How to Pronounce a Word | Free Ways to Check Single Words and Sentences

You run into a word you do not know and you cannot tell how it is said. There are several places to go: a dictionary, a site collecting native speakers' recordings, a site that finds the word inside real video, the read-aloud button in a translation service. All of them are tools for checking how English is pronounced.

But which one to open is not a matter of taste. It is settled almost entirely by the kind of word. A dictionary only returns words that have been entered as headwords. Large place names are sometimes there, but people's names, company names and new product names usually are not. Conversely, when what you want to know is which part of a sentence to stress, hearing the word on its own will not answer it.

So the mapping between what you want to check and where to go comes first. After that, we’ll look at what each of the five tools actually returns, according to their official pages.

The short version. Words that are in the dictionary: Cambridge Dictionary. People, places and product names: Forvo. Stress and linking inside a sentence: YouGlish. Turning long text into audio: NaturalReader. A rough check on whether you get through: Google Translate. Between the 5, most situations are covered. And all of them give you the correct sound; none of them can tell you whether your own pronunciation matches it. That difference is what the last section is about.

Key takeaways

• Which one you open is not a preference. It is decided by the kind of word you want to check
• Words that are in the dictionary: Cambridge Dictionary. You get the transcription plus both British and American audio
• People, places and product names: Forvo. Real speakers recorded them, so words missing from dictionaries turn up
• Stress and linking inside a sentence: YouGlish. You can find the word inside real video
• Long text you want as audio: NaturalReader. You can build a model out of your own material
• A rough check on whether you get through: Google Translate is enough
• Every tool here returns the correct sound. None of them returns whether your own sound matches it

The kind of word decides where you go

Each of these tools collects its sound in a different way. A dictionary holds edited headword audio, Forvo holds recordings its users contributed, YouGlish goes and finds the sound inside video. Different collection methods mean different strengths. Here is the mapping.

The bottom row of that table is the thing most worth taking away. Everything above it tells you what the correct sound is. None of it tells you how far your own sound sits from that. Listening and comparing are two different jobs.

Dictionaries — the symbol and the sound in one place

If the word is in the dictionary, If the word is in the dictionary, looking it up there is the fastest and most reliable way to learn it. The reason is that you get the transcription as well as the audio, and once you can read the symbols, you can read the next word with the same pronunciation without looking it up. If you learn a word by ear alone, you have learned exactly one word.. The reason is that you get the transcription as well as the audio, and once you can read the symbols you can read the next word of the same shape without looking it up. Learn a word by ear alone and you have learned exactly one word.

Cambridge Dictionary plays British and American English side by side. The official site says this audio is not synthesised but recorded by real people, and that every word from CEFR A1 through C2 is covered. In other words, for anything a learner is likely to meet, assume it is in there.

One thing worth looking at while you are in the dictionary is how the pronunciation shifts with part of speech. Some words spelled identically are stressed on a different syllable as a noun and as a verb, and the dictionary lists those separately. Since a shift in stress can make a word sound like a different word, glancing at the stress mark inside the transcription while you listen pays off later.

Forvo — for the words the dictionary did not have

Somebody's name, the name of a town, a company, a product. If you use English at work, these are the ones that actually cause trouble. They are often missing from dictionaries, and they are exactly the kind of word where guessing from the spelling goes wrong.

Forvo calls itself the pronunciation dictionary on its own site, and that is what it is: a collection of native speakers' recordings of words. It does not score. Its value is that you can hear the same word from different speakers. A dictionary settles on one sound and presents it, but in practice not everyone says it that way. Play several recordings in a row and you can see what is shared and what is down to the individual.

For proper nouns, picking a recording by a speaker from that place is the safe move. Place names can be said one way locally and another way in the English-speaking world, and which one you want depends on who you are talking to. With somebody from there, leaning towards the local pronunciation lands better. In something like an international meeting, the form current in English can cause less confusion.

YouGlish — stress and linking inside a sentence

You have the word right and the sentence still does not get through. This is not unusual. In English, neighbouring sounds join together and drop out, so you can pronounce each word correctly, join them into the wrong shape, and still not be understood.

YouGlish shows you how the word is actually pronounced in real videos, rather than in a dictionary recording. You can narrow your search by part of speech, speaker gender, or a specific person, so you can see how the same word changes depending on the situation. It includes both single words and whole phrases., rather than in a dictionary recording. You can narrow by part of speech, by speaker gender and by a specific person, so you can see how the same word shifts with the situation. It takes single words and whole phrases.

The knack when using it is not to decide from the first clip. Play 3 to 5 in a row and take out what they share. Copy one person exactly and you take their habits along with the word. Extracting the common elements from several recordings makes them easier to retain.

Turning your own material into sound

Everything so far has been about finding sound that already exists. The other kind of tool turns text you supply into sound. Set phrases you use at work, or what you want to say in tomorrow's meeting, are sentences whose audio does not exist anywhere, and they belong here.

NaturalReader reads English you type and NaturalReader reads English you type and turns it into audio.. It does not score. The official site states a Free forever plan and No credit card required, and long documents and PDFs can be turned into audio as they are. It runs in the browser, as a phone app, and as a Chrome extension.

What this kind of tool returns, though, is synthesised speech. It can differ slightly from how a person would actually say it, and the gap shows most in sentences carrying emotion or emphasis. For checking set phrases it is more than enough, but keep it as your only model and your speech drifts towards the machine's.

You can start without being able to read the symbols

You look a word up, cannot read the transcription, and end up writing the sound down in the letters of your own language instead. It is a common dead end. But you do not have to learn every symbol before you start. The sounds that give you trouble are fixed, and there are usually only a handful of them.

Look at the symbols for the words you actually looked up, alongside the sound. Do that repeatedly and the frequent symbols become readable in order of frequency. Vowel symbols are numerous and hard to hold at first, so starting from the consonants and the stress mark moves faster. The stress mark is the simplest thing in the transcription to read, and it bears directly on whether you get through.

Writing an English pronunciation out in the letters of your own language is convenient as a way in, but stopping there causes trouble. A sound you learned by substituting the nearest sound your own language has stays learned as that substitute. Inside your own head it is correct, so there is nothing to notice until somebody points it out.

Should I use British or American English, and which one should I check against?

Either is fine, but deciding before you look things up is easier. A dictionary puts both in front of you, so you can listen to the two and set your own reference. Mix them and you end up saying the same word differently on different days, and you Either is fine, but deciding before you look things up is easier. A dictionary puts both in front of you, so you can listen to both and make your own choice. Mix them and you end up saying the same word differently on different days, and you lose track of which one was right..

As a rule of thumb, use the variety of English spoken by the people you speak to most. If the people you work with are American, use American English. If you deal mostly with Britain or Europe, British English. If you cannot decide, matching the side used in your coursebooks or your exam makes material easier to come by.

Either choice gets through, incidentally. Most of the people using English are not native speakers, so it is not the case that you have to commit to one side to be understood. You decide in order to stop hesitating.

Understanding it and being able to say it are not the same

Everything so far exists to get you the correct sound. Once you have it, whether you can produce it is a separate question that stays open. And these two are hard to tell apart from the inside.

In second language research, the moment a learner notices the difference between their own version and the correct one In second-language research, the moment a learner notices the difference between their own production and the correct form is seen as the first step toward acquiring that form. Put another way, a sound you hear without noticing it will not be acquired. Listening to a model and simply thinking that it makes sense will not change what your mouth produces.. Put the other way round, a sound you let pass without noticing does not go in. Playing a model and thinking that makes sense does not change what your mouth does.

This is the limit of the tools for checking. They can play the correct sound, but they have no mechanism for directing your attention to the difference between that and the sound you are producing right now. To notice a difference, your own sound has to be present. Either record yourself and listen back, or use something that returns the gap.

The awkward part is a way of saying something that has already set. A word you have said the same way for years does not change just from hearing the correct version. That learners build a system of their own, different from both the target language and their first language, and that parts of it stay put however long study continues, has been argued for half a century.

And prising something loose once it has set depends on the kind of intervention. An analysis pooling classroom corrections reports that making clear what is different, and how, has a larger effect than saying the correct form back for the learner to hear. That is why playing a model is not enough on its own.

So tools for checking and tools for comparing your own sound have different jobs. The first returns the correct sound; the second returns the gap between that and yours. To fix your pronunciation you need both. The second kind is generally called a scoring tool, and it sits on a separate shelf from the checking tools.

Straight from the search box

You do not have to open any of these to hear a word. Type the word with pronunciation after it into a search box and it plays. It is faster than opening a dictionary and needs no account. For a single word, that is often all you need.

The feature that listens to you and grades you is a different matter. It is a limited beta, restricted by country and by language pair, and the pairing it runs on is practising a language other than the one you already speak. There is no pairing in which an English speaker practises English, so that side is not available to you. What is available everywhere is plain speech-to-text: say the word, read back what it typed, and you learn whether the machine heard the word you meant. That is a pass or a fail, not a score.

What to do after you have checked

Once you have the sound, saying it out loud 3 times on the spot makes it stay. Listen in silence and move on, and you will be looking the same word up again next time you meet it. Saying it aloud also sometimes catches, right there, the gap between it and the sound you had in your head.

Write down the words you checked and come back to them a few days later. Coming back briefly, many times, with gaps in between, stays with you better than setting aside one long block. 5 minutes at a time is fine.

And a sound you verified on a single word is easy to lose inside a sentence. If you can say the word but not the sentence, the cause is not the sound itself but the joining or the rhythm. When you get that far, it is time to move from tools that look up words to tools that work at the level of sentences and conversation.

You played it but still could not make it out.

You opened the right place and the sound itself is still not clear. That is usually not the tool. It is happening on the playback side. The first thing to suspect is your speakers. Built-in laptop and phone speakers are weak at the low and the high end, and fine consonant differences get flattened. Pairs like s and th, or f and th, where the difference is small to begin with, take the worst of it. ★How much any given device flattens has not been measured here, so this is not a claim that you cannot tell them apart. Switching to earphones and playing the same word again is the fastest way to separate a tool problem from a playback problem.

Speed is next. Native recordings are not slowed down for learners, so not catching it on the first pass is normal. If the tool lets you slow it down, slow it down. If it does not, repeat the same stretch. With the kind of tool that finds the word inside video, listening to the second on either side of the word shows you the sounds that are joining and disappearing.

If it still will not come, look at the transcription. Sounds your ear cannot separate are clearly different when the symbols differ. Listening again with the symbols as a handle can change what you hear. The reverse order, listening first and then reading the symbols, makes you read the symbols the way you already heard them, so looking at the symbols first is safer.

Increase the number of words you can read without looking them up you can read without looking them up

English spelling is often said to be irregular in its relationship to sound, but not all of it is. Most words can be read by rule. The more words you can read by rule, the fewer times you look things up, and the more of that time goes to the words that genuinely need it.

The endings are what pay off fastest. Words ending in -tion all read the same way; the past tense -ed settles into 3 forms according to the sound in front of it; the consonant c takes an s sound before e, i and y. There are only so many of these, so learning them removes whole groups of words from the list you have to look up.

Even so, learning the rules never takes the count to zero. The exceptions cluster in the words used most, and the shorter the word the more likely it is to be one. Words that come up daily have been in use longest and have kept older spellings. The words you assume you can read by rule are exactly the ones worth looking up once to be sure.

Pick one place to keep what you checked

Use a word once and drop it, and you will look it up again next time you meet it. In practice the words people look up twice are a fixed set, and the same ones come round repeatedly.

Pick one place. A vocabulary app, a paper notebook, a notes app; the type does not matter. What matters is not splitting it. Split it in two and you start by working out which one you wrote it in, and before long you stop writing at all.

There is a knack to what you write. Sound does not survive in letters, so writing the spelling alone will not let you reproduce it later. Write down where you tripped as well: whether you had the stress in the wrong place, whether the vowel quality was different, or whether you simply did not know the word at all. That one line tells you what to check when you come back to it.

Summary — which one to open, by kind of word

• A word that is in the dictionary → Cambridge Dictionary (the transcription and both British and American audio at once)
• People, places, product names → Forvo (recordings by real speakers; you can hear the same word differ between them)
• Stress and linking inside a sentence → YouGlish (finds and lines up real video)
• Long text or your own material → NaturalReader (reads English you type and makes the model)
• A rough check on getting through → Google Translate (see what voice input heard; it does not score)
• Whether your own sound matches → none of the above can tell you. That needs a separate tool, one that returns the gap

Worth remembering

• Where you look is not a preference. It is decided by the kind of word
• Words that are not in dictionaries, such as people, places and product names, will not appear however long you search a dictionary
• You do not have to learn every symbol. Look at the symbols for the words you actually looked up, alongside the sound
• Starting from the consonants and the stress mark moves faster. Stress bears directly on whether you get through
• With YouGlish, do not decide from the first clip. Play 3 to 5 and take out what they share
• Respelling in your own letters is convenient as a way in, but stop there and the substituted sound is what stays
• Decide between British and American before you look things up. Mix them and you lose track yourself
• Checking gets you the correct sound. The gap between that and your own does not come back
• A way of saying something that has set does not come loose from hearing the correct version alone
• Say a word you checked out loud 3 times on the spot, and come back to it a few days later