keeps the sky you arrived under. • The ride scene now respects Reduce Motion, and repaints far less often.