Which of the following statements about probing language models for syntactic structure is true?

Group of answer choices

There is enough information in the vector representations of words in a sentence to decode $($using a linear transformation$)$ the distances between pairs of words in the syntactic structure.

Large language models can only produce a vector representation for a sentence if that sentence appears in the Google n$-$gram corpus.

The probability assigned to each word in a sentence by a language model is directly proportional to its distance from the verb in the syntactic structure.

Language models have not been shown to contain any information about syntactic structure, because they are only trained to predict the next word.