The following is a comparison of ABA/VB and some applications of discrete trial teaching (DTT). I suggest you read both Teaching Language to Children with Autism Or Other Developmental Disabilities by Sundberg and Partington and The ME Book by Lovaas for complete descriptions. These are generalizations to a point, but the basic concepts are valid. However, it should always be remembered that the choice of techniques, prompts, curriculum, etc., should all be determined by the CHILD, not by labels. This is the same science, different applications. For many students, a balance of both DTT and NET are necessary for optimal learning.

It should also be stated that any good program is completely individualized for the student. There are great ABA/VB programs and poor ones, just as there are great DTT programs and poor ones. Generally the thing that makes a program great is that the teaching is shaped by the student's learning and not by a preset format.

Applied Behavior Analysis (with VB)
Some Discrete Trial Teaching (DTT)
Teaching is done both in formal/table settings and in the natural environment. Teaching is done mainly in formal/table settings and generalized to other environments.
The function of inappropriate behaviors is determined through functional analysis, with interventions chosen to target the function of behavior. Special attention is paid to the MO/EO (motivation) of the student and how that can be manipulated to both decrease the value of escape and stim and to teach functional replacement behaviors. Inappropriate behaviors are also often addressed with interventions based on functional assessment and are directly targeted through extinction and other correction procedures. However, attention is not necessarily paid to the matching law and its role in undesired behavior.
Students are "eased" into instruction, since teachers devote time to pairing themselves with reinforcement, gradually interjecting more demands into the situation. Students are sometimes worked through an "extinction burst" during the onset of instruction, when they are escalating escape behaviors prior to learning the contingencies of the teaching situation.
Most teaching is initially done in the natural environment, with table time gradually introduced. Much teaching is done at the table, with time in the natural environment more limited or sometimes used as "breaks" from instruction.
Manding is a major focus of instruction and capturing and contriving MOs/EOs, both in formal and natural settings, sets the stage for teaching opportunities. Language is first taught through mands and echoics. Manding is sometimes considered an advanced language program and is not initially addressed. Language is first taught through verbal imitation.
Teachers generally use errorless learning procedures to increase rate of reinforcement, decrease value of escape, and indirectly eliminate inappropriate behavior. Quick prompting, fading, and transfer procedures result in low or no rates of student errors, as well as fluent responding and lack of prompt dependence; most-to-least prompting. Teachers sometimes (though not always) use a "no-no-prompt" procedure in which the student is given opportunities to respond (and potentially to make several errors) before prompts are delivered, often resulting in increased student errors and escape behaviors; least-to-most prompting. However, errorless learning procedures are also used in some cases.
Instructional targets are mixed from the outset, teaching various functions and classes of operants and skills simultaneously. Teaching is a moment-to-moment enterprise, so teachers are required to identify and contrive teaching opportunities "on the fly," drawing from a knowledge of the science. This results in a dynamic form of teaching that is constantly shaped by the student, the environment, and the teachers' skills and creativity. The sequence of instruction often progresses from mass trialing (isolation) of one target to introduction of one or more distracters, to random rotation; skills are generally taught separately.
Skinner's (1957) functional analysis of language guides language instruction. Language instruction is not necessarily guided by Skinner's analysis of verbal behavior.