The difference in one line
Phonics teaches a child to work a word out from its letters. Sight words are words recognised whole, without working them out. One is a method; the other is a small set of exceptions.
Why the argument exists
For decades some schools taught reading mainly by having children memorise whole words and guess from pictures and context — 'whole language', later 'balanced literacy'. It produces children who look like readers early and then stall around age seven, when the books contain words they have never seen and pictures stop giving the answer away.
Decades of research now point clearly to systematic phonics as the effective approach. That is settled; the practical question is what to do about the genuinely irregular words.
Why some words still have to be memorised
English has words that simply do not decode: the, was, said, one. They are also among the most frequent words in the language, so a child cannot read a single real sentence without them. They must be learned by sight — not because sight-reading is a good method, but because these particular words leave no alternative.
The approach that works: heart words
Do not treat an irregular word as one unreadable lump. Take said: the s and d are perfectly regular — only the ai is odd. Mark just that part as the bit to know 'by heart' and decode the rest as normal.
This keeps decoding as the default strategy even for exceptions, and it dramatically reduces how much a child has to memorise.
