Text to image, prompts including 2 words or more are always misspelled and butchered.

When using text to image and correctly prompting in numerous attempts, simple to advanced prompts, it always generates the image with incorrect spelling if involving 2 words or more