This is usually because a translation is missing. If you define a language for one UI element (e.g. label::English instead of just label), you need to make sure all UI elements including media also define a language. You can read more at https://docs.opendatakit.org/form-language/.
From a quick look at your form, I think what is happening is that the default language defines images and English defines text. If you change the language to English, you should see the text as expected.