The Often Subtle Linguistic Cues in Advertising

Larry Percy, HBM/CREAMER, Inc.
ABSTRACT - In trying to communicate, regardless of the situation, concern is always focused upon "what to say.' Beyond this obvious point, however, if one is to be an effective communicator, one must also be concerned with the semantic properties of the message, extending beyond simple verbal meaning. Language can be subtle at times in its effects. Weiner and Mehrabian (1968) have pointed out that the words of communications transmit information that is complementary, supplementary, or redundant to the information transmitted in other components in the communication. As Weiner and Mehrabian go on to point out, careful consideration will show that what appears to be the same thing (i.e. the-same content) said with different words, can be a basis for inferring quite different feelings or attributes from the message. This paper explores these points, and applies their implication to advertising.
[ to cite ]:
Larry Percy (1988) ,"The Often Subtle Linguistic Cues in Advertising", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 269-274.

Advances in Consumer Research Volume 15, 1988     Pages 269-274

THE OFTEN SUBTLE LINGUISTIC CUES IN ADVERTISING

Larry Percy, HBM/CREAMER, Inc.

ABSTRACT -

In trying to communicate, regardless of the situation, concern is always focused upon "what to say.' Beyond this obvious point, however, if one is to be an effective communicator, one must also be concerned with the semantic properties of the message, extending beyond simple verbal meaning. Language can be subtle at times in its effects. Weiner and Mehrabian (1968) have pointed out that the words of communications transmit information that is complementary, supplementary, or redundant to the information transmitted in other components in the communication. As Weiner and Mehrabian go on to point out, careful consideration will show that what appears to be the same thing (i.e. the-same content) said with different words, can be a basis for inferring quite different feelings or attributes from the message. This paper explores these points, and applies their implication to advertising.

INTRODUCTION

No one doubts the importance of the words chosen in verbal communication in determining just how effective that communication is likely to be. And while attention may be paid to insuring that descriptions or attributes within a target message reflect those things most likely to be meaningful to the target receiver, very little consideration seems to go into possible interactions among those descriptions or attributes, and even less to grammatical consideration. One hopes that there will be a positive, additive relationship among multiple descriptive elements in a communication and that they are well ordered; but that isn't always the case. To the extent that there are incongruities between these verbal elements in a communication, potential dissonance or confusion may result.

Clearly, the components of our language and the way they are assembled will influence how well a message is communicated. Schloss (1981) has remarked that there seems to be something about the English language that causes people to react to certain words or sounds differently than they react to others.

SOME SEMANTIC PROPERTIES IN PROCESSING

It is not unusual to remark in the study of semantics the fact that the same words may have different meanings or that different words may have the same meaning (although when you stop to think about it, no two words are likely to have exactly the same meaning). Of course, we know these words as synonyms and homonyms; and there has certainly been a great deal written about their strong effects in language. For example, when subjects are asked to recall lists of words that contain synonyms, it is not unusual for them to substitute another synonym for that word in playback (cf. Grossman and Eagle, 1970; Kausler and Settle, 1973). With homonyms, the problem is that when receivers hear or see words with diverse or multiple meanings, they are quite likely to be immediately reminded of the several meanings of the word (Conrad, 1974). Associated problems with synonyms and homonyms in advertising have been discussed by Percy (1982).

While it is obvious that different words may have different meanings, as Palmer (1976) points out, their simple meanings are in and of themselves not of much interest. Only when these different meanings are in some way related does one become concerned with their probable impact upon comprehension in communication.

One of the more important considerations in writing copy for advertising should be the extent to which any set of commonly elicited attributes or descriptions are in fact seen as compatible or incompatible in the sense of a synonymic usage. There are, of course, a number of analytic techniques available that help delineate this distinction. Two such methods are discussed below.

Given the potential conflict among product descriptors, and the possibility that underlying semantic incongruity could affect product evaluations, it is important to fully understand just what these semantic relationships might be for any given set of potential product attribute claims. Additionally, it is helpful to assess the cumulative effect of such product attributes, assuming semantic compatibility, upon preference judgements.

A stimulus set of potential product descriptions for a new light meal product was elicited from a series of exploratory focus groups. Once this set of attributes was determined, a quantification of consumer perceptions of the interchangeability of the attributes in describing various types of meals was gathered, followed by a rank ordering of various combinations of these attributes in describing a new product for people interested in a better way of eating.

The original exploratory focus groups (following Calder's distinction, 1977) provided a set of seven potential product descriptions: convenient, easy, less filling, light, lower calorie, quick, and simple. Within this context, each might be considered part of a group of synonyms. A quantitative study then followed among 150 female heads-of-household aged 18-54. Subjects were recruited via mall-intercepts; 50 interviews in each of Atlanta, Boston, and San Francisco. Data were collected to be analyzed with multidimensional scaling and conjoint measurement techniques.

Multidimensional Scaling - In our first check on the compatibility of the attribute set, subjects were given a set of 21 cards, and listed on each card was one of the 21 pairs of the seven attributes. The card set was presented to subjects as a set of cards where on each card you will find "two words or phrases that people have used to describe various types of meals." They were first asked to separate the cards into two piles: cards with words that could be used almost interchangeably in describing a particular meal and cards containing words that they would not use interchangeably to describe a meal. They continued, ranking the word pairs from the most interchangeable to the least interchangeable. Subjects then considered the set of cards containing words or phrases not seen as interchangeable, and ranked them from the card containing the two words or phrases least likely to be used together in describing a meal. These data were then analyzed using Kruskal's MDSCAL, and the results are shown below as Figure 1.

As the space suggests, subjects are likely to see the attribute "easy" and "simple" as highly interchangeable in describing a meal; and while quick and convenient also tend to be clustered in the same cognitive environment, they are not perceived to be quite so interchangeable. The attributes "lower calorie" and "less filling" are both seen as significantly different from the ease and convenience attributes, yet are not themselves perceived as interchangeable. The word "light" falls in the center of the space, about the same distance from the easy and convenient cluster, less filling and low calorie.

FIGURE 1

MDSCAL 5-M CONFIGURATION OF SEMANTIC ALTERNATIVES

From a practical standpoint, it is important to know that although subjects feel each of these seven descriptors is appropriate for a new product marketed to those interested in lighter eating, they reflect several possible product configurations. The fact that subjects describe this new product concept in these terms does not mean they are compatible semantic alternatives.

Conjoint Measurement - Given the fact that there are various alternative semantic orientations for this new product, consideration was given to which combination of these attributes are most likely to be seen as best describing the product. Specifically, subjects were given 8 cards with various combinations of the seven attributes based upon a random reduced block orthogonal array. Subjects were asked to look through the set of cards, and thinking about a new product like this, rank order them from the one that they felt best described the new product to the one that they felt would be least descriptive. They were reminded that there were no right or wrong answers, that we were only interested in their opinions.

This rank ordering was analyzed with a CONJOINT measurement algorithm, providing utilities for each semantic alternative. These results are provided in Table 1. What these utilities suggest is that within the context of the seven semantic alternatives generated by subjects in describing the new product, the word "light" is considered by far the best description, unpaired with any other word. The next most favorable words would be "convenient" and/or "easy." While these two words are seen by subjects as compatible, we know from the multidimensional scaling that they are not necessarily additive with the attribute "light." In some cases this may be true, but it does not necessarily follow for all perceptions of this new product. The words "quick" and "simple," while not as good as "convenient" or "easy", are nonetheless semantically congruent and generally interchangeable as descriptors of this new product. The remaining alternatives are not reasonable considerations.

TABLE 1

PART-WORTH UTILITY VALUES FOR SEVEN SEMANTIC ALTERNATIVES

It is interesting to note that given an opportunity to "combine" these attributes in describing the new product, subjects tended to utilize their understanding of the semantic compatibility of the words. There was no mixing at all of attributes perceived not to be interchangeable; yet those felt to be interchangeable "convenient," "quick," "easy," and "simple") were generally felt to be equally appropriate as descriptors.

In effect, then, we have found that within this product environment the attributes of "convenient," "easy," and "simple" are almost certainly used synonymously; but that in only certain circumstances (i.e. for specific referent products) would the word "light" or the phrase "less filling" or "lower calorie" also be a part of this synonym group.

IMAGE VALUE OF WORDS AND SENTENCES

The imagery values of words and pictures is strongly related to their ability to enhance effective communication. The foundation of high imagery is measured by the concreteness of words. Concrete words are generally described as those which refer to objects, persons, places, or things that can be seen, heard, felt, smelled, or tasted. They have been found to be better remembered and more meaningful (Yuille and Paivio, 1969) as well as better comprehended (Begg and Paivio, 1969; Paivio, 1971; Sheehan, 1970.)

In a study of abstract vs. concrete copy in advertisements, Rossiter and Percy (1980) found concrete copy to stimulate almost twice the favorable attitude toward the advertised brand. While somewhat less research has been done on the imagery value of sentences and other large verbal units, what is available strongly implies that greater imagery yields better communication. Several researchers have found that high-imagery sentences are more easily and correctly understood (Holyoak, 1974; Jorgensen and Kintsch, 1973). Overall, one might safely conclude that the use of high-imagery words and constructions in advertising copy, especially in headlines, will enhance successful message processing and communications.

In creating advertising for several beer brands, a set of product descriptors was developed from exploratory qualitative group work. Table 2 below details the key attributes that were developed, along with their concreteness ratings from Toglia and Battig (1978). Their work followed very closely that of Paivio et al. (1968), the only major difference being their usage of "low concrete" as the low end anchor rather than "high abstract," consistent with Spreen and Schultz (1966).

TABLE 2

CONCRETENESS OF KEY ATTRIBUTES

As a result of this work, advertising copy for one brand included the words: "brewed fresh and natural for a smooth taste"; and copy for a second brand included the so clean, refreshing, and light."

SENTENCE CONSTRUCTION

A number of considerations involving sentence construction that have obvious implications for advertising have been studied. For example, how long should a sentence be for maximum communication? This has clear implications for writing headlines in advertising. Coleman (1962) has found that when technical passages are divided into short sentences, comprehension is significantly better than when the same message is written with longer sentences. Wearing (1973) found that both recall and comprehension improved as the number of words in a- sentence increased from five to seven, but fell off significantly for sentences of greater length. In a review of the "read most" scores for 78 corporate ads, a remarkably similar result was noted by Percy ( 982): the relationship between read-most and number for words in a headline increased to a maximum in the range of five to eight words, then fell off significantly for headlines containing more words. Perhaps the reason for this may be that longer sentences tend to be both grammatically and psychologically more complex because of the addition of phrases and clauses. In fact, grammarians have identified sentences that contain both an independent and dependent clause (something they call "self-embedded") as the most difficult type of sentence to remember or comprehend (Fodor and Garrett, 1967; Forester and Ryder, 1971). Holding sentences to simple constructions of about seven words in length would seem to be the ideal.

GRAMMATICAL STRUCTURE IN PROCESSING

We also know that active declarative sentences are significantly more easily processed than active questions, which in turn are easier to process than passive sentences; that negative sentences are yet more difficult to process, and so on (c.f. Wason, 1965 Slobin, 1971; Kanouse, 1972). Various verb forms are more or less likely to stimulate deductive vs. inductive reasoning (Kanouse, 1972); the number of words in a sentence (Wearing, 1973), whether they are concrete or abstract (Paivio Yuille, and Madigan, 1968), all influence how well we are likely to process a piece of communication, as we have seen.

Given the critical importances of the grammatical structure of a sentence in mediating cognitive processing, results from work (reported in detail by Percy, 1987) that looked at the communication response to the four sentence stimuli listed below are discussed, illustrating the potential differences occasioned by nothing more than grammatical changes.

It's not just the calories that count, it's the taste.

It's the taste that counts, not just the calories.

It's the taste, not just the calories that count.

It's not just the calories, it's the taste that count.

Adult male and female subjects were randomly assigned to one of four experimental cells representing the four stimulus sentences and asked for the first thoughts that came into their mind on hearing the sentence. Next, they were asked whether or not they felt the sentence implied that taste or calories was more important in a product that was so described.As the table below indicates, regardless of the sentence heard by the subjects, roughly the same number of words were elicited in response.

TABLE 3

AVERAGE NUMBER OF WORDS ELICITED

Each of the words elicited was classified according to Laffal's (1973) concept dictionary. What he provides is a collection of words from the English language that have been organized to reflect the "kinds" of content represented in the words. These concepts, or categories of content, are presumed to reflect cognitive-conceptual sets which are evoked whenever a pertinent word is encountered. By classifying the elicitations for each sentence we are in effect doing a content analysis of the response utilizing the semantic structure of the language used in response stimulated by the sentence. The null hypothesis would surely be that given a common semantic base, the thoughts generated by those words should be similar. But as the correlations in Table 4 reveal, the profile (as measured over 20 measured concepts) of the responses, while certainly not unique, are far from common.

Looking further at these classifications, we find that of the 20 concepts utilized by the subjects in response to these sentences, half were common to all sentences; and in fact the bulk of the words used in each case could be categorized by these 10 concepts. Significantly (x2 = 9.74; n = 3), however, sentence four is much less likely to elicit words common to those from the other sentences: 83%, 79%, and 87% of the words elicited for the first three sentences reflected a common underlying cognitive base, but only 68% of those elicited by the fourth sentence.

TABLE 4

PEARSON CORRELATION MATRIX OF LAFFAL CONCEPT PROFILES

Another way of looking at the words elicited by these sentences is their concrete and imaging value. If differences were to occur here, we would have more than the meaning differences implied by the Laffal concepts to deal with; we would have potential learning differences as well. As we have seen, in terms of concreteness, it has been well established that more concrete words are better remembered, tend to be more meaningful, and as a result better comprehended (Yuille and Paivio, 1969) . Would not be reasonable to assume that if a stimulus elicited more concrete thoughts it too would enjoy some of these same attributes? In any event, the resulting cognitive response, which reflects its interpretation in memory, certainly does.

Beyond this question of memory and learning, Paivio (1971) has pointed out that both recognition and recall occur more accurately and faster for concrete words. Additionally, a number of other studies have shown that concrete words are more positively associated with comprehension (Begg and Paivio, 1969; Paivio, 1971; Sheehan, 1970). Given this, one should expect better sentence comprehension to follow from more concrete responses. Finally, Rossiter and Percy (1978) found that representation of advertising claims for a hypothetical new beer when concrete generated almost twice the favorable attitude for the product as more abstract ones.

Each of the cognitive elicitations for the four stimulus sentences were checked for concreteness as reported by Toglia and Battig (1978). Approximately 80% of the thoughts elicited by the stimulus sentences were classifiable by the Toglia and Battig word norms. And as the data in Table 5 reveal, there was no significant difference in the types of responses that were not able to be classified.

TABLE 5

ELICITATIONS NOT CLASSIFIED BY TOGLIA AND BATTIG WORD NORMS

Looking at the "concreteness" ratings for all of the responses classified, the 56% above the Toglia and Battig mean of 4.4 for sentence three (versus 44%, 40%, and 45% for the other sentences) suggests that the stimulus sentence three "It's the taste, not just the calories, that count" tended to elicit more highly concrete thoughts. While only marginally significant (at about the 86% confidence level), it is interesting that sentence three is the only clearly center-embedded sentence. Table 6 below details the percentage of words classified above the mean for the concreteness of elicited thoughts.

Closely related to the concept of concreteness is that of imagery value in words (Paivio, 1971). As a result, even though it is not strictly impossible for more abstract words to evoke visual images, unless learned specifically, it is much less likely (Rossiter and Percy, 1983). For example, although words like "fantasy" or "dream" are highly abstract, they may exhibit high imagery value.

Unlike concreteness, research on the imagery value of large verbal units such as sentences is sparse. Among the few studies available to us, however, there is a strong suggestion that imagery value enhances communication. Jorgensen and Kintsch (1973) have shown that sentences with higher imagery value tend to be evaluated significantly faster as true or false; and Holyoak (1974) has found them significantly easier to understand than sentences rated low in image value. And as Percy (1982) has suggested, following the work of Williams (1979), who found that high imagery syllogisms were faster and more accurately solved than the same syllogism made up of low imagery words, advertising claims which more easily arouse visual imagery of items and relationships should be more easily comprehended. Hence, the sentence stimuli eliciting higher imagery thoughts should be more easily comprehended.

Looking at the imagery word norms for all of the >-responses classified, the 62% above the Toglia and Battig mean of 4.55 for sentence three (versus 40%, 37%, and 45% for the other three sentences) again suggests that stimulus sentence three tends to elicit more high imagery thoughts. Here, the difference tends to be much more significant (the 94% confidence level). The parallel between the concreteness and imagery ratings is consistent with our expectations from the literature.

TABLE 6

HIGH CONCRETE AND HIGH IMAGERY ELICITATIONS

After the subjects were asked for the first thoughts that were stimulated by the sentences, they were asked whether taste or calories was most important to the product described. The correct inference in each case, of course, was taste; and indeed, the majority of subjects did in fact say taste. However, among the minority saying calories, there was a significant difference between sentence stimuli. As the data in Table 7 indicates, those exposed to sentence two, "It's the taste that counts, not just the calories," were significantly more likely to incorrectly infer that calories was the most important consideration in the product.

TABLE 7

INFERENCE DERIVED FROM STIMULUS SENTENCES- 1

One explanation of this result might be found in the recency literature, which suggests that words that have been heard (or read) more recently are retrieved more rapidly than words which occurred longer ago. And this coupled with the increasing difficulty of correctly dealing with the negative, right branching clause ("not just the calories") in processing, could be causing subjects to be more likely to misattribute calories rather than taste as the dominant product attribute in this grammatical construction.

Also, even though it might be argued that "taste" should be expected to have been more likely to have been correctly processed and remembered simply because it is a more frequently used word than "calories" in the English language (used over 100 times per million vs. about 4 times per million for calories according to Thorndike and Lorge, 1944), this issue of word frequency has been shown in a study by Scarbourgh, Cartese,and Scarbourgh (1972) to be confounded by recency effects.

SUMMARY

The specific results reported here from several studies, while not always highly significant, nevertheless indicate that linguistic elements and grammatical structure do indeed function as mediating variables in message processing. Specifically, we have seen where grammatical variations of an advertiser's claim meant to communicate that taste is the critical attribute of their product, not merely lower calories, can significantly influence the likelihood that the message will be correctly communicated. Additionally, we have looked at the compatibility of words in terms of perceived congruence and discussed several applications of psycholinguistic principles to advertising. The point is, these often subtle cues in language can have significant impact upon how advertising will be processed. This clearly argues strongly for testing advertising to be sure one knows just what is likely to happen in the market because of the way the copy has been written.

REFERENCES

Begg, I. and Paivio, A. (1969), "Concreteness and Imagery in Sentence Meaning."Journal of Verbal Learning and Verbal Behavior, 8, 821-827.

Calder, B.J. (August, 1977), "Focus Groups and the Nature of Qualitative Research." Journal of Marketing Research, 14, 353-364

Coleman, B.B. (1962) "Improving Comprehensibility by Shortening Sentences."Journal of Applied Psychology, 46, 131-34.

Fodor, J. and Garret, M. (1967) "Some Syntactic Determinants of Sentential Complexity."Perception and Psychophysics 2, 289-96.

Forester, K.I. and Ryder, L.A. (1971), "Perceiving the Structure and Meaning of Sentences."Journal of Verbal Learning and Verbal Behavior, 10, 285-296.

Grossman, L. and Eagle, M. (1970), "Synonymity, Antonymity, and Association in False Recognition Response."Journal of Experimental Psychology, 83, 244-248.

Holyoak, K. (1974), "The Role of Imagery in Evaluation of Sentences: Imagery or Semantic Relatedness?" Journal of Verbal Learning and Verbal Behavior, 13, 1 6-3-66

Jorgensen, C.C. and Kintsch, W. (1973), "The Role of Imagery in the Evaluation of Sentences."Cognitive Psychology, 4, 110-116.

Kanouse, D.E. (1972), "Verbs as Implicit Quantifiers."Journal of Verbal Learning and Verbal Behavior, 11, 1141-1147.

Kausler, D.H. and Settle, A.V. (1973), "Associate Relatedness vs.Synonymity in the False Recognition Effect." Bulletin of the Psychonomic Society, 2, 129-131.

Laffal, J. (1973), A Concept Dictionary of English Essex, Connecticut: Gallery Press.

Paivio, A. (1971), Imagery and Verbal Processes New York: Holt, Rinehart, and Winston.

Paivio, A., Yuille, J.D., and Madigan, S.S. (1968), "Concreteness, Imagery, and Meaningfulness Values for 925 Nouns." Journal of Experimental Psychology: Monograph Supplement, 76 (3,P.2).

Palmer, F.R. (1976), Semantics: A New Outline Cambridge: Cambridge University Press.

Percy, L. (1987) "Exploring Grammatical Structure and Non-Verbal Communication" in S. Hecker and D.W. Stewart (eds.) Nonverbal Communication in Advertising Lexington, Mass.: Lexington Books.

Percy, L. (1982), "Psycholinguistic Guidelines for Advertising Copy." In A.A. Mitchell (Ed.), Advances in Consumer Research: Vol IX Ann Arbor: Association for Consumer Research.

Rossiter, J.R. and Percy, L. (1978), "Visual Imagery Ability as a Mediator of Advertising Response." Proceedings of the Association for Consumer Research, 621-628.

Rossiter, J.R. and Percy, L. (1983), "Visual Communication in Advertising." In J.R. Harris (Ed.), Information Processing Research in Advertising. Hillsdale, N.J.:Lawrence E. Erlbaum & Associates.

Schloss, I. "Chickens and Pickles." (1981) Journal of Advertising Research, 21, 47-50.

Sheehan, P. (1970), "The Relation of Visual Imagery to True-False Judgement of Simple Sentences." Unpublished Master's Thesis, University of Western Ontario.

Slobin, D.I. (1971), Psycholinguistics, Glenview: Scott, Foresman and Company.

Spreen, O. and Schultz, R.W. (1966) "Parameters of Abstraction, Meaningfulness, and Pronounceability for 319 Nouns." Journal of Verbal Learning and Verbal Behavior, 5, 459-468.

Thorndike, E.L. and Lorge, I. (1944), The Teacher's Word Book of 30,000 Words New York: Bureau of Publications, Teachers College.

Toglia, M.P. and Battig, W.F. (1978), Handbook of Semantic Word Norms Hillsdale, N.J.: Lawrence E. Erlbaum & Associates.

Wason, P.C. (1965), "The Contexts of Plausible Denial." Journal of Verbal Learning and Verbal Behavior, 4, 7-11.

Wearing, A.J. (1973), "The Recall of Sentences of Varying Length." Australian Journal of Psychology, 25, 155-161.

Weiner, M. and Mehrabian, A. (1968), Language Within Language: Immediacy, a Channel in Verbal Communication. New York: Appleton-County-Crofts.

Williams, R.L. (1979), "Imagery and Linguistic Factors Affecting the Solution of Linear Syllogisms." Journal of Psycholinguistic Research, 8, 123-140.

Yuille, J.C. and Paivio, A. (1969), "Abstractness and Recall of Connected Discourse." Journal of Experimental Psychology, 82, 467-471.

----------------------------------------