How many words do you need to get by as a tourist?

Budget about 50 phrases for each situation you expect to handle on your own, and most trips of one to two weeks come down to three such situations. That is roughly 150 lines. The first 20 of them will carry more of your week than the hundred that follow.

Notice that it is a phrase count. Words are the unit of a dictionary, not of a counter in a train station. In raw vocabulary, 150 travel phrases are smaller than they sound, because the scaffolding repeats: please, I would like, how much is it, where is, two of these, sorry, again but slower. Build 150 lines on six or seven frames and the distinct word list behind them runs to a few hundred entries, most of which you will only ever recognise rather than produce.

Why the numbers you have seen are bigger

Vocabulary-size targets answer a different question. They estimate how much of a language will make sense to you when it arrives unannounced: a newspaper column, a podcast, a stranger with an opinion. That is a comprehension goal, and it genuinely is a thousands-of-words problem.

Yours is narrower and far more mechanical. You need to produce a small, predictable set of utterances out loud, at speed, in perhaps eight recurring moments, and understand enough of the reply to act on it. The two projects have almost nothing to do with each other, which is why a year of general study can still leave you pointing at a pastry.

A word in a list is not a word in your mouth. Under even mild pressure, three things fail at once: retrieval, which is finding the word while someone waits; assembly, which is putting it in an order and an ending the local ear accepts; and pronunciation, which is making the whole thing land as sound rather than as spelling read aloud.

A memorised whole utterance sidesteps all three. You are not building the sentence at the counter, you are recalling it. Knowing the noun for bill is not the same as having the line that asks for it, and the gap between those two states is exactly where people decide they are bad at languages. They are not. They have been studying the wrong unit.

Production is small, comprehension is not

You control your half of the exchange and nothing of the other half. The reply comes back at a speed you did not choose, in a register nobody adjusted for you. So your 150 lines are a production budget, and you buy comprehension separately with two cheap things: a few repair lines that slow the other person down, and the willingness to point, show a screen, or hold up fingers.

This is also why counting words flatters you. You will pass a vocabulary quiz on 400 words and still stall on the one sentence you needed. The honest test is whether you can say the line out loud, from cold, without a pause to assemble it.

Write out your actual trip and you will find it has far fewer language situations than days. Nine days somewhere is not nine days of talking. It is perhaps thirty transactions of six or seven types, repeated.

A typical week produces something like this:

• Ordering and paying in a cafe or restaurant, several times a day
• Buying a ticket, asking which platform, asking whether this train stops at your station
• Checking in, asking for the key again, reporting that the shower is cold
• Buying something, asking the price, asking for a different size
• Asking directions and then understanding roughly half of the answer
• The occasional unplanned moment: a pharmacy, a lost bag, a closed road

Most of that list repeats the same five or six structures. You are not learning thirty situations, you are learning six, then saying them again on Tuesday with a different noun.

The arithmetic is therefore simple. Take the number of situation types you will genuinely face alone, multiply by the lines each one needs, and add a fixed repair kit. For most travellers that calculation lands between 60 and 150, which is why the 50 per scenario figure is a reasonable planning unit rather than a round number picked for the look of it.

If you only take ten lines abroad, take these ten. They are the ones that convert a stalled exchange into a slow successful one, and they work in every situation on your list.

• An apology and a declaration: sorry, I speak only a little
• The permission question: do you speak English, asked in their language first
• The slowdown: again, more slowly, please
• The number rescue: can you write it down, or show it on the screen
• The confirmation: yes, that is right, or no, sorry
• The pointing line: this one, please
• The quantity line: two, three, a little, no ice
• The price question, and the ability to recognise the answer
• The location question: where is the toilet, the station, the exit
• A proper thank you, delivered like you mean it

Notice how little vocabulary that is and how much of a trip it covers. Learning these first also changes the mechanics of the rest of the week, because an opener in the local language, even a clumsy one, tends to buy patience, and patience is what every line you have not learned runs on.

The binding constraint is not your memory, it is the number of days between now and the flight. A phrase met once on a Tuesday is not a phrase you own on Saturday. It needs to come back at spaced intervals, each one a little longer than the last, and each return has to be an attempt to say it rather than a glance at it.

That is what a spaced repetition scheduler does. Phrazl uses SM-2, the classic interval algorithm, with one change that matters for travel: the intervals are compressed against your departure date, which is treated as a hard deadline rather than a preference. Ordinary spaced repetition is happy to push a phrase six weeks out, because it assumes you have forever. You do not. If you fly on the 19th, a review scheduled for the 24th is worthless, so the schedule tightens as the date approaches and the final days get denser.

So treat the size question as a calendar problem rather than a memory one. Nobody can tell you how many attempts your particular mouth will need on a particular phrase, and we are not going to pretend to a figure for how many minutes a day this costs you. What is reliable is the shape of the trade. Every line needs more than one spaced attempt, the number of attempts that fit between today and your flight is fixed, and the list is the only part of that equation you control.

Hold the list small and each line gets its repeat attempts inside the time you have. Push the target into the many hundreds and the calendar does not stretch to meet it, so the schedule either thins the repetitions per phrase or runs past your departure date. The failure mode is arriving having met a lot of language and owning very little of it. Fewer lines, properly spaced, beat a longer list skimmed once.

The date is the one field worth checking twice

Because the whole compression hangs off your departure date, it is the setup answer to get exactly right, more than the destination and more than the scenarios. Setup is three steps, destination, date and scenarios, and that flow is where a trip gets pointed at a date. On the free preview you hold one trip at a time, so running setup again re-points the trip you already have rather than starting a clean second one. If your flight genuinely moves, re-point the trip against the real date, and plan on re-covering ground rather than assuming every phrase keeps its place in the schedule. Re-pointing at a different country is a bigger act again, because it changes the language the set is drawn from, and the coverage figure starts over from zero. Change the country only if you have genuinely changed country.

The number on the screen is coverage, not a score

There is one progress bar, labelled as the share of your trip's curriculum you have covered, and there are no streaks and no XP behind it. That is deliberate: a deadline already supplies all the urgency a six week run needs, and a daily counter mostly measures attendance. We have taken that argument apart properly on the sibling page about whether streaks are bad for learning, so it is not repeated here.

A phrase count only means something if the lines leave your mouth. Reading a list builds recognition, and recognition is the thing that collapses when a waiter is standing there. So the rehearsal has to be spoken, from cold, at something near the speed a real exchange runs at.

Phrazl gives each phrase three ways through. Flashcard, where you say the line before you see it. Listen and repeat, where you hear the phrase and copy it back. Fill in the blank, where one element of a line you half know is missing and you have to supply it. Every phrase plays at speed or slowed down so you can copy the shape of it, and each one carries a written pronunciation guide for the sounds your mouth has not made before.

Where the modes have their edges

The voice on every phrase is synthesised rather than a native speaker, and the app labels it that way on the screen instead of hoping you will not look. What that does and does not let you copy is an argument of its own, and we have it out on the sibling page about whose voice the audio is, so it is not repeated here.

Fill in the blank has two limits worth knowing before you lean on it. It splits a phrase at the spaces, so it never blanks a word at all in Japanese, Chinese or Thai: if you are flying to Tokyo, Shanghai or Bangkok, plan on flashcard and listen and repeat doing the work and spend the third mode's time saying the line again. And the correct option tends to land at the end of the list rather than anywhere in it, so grade yourself on whether you could have produced the word, not on whether your thumb found the right box.

Either way, the measure to hold yourself to is the same one as the count itself: not how many lines you have seen, but how many you can produce out loud without a pause to assemble them.

Two travellers learn 60 phrases each. One learned the 60 most useful for their destination and scenarios, the other learned 60 in whatever order an app served them. Those are not comparable weeks.

So rank before you count. Phrazl's phrase sets are pre-curated per language and scenario and ordered by trip relevance, so the top of the list is the material you would miss first, not the material that happens to be grammatically simplest. The two orderings diverge quickly. A beginner curriculum opens with colours and family members. A trip curriculum opens with asking for a table, asking the price, and saying that you have an allergy.

What belongs in your first twenty

• The three lines that start a transaction, because the opening sets the whole exchange
• Anything involving money or quantity, where a misunderstanding costs you something
• The repair lines, because they rescue every phrase you have not learned
• The one line specific to your trip: a dietary restriction, a station name, a medication

What does not belong in the first twenty: pleasantries beyond thank you, anything about the weather, and every noun you could point at instead.

Three scenarios to a trip, which is where 150 comes from

Setting up a trip takes three things: the destination, the departure date, and up to three scenarios chosen from restaurant, transport, hotel, emergency and shopping. Three is a cap on one trip rather than a cap on you, and at 50 phrases a scenario it is exactly the 150 this page keeps arriving at, which is about as much as a six week run will carry. A trip refuses the fourth pick rather than stretching to hold it.

If your trip genuinely wants four or five, it is worth knowing how the trips themselves behave before you plan around it. A trip pack unlocks one trip, which means a second trip to the same destination starts again at the free 20 and is offered the pack again. The honest route to more than three scenarios, or to more than one trip on the go, is the full bundle: every destination, all five scenarios, and new trips set up any time with no further purchase, with each individual trip still taking three scenarios at a time.

The destinations the sets are built for

Phrazl covers ten travel languages, chosen as destinations rather than as a language list: Portugal, Spain, France, Italy, Germany, Japan, China, South Korea, Thailand and Greece. Dutch is not among them, so if you are flying to Amsterdam this is not your tool yet, and we would rather say that here than let you find out at the destination picker.

Be clear with yourself about what the number is. A hundred and fifty rehearsed phrases is a coverage figure rather than a guarantee: it describes how much of your trip's likely language you have practised out loud, and no honest progress bar can promise more than that. Inside that coverage, the thing you are aiming at is ordinary transactions you can open and close in the local language, with enough of the reply recognised to act on it.

What it will not give you is conversation. It will not carry you through a chat with your host's father-in-law, a dispute with a landlord, or a medical appointment with any nuance. If a situation turns genuinely complicated, the right move is an interpreter, a phone translation app, or someone bilingual, not the fourteenth phrase in your transport set.

The emergency set deserves the same honesty. Those lines are for general communication practice only and are not a substitute for contacting local emergency services or a qualified interpreter, which is stated on the scenario selection screen and in the terms rather than buried.

And expect the first attempt to miss. You will get the stress on the wrong syllable, they will reply at full speed, and you will reach for the slow-down line. That is the system working rather than failing. The second attempt, delivered slower, is the one you are rehearsing for, which is why the repair lines earn their place ahead of another noun.

Trips move. Three weeks becomes nine days, and the sensible response is to cut the target rather than skim a bigger one.

The triage, in order

• Make sure the trip holds the real date. The compression works against the date the trip was set up with, not the one in your head, so if the flight moved, run setup again against the new date before you start cutting lines.
• Pick one scenario. The one you cannot avoid. For most people that is restaurant or transport. Learn its top lines properly and handle everything else in English or by pointing.
• Take the repair kit anyway. Ten lines, no exceptions. They are the highest return per minute of anything on this page.
• Practise out loud, not in your head. Silent reading rehearses recognition. You need production, which means your own voice at something close to conversational speed.
• Front-load the awkward sounds. Nasal vowels, rolled consonants, tones. If a sound is new to your mouth, it needs more attempts than a new word does, and no amount of reading will substitute.
• Let the schedule compress. With a near departure the intervals shorten rather than the list growing. Twenty phrases reviewed daily for a week beats sixty seen once each.

Twenty well-drilled lines is not a consolation prize. It is a visibly different trip from arriving with none.