The data I downloaded had two data columns (excepting the margin of error estimates): Number of Speakers and Number who spoke English "Less than very well". Here are the top five languages other than English spoken at home, along with the English only numbers for comparison. The proportion column represents what proportion of all speakers surveyed each language represents.
|Language||Number Of Speakers||Proportion|
"Chinese" represents people who wrote down "Chinese" as another language spoken at home. They also have reported numbers for Mandarin and Cantonese separately, but there's no way to apportion "Chinese" responses to one dialect or the other.
I wondered whether there was a relationship between how many speakers of a language there were, and how many of those speakers spoke English less than very well. You might think that the larger the available speech community for a non-English language, the less need for English there would be.
The answer (at the national level mind you) looks to be "maybe a little," but there's a lot of variation.