Abel Smith



Most experts agree that semantics is the first area of language to deteriorate.  The question asked in this paper is: does the normal aging process affect our ability to recall words from memory?  This may seem that it is obviously yes, but there is still much debate concerning the answer to this question amongst researchers.   More specifically, the question that is being debated currently is whether there is a steady decline in semantic recall throughout the lifetime.

To further pinpoint the answer to our question, this paper will address the very definition of age.  By what characteristics do we define a normal elderly individual? 

It is also necessary to know what is not the focus.  For example, naming tests are most commonly referred to in relation to Alzheimer’s patients, mainly because they have found the onset of a high rate of these errors is an indication of Alzheimer’s.  A higher rate of errors in naming indicates a large loss in semantic memory recall.  However, this paper does not focus on individuals with Alzheimer’s.  This debate revolves around the average elderly individual.  Not as much research has been aimed specifically at this group.  They have been used as control groups in comparison to Alzheimer’s patients, but not until the last 20 years or so has research been done on the average elderly individual 

Two articles will be discussed which have gathered the current research and analyzed it.  They each ask the same question:  Does naming ability decline with normal aging?  Both studies disagree with each other on the point of whether memory does decline across the lifespan.  These two studies will be explained, discussed, and contrasted.  After these are discussed, I will discuss additional considerations, as well as ideas for possible future research into the answer to this question. 




1.1.1  BACKGROUND.  In the current research, several different methods are used to measure the semantic recall of individuals.  One way in which semantic memory is tested is through observing naming abilities.  Naming refers to someone’s ability to identify or define the meaning of what a picture represents.  The method of testing most commonly used is the Boston Naming Test.  One type of this test is the object-naming test.  Others are action-naming tests, which have also shown similar results, but they are not discussed here.  In the object-naming test, either 85 or 60 line drawings of objects are presented.  At the beginning, they are very common items (e.g. bed) and near the end they are more complex (e.g. abacus).  The individual taking this test is asked to attach meaning in a brief and concise manner to these objects.  This is a standardized test and has specific directions for the examiner to follow.  This is only one example of different kinds of tests that are performed to test this ability.  Most other tests used in this study were based on similar philosophies, but were created by the examiner’s themselves.

            Two researchers, Pierre Goulet et al. and Pierre Feyeresein, sought a general consensus among the research.  Some argued against a correlation, others for a correlation.  Both felt unsatisfied in the current research in that the definition of aging was fairly vague.  They argue that age alone, as a mere number, does not take into account other important factors which could influence semantic loss.  However, they disagree on one current hypothesis.

            As cited in Goulet’s findings, many researchers have found that after the age of 70, semantic recall declines sharply.  However, Goulet does not find evidence of this occurrence as believeable.  Feyeresein argues that there truly is a drop in naming scores after the age of 70, thereby proving that semantic abilities are directly linked with age.  Neither agrees that age is the only factor, however they both disagree on how age should be interpreted in the data.     

1.1.2  PIERRE GOULET.  The first study discussed was the first to come out.  It was written by Pierre Goulet in 1994.  Goulet gathered 25 different studies and sought to find a consensus amongst them.  He wanted to know whether or not the data could conclude that normal aging leads to a decline in naming ability.  In these studies, not all used the aforementioned Boston Naming Test.  Many similar types of tests were used, however.  Additionally, he points out that not all the research was a result of the same initial research question.  Seven of the 25 articles were not searching for the answer to this question directly, but found related data nonetheless.  Upon analysis of the data, his conclusion is that normal aging does not play a role in the increase of naming errors with age.  He gives several reasons, although the reasons below are not a completely exhaustive list. 

First, he cites that the items used for testing may be biased.  He argued that in some of these studies, when objects which were more common 50-70 years ago were used, older adults performed better than younger adults.  Therefore, there was a generation bias involved in some of these tests.  Secondly, the methods of issuing the test were variable.  Some examiners’ interpretation of the directions on the Boston Naming Test were incorrect.  For example, some examiners varied in the way they handled the situation when a patient was unsure of a word.  Some were overly helpful while others were less helpful than instructions indicated they should be.

His third reason for defending his argument is a critical one.  It strikes at the issue of the definition of aging.  He argues for a distinction to be made between primary and secondary aging.  Primary aging are consequences due to normal aging processes.  Secondary aging are supplementary measures used to deal with disease.  Examples are medication, institutionalization, etc.  He claims that none of the studies he gathered took into account either of these issues.  Instead, all studies referred to chronological aging (age only), but didn’t take into account the psychological and cognitive processes common with primary aging.  Additionally, Goulet claims that since secondary aging characteristics are not common to the majority of the elderly population that subjects who present these characteristics should be thrown out of consideration.  Therefore, he concludes, chronological age cannot be the sole cause of naming errors, nor should it be considered as a criteria of evalution.   

But what are the characteristics that make up an average aging individual?  Goulet responds with some of the following criteria:  free of visual problems/without any disease/taking no medication/noninstitutionalized/highly educated/culturally, economically, and professionally favored individuals.  Therefore, Goulet concludes that studies have been ignorant of other factors and should not jump to conclusions that chronological age is the unequivocal cause of decline in naming errors.  Additionally, he concludes that because no single study took this into account, none can convincingly argue that normal aging leads to a decline in naming errors:

Chronological age alone cannot account totally for the controversy surrounding picture naming and age. (Goulet, 1994: 631)


1.1.3  FEYERESEIN.  Three years later, in response to Goulet’s article, a Belgian professor, Pierre Feyeresein, wrote a response to Goulet’s article.  Using nearly all the same data, Feyersein claimed that he did find evidence for a decline in naming throughout age.  He used a method called meta-analysis to evaluate Goulet’s argument.  A meta-analytic procedure compares results from studies that have similar group sizes, research questions, etc. 

Using this analysis, Feyeresein identified several weaknesses in Goulet’s analysis of the data.  For example, he noticed that the size of the group can cause a lot of unnecessary variance.  In a small group, a few errors will give a poor score, where the same number of errors can yield a high score if more questions are asked.  Additionally, he tried to evenly match data with groups whose focus of study was the same.  As mentioned previously by Goulet, not all the study groups began their studies with same question in mind. 

Most importantly, Feyeresein felt that Goulet had been too ignorant in the description of chronological age.  Goulet had only referred to the distinction between “younger” and “older” individuals, even though age groups had been provided in all the studies.  Therefore, Feyeresein sought to eliminate the differences between studies and only compare the studies that were more similar in group size, research pursuits, etc. and also to set a more clear guideline for chronological age.

Next, he began the filtering process.  To begin with, he took all of Goulet’s 25 studies and added 7 more of his own.  Then he kept only the ones that had sufficient statistical information.  This left 19.  Next, he continued to narrow it down by matching studies whose subjects had similar ages.  He established three important age groups:  1) below 50 2) between 50 & 69, and 3) more than or equal to 70.  The naming scores of these groups would be compared to each other to distinguish if there is indeed a decline according to chronological age.  After this, he was able to narrow the focus of the study down to 11 studies.  The ratio of arguments refuting to accepting age as a factor in decreased word-finding ability was 3:11.  This was essentially identical to the ratio that Goulet used (7:25 against:for a decline in aging). In table 1 (p. 9), an increase in naming errors can be seen throughout the life span.  From this data, Feyersein concludes that a steady increase in the number of naming errors indicates a decline in semantic recall through these age groups.  Feyeresein additionally points out that 2 of the remaining 3 studies, which argued against a correlation of age and naming ability, did not use subjects over the age of 70. 

Although Feyersein agrees that a distinction should be made between primary and secondary aging, he argues that Goulet had erroneously ignored chronological age.  He agrees with Goulet in that by itself  chronological age cannot be the only consideration, but also states that it should not therefore be thrown out of consideration.  Regardless, they both seem to agree that additional considerations should be examined.  Feyeresein said the following:



The fact that picture naming accuracy does decline with advancing age is only

            part of the statistical truth.  Goulet and others have rightly stressed that

            chronological age per se cannot be the cause of difficulties in word finding

            or other cognitive skills.  (Feyeresein: 1328)




2.1.1  In research published last year, Susan Baressi released the findings of her study on this subject.  She sought for an answer to the same question, but she also wanted to study what kind of activities are performed by those who exhibit high scores in their old age.  In her study, she surveyed 11 60 year olds, 11 forty year olds, 11 70 year olds, and 7 80 year olds.  This survey was done 3-4 years after the subjects had been tested on the Boston Naming Test (BNT) a third time over a period of ten years.  She found two things:  One, older age and less education contributed to lower scores on the BNT.  Second, in relation to activities that have significance on scores on the BNT, she found that television viewing significantly decreased scores on the BNT.  She suggested that this may be because television is a passive language activity compared to everyday conversational language tasks.  Baressi’s correlation with TV viewing further argues that the cause is most likely not due to one factor alone.     




3.1.1  QUESTIONS.  It appears that Goulet’s article, has helped forward more research in additional areas of consideration.  Although there is still disagreement on chronological age’s correlation with naming ability, there has been agreement that age alone is not the only factor that influences a loss in word-finding ability.  Criteria must be set as to what a “normal elderly individual” consists of.  Goulet’s criteria seem like a good start, but I wonder if they are some of the criteria proposed are truly characteristics of an average elderly individual or just a healthy elderly individual.  Taking no medication and free of visual problems do not seem to be the mainstream of the elderly population. These criteria should be representative of the elderly population as a whole.  More consideration into this matter should take place. 

I agree with the meta-analytic methods used by Feyeresein.  To find more consistent results, we need to have research that is similar in design and methodology.  Although I believe in Feyeresein’s findings about chronological age, he also showed some ignorance of his own.  Goulet’s assertion that no single study took into effect all secondary aging factors is a good one.  Although Feyeresein was able to find an increase in naming errors, he didn’t take into account other secondary aging characteristics. 

The question raised by Barressi’s study (what are active language activities?) is a good one and should also be pursued further.  Her study, published after Goulet and Feyeresein, shows improvement to analyze more than just choronological age.  If these criteria continue to be more solidly established and uniform testing procedures adopted, more consistent results will be the result.  Most importantly, in clinical applications, we will be able to identify the norm for the average elderly populations and possibly recommend activities that may slow down this semantic meltdown. 
















Abel Smith

November 15, 2001












Although many of us associate the elderly with forgetfulness, the the causes for this semantic decline are still under debate.  Traditionally, age (20 vs. 75) is used to measure this decline.  Additionally, other testing methods are used to measure semantic loss, such as naming tasks.  An decrease in the ability to perform naming tasks shows should indicate a loss in semantic memory.  Some older individuals have been shown to exhibit a decrease in ability to perform these tasks after the age of 70.  Constrastingly, other studies have shown no significant decrease throughout the life span.  This hypothesis is referred to as the 70’s hypothesis.  This paper contrasts the two opposing viewpoints on this issue.  They argue the same question:  Does normal aging correlate with an increase in inability to recall words?  One says yes and the other no.  These viewpoints are explained and contrasted.   Although they disagree on how to interpret the definition of aging, they do agree that age alone cannot be the only factor in semantic loss.  They propose additional considerations to more clearly define the concept of aging.  Additional activities to possibly decrease semantic deficit are also briefly discussed.  Possible continued areas of focus of research are proposed to clarify this issue. 


















Feyeresein, Pierre.  Journal of Speech and Hearing Research, 1997,40,6, Dec.  pp. 1328-1333.

Goulet, P., Ska, B., Kahn, S.  Journal of Speech and Hearing Research, 1994, 37, 3, June.  pp. 629-644.

Hamilton, Heidi E.  Language and Communication in Old Age  pp. 77-90.   (New York, NY, Garland Publishing).

















Study #1





Study #2





Study #3





Study #4





Study #5





Study #6





Study #7





Study #8





Study #9















Studies       Age-related decline?          < 50                  50-69                 > or = 70

                                                            vs. 50-69         vs. > or = 70              vs. < 50


Mean difference:                         .08                     .88                    1.23

Table 1.   Comparisons of different age groups show the greatest difference in age groups of  < 50 vs.  > or = 70.  Note:  3 of the final studies used in Feyeresein’s analysis were not part of Goulet’s 25 studies.  (Adapted from Feyeresein: 1331)