Center embedding in linguistics is a phenomenon where one phrase is placed, or embedded, within a larger phrase or sentence. Different languages accommodate this construction in various ways, but many of them allow for instances where a smaller, or more precise, unit of speech can be included in a fuller sentence. When this phrase is integrated whole into the larger one, it is often referred to as center embedding.
One of the most common examples of center embedded phrases involves relative clauses that get injected into larger sentences. One basic example is a sentence like this one: “The man that the woman heard left.” – In this example of center embedding, if the relative clause was entirely taken out, a shorter sentence might read like this: “The man left.” The inclusion of the relative clause serves to show the reader that there was a woman who heard the man, and this leads to the appearance of the phenomenon of center embedding in the sentence.
The rhetorical technique of center embedded speech or writing often creates problems with language comprehension. In the above example, it can be confusing to readers or listeners to figure out exactly what is meant, as the context of the woman hearing the man is rather unclear, especially when read silently, without any hint from the speaker's intonation. Some might question whether the woman heard the man leave, or rather, simply heard him in some other context. In addition, multiple instances of center embedding can create confusion about subject/object relationships, and make a longer monologue much more difficult to understand. Many academics who teach writing would counsel students to avoid excessive practices of center embedding in writing.
In addition to the difficulties with center embedding in actual language comprehension between humans, there are other problems with this phenomenon that relate to the use of technology to understand natural language. Experts in using today’s technology to interpret human speech have pointed to significant issues with linguistic techniques like center embedding that inhibit effective parsing of the language for purposes of accurate interpretation. Although complex modeling can assist in making machines more able to deal with complex speech, some of today’s top authorities on linguistics have generally contended that most languages are somewhat resistant to parsing, indicating that building technological ability for natural language processing will be a long, complex, and likely imperfect process.
New resources like simple recurrent networks and devices for lexical category distinctions have helped scientists to get closer to a deeper understanding of how to interpret natural language. Part of this has involved the study of center embedded structures and other alternative sentence structures. There’s also been an extensive effort to create sophisticated modeling for subject and object relationships.