A computer adjusts digital content to match a user language proficiency level. The computer receives digital content including words and monitors word consumption activity of a user. The computer identifies troublesome words associated with content consumption delays and establishing a user language proficiency level based on a difficulty value associated with said troublesome words. The computer identifies content words having difficulty level above the language proficiency level. The computer identifies from a word corpus, exchange candidate words having a difficulty level equal to or below said language proficiency level. The computer generates using a natural language processing (NLP) algorithm, word embeddings for said target words and said exchange candidate words. The computer selects from the exchange candidate words, replacement words having word embeddings substantially the same as the word embeddings of said target words. The computer adjusts content by replacing the target words with the replacement words.