Priming effects and prime time processing

The current psycholinguistic literature on visual morphological decomposition makes a great use of the visual masked priming design. Generally speaking, the priming design consists of two stimuli being presented one after the other; the first stimulus is called prime, the second is called target. The experimental task always concerns the target stimulus, and it is typically a lexical decision task, in which subjects are asked to decide whether the target stimulus is an existing word by pressing a response button box. It has been shown that recognition of the target stimulus is faster when it is related to the prime stimulus to some degree: e.g., the target word CAT is recognized as a word faster when it is preceded by a semantically related prime word like kitten than when it is preceded by an unrelated prime word like hand. Visual morphological decomposition is typically investigated by adopting a sub-type of the visual priming design: the masked priming design (Forster & Davis 1984), in which ‘mask’ (that is, a series of hashmarks, #######) is presented for about 500 ms before the prime stimulus, which is usually presented for 30-70 ms. At such a short duration, primes are not consciously recognizable to subjects. The advantage of the masked priming technique is two-fold:

  1. it prevents facilitation effects due to strategic processes triggered by episodic memory; and
  2. it elicits them before full lexical access of the prime word is complete.

The target immediately follows the prime and usually stays on the screen until the subject performs the task on the target (e.g., lexical decision). It is important to acknowledge, however, that the whole enterprise on visual morphological decomposition hinges on the assumption that the masked priming design ensures that the processing of prime stimulus stops as soon as it is visually replaced by the target stimulus on the screen (‘no-overlapping theory of priming’). This assumption is crucial to interpret the priming data as indicative of properties of the mechanism of early decomposition. On the other hand, an alternative theory of masked priming has been proposed, in which the analysis of the prime stimulus do not stop at the presentation of the target; rather, it runs in parallel with the analysis of the target. In this sense, priming arise as a result of the analysis of both prime and target stimuli (‘overlapping theory of priming’; Forster et al. 2003). From this perspective, it follows that the masked priming response cannot be used to explore the mechanisms of early morphological decomposition and, more generally, on early stages of visual word processing. Teasing apart these two theories of priming is essential to validate the feasibility of priming as a probe for morphological decomposition.

One way to assess the contribution of the pre-target-onset prime processing onto the priming response is to implement a backward mask in the masked priming design. The idea is not new (Forster et al. 2003, Morris et al. 2007, Grainger et al. 2003, Holcomb & Grainger2006) and involves the use of a mask (i.e., #####, usually lasting around 20-30 ms) being sandwiched between the prime and the target stimuli (so the presentation of a typical prime-TARGET pair flows as follows: ##### (500 ms) → fuss (30-40 ms) → ###### (20-30 ms) → FUSS). The backward mask is meant to prevent processing of the prime stimulus from overlapping with processing of the target stimulus and therefore allow us to quantify the amount of processing of the prime occurring within the duration of prime presentation. This fact suggests that particular attention will be needed when choosing the duration of the backward mask.} The investigation will need to run a series of masked priming experiments across conditions with different relatedness dimensions (e.g., identity pairs: fuss-FUSS; morphologically transparent and opaque related pairs: boneless-BONE, belly-BELL; orthographically related pairs; canvas-CAN), both with a constant backward mask (backward-mask condition) and without it (no-backward-mask condition). The no-overlapping theory of priming predicts the priming response to be the same for all conditions, regardless of the presence of the backward mask; for example, boneless-BONE should therefore prime in both the no-backward-mask and the backward-mask conditions. The overlapping theory of priming, instead, predicts the priming response to be a function of the presence of the backward mask; for example, boneless-BONE should prime in the no-backward-mask condition, but should not prime in the backward-mask condition.