Certain sounds in Japanese are spelled using two kana, kind of like various two-letter compounds used in English that are used to spell a single sound -- technically called "digraphs", things like "th" that uses two letters to spell either voiced //ð// as in "the" or voiceless //θ// as in "thing", or "ng" that uses two letters to spell the //ŋ// as in "thing".
In English, digraphs aren't visually presented specially -- you just have to learn them.
In Japanese, digraphs write the second kana -- the part expressing the vowel value -- smaller than normal. For example, here's a comparison of regular-sized kana and the smaller versions used as the second kana in digraphs:
The name for two-letter combinations in Japanese that use the small "y" or small "w" kana is 拗音【ようおん】 (yōon). You won't usually run into the small "w" variants anymore, as those sound combinations died out in mainstream Japanese some time in the last 100 years or so. Sometimes you'll still bump into them, such as in foreign borrowings or manga-style sound effects.
The "y" combinations you'll run into normally are:
Notice that the first kana, the regular-sized one, is always from the "i" column. This is because the vowel portion starts with the semi-vowel sound "y", which is basically a kind of bendy form of "i". The pronunciation of ちゃ, for instance, is a little like "chiya", where the "i" is abbreviated so much that it basically disappears. Sometimes you'll hear speakers say "chya" in formal or careful speech. In everyday speech, it comes out more like "cha".
(The "w" combinations are always spelled with the regular-sized kana from the "u" column, as the semi-vowel "w" sound is basically a bendy form of "u".)
The combinations that start with a fricative (a consonant that is pronounced with a kind of friction, like "sh" or "ch" in English), specifically the ones with し and ち, often lose the "y" value entirely in everyday speech, so しゃ sounds like "sha" and ちょ sounds like "cho". The others all keep the "y", so にゃ sounds like "nya" (similar to how children taunt each other in the US), and きゅ sounds like the name of the letter Q.
Also, note that each of these yōon digraphs is pronounced as one mora, or one "beat" in Japanese. ちや with a big "や" is pronounced as "chi ya", with two beats, while ちゃ with a small "ゃ" is pronounced as "cha" with just one beat. The mora count is a very important feature of Japanese, and you'll hear this very clearly if you listen to Japanese songs or poetry.
Hope this helps! Please comment if anything above is unclear or incomplete, and I'll update.