Scoring Referential Activity (RA)

This page concerns the scoring of RA by judges.  For information about computerized scoring using the Weighted Referential Activity Dictionary (WRAD) please visit the page below:

RA was initially measured using ratings by judges on four scales: Concreteness, Clarity, Specificity, Imagery, based on Strunk and White’s Rule 12, which reads as follows:

Prefer the specific to the general, the definite to the vague, the concrete to the abstract... If those who have studied the art of writing are in accord on any one point, it is on this: the surest way to arouse and hold the attention of the reader is by being specific, definite, and concrete. The greatest writers—Homer, Dante, Shakespeare—are effective largely because they deal in particulars and report the details that matter. Their words call up pictures... In exposition and in argument the writer must likewise never lose his hold upon the concrete; and even when he is dealing with general principles, he must furnish particular instances of their application. (1972, pp. 16-18).

When scoring referential activity language is first segmented into "idea units" by trained judges and then scored on each of these four scales. Each idea unit is given a score between 0 and 10 on each scale; the four scores are averaged to achieve the RA score.  What follows are brief definitions of these scales as defined for judges' by Bucci, Kabasakalian-McKay & the RA Research Group (1992); See the current RA Manual (2004) for the current scoring guide.  For complete definitions and scoring criteria please see, Bucci, et al. (1992).  Following the scale definitions are descriptions of the rules for segmentation and of computerized measures that have been developed to score RA.


Concreteness reflects the extent to which verbal expressions refer to sensate properties of actual things or events or to anything that is experienced as sensation or feeling. This may include reference to imagery in any sense modality; somatic or visceral experience; or representations of motoric activity, i.e., any components of the nonverbal processing systems.

The following excerpt from a poem by Wallace Stevens was judged as highly concrete on this scale.

The red-ripeness of round leaves is thick.
With the spices of red summer.
But she that he loved turns cold at his light touch.


In general, the Clarity dimension may be understood as reflecting the speaker's awareness of the communicative intent of discourse; the degree to which the speaker appears to be aware, on some level, of the perspective of the listener, and to be attempting to bring the listener to where she is.

The following segment was scored high on the clarity dimension.

One day I was sitting in class and I had my T square - I was taking mechanical drawing. I took a rubber band and made a cross bow out of it. I had a paper wad and I'd been shooting 'em out the window. I was standing there, and I had a lot of tension on the rubber band, I mean I really stretched it - it was about that thick. I really stretched the thing and it slipped. (Bucci et al. 1992, p.29)


A highly specific text is detailed and informative, refers to particular objects, persons, places, times; specifies precise quantities; describes the subject or object of the discourse, whatever that may be, in detail.

The following is a segment that was scored as highly specific on this scale: 

Just before I went to sleep on the first night, I had a good look around the room, opening all the drawers and cupboards I could see. I hadn't expected to find anything, but in the last drawer, which was the one in the bottom of the wardrobe, I came across a quite amazing collection of old junk. There were old broken dirty shoes, a greasy old recipe book, bottles half full of over-the-counter medicines, a large moth-eaten embroidered pincushion, a lot of shoe trees, some plugs and bits of wire, and two beer bottles. I couldn't have been more surprised if I had really found a skeleton. (Bucci et al. 1992, p.28)


This refers to the judge's overall impression of the vividness and effectiveness with which the speakers' language is reflecting and capturing imagery or emotional experience, in any sense modality.

The segment below was scored as high on the imagery dimension.

He was a hard-working man. He left home very early in the morning. I remember in the winter he left home while it was still dark, and it was so cold. And he arrived home exhausted, haggard, sweaty, uncommunicative. (Bucci et al. 1992, p.31)


Prior to applying the scales, texts need to be segmented into units, such as the segments above, that can be scored on these dimensions. The size of unit depends on the research purposes and may range from major theme units to  idea units at varying levels of precision. Units that are judged to capture a single "shot" or "frame" of a narrative are usually most suitable for RA scoring. Raters can often recognize idea or theme unit boundaries without defining explicitly the basis on which the judgment was made. The knowledge of
what an idea is, and where one ends and another begins, is a part of linguistic
competence, that is shared by speakers of a language and applied intuitively without explicit awareness of the underlying principles or rules.  Detailed instructions for segmenting narratives and assessing reliability of segmentation are available in Bucci et al. 1992.

Computer Scoring

The original empirically derived computer model of this scoring was developed by Mergenthaler and Bucci (1999); this was modified and enhanced so as to include a dictionary with variable weights by Wilma Bucci and Bernard Maskit (2006). The basic structure of this model is contained in the Weighted Referential Activity Dictionary (WRAD), which contains 697 highly frequent items, primarily function words, accounting for about 85% of spoken language. Texts are scored using this dictionary by the Discourse Attributes Analysis Program (DAAP), which uses smooth local weighted averaging to capture the ebb and flow of RA and other related variables.


Bucci, W., Kabasakalian, R. & the RA Research Group (1992). Instructions for scoring Referential Activity (RA) in transcripts of spoken narrative texts. Ulm, Germany; Ulmer Textbank.

Bucci, W. & Maskit, B. (2006). A weighted dictionary for Referential Activity. In J.G. Shanahan , Y. Qu, & J. Wiebe (Eds.) Computing Attitude and Affect in Text; Dordrecht, The Netherlands: Springer; pp. 49-60.

Mergenthaler, E. and Bucci, W. (1999) Linking verbal and nonverbal representations: Computer analysis of Referential Activity. British Journal of Medical Psychology, 72, 339-354.

Strunk, W., Jr., & White, E.B. (1972). The elements of style. NY: The Macmillan Co.