prepared by Jennifer Ganger
This is a review of the published literature on this subject. The Twins Study @ Harvard will soon have results to add to this question.
II. The Early Years
IV. Commentary and Further Reading
There is a pervasive assumption in the twin and language development literature that twins are somewhat delayed in language development and more prone to language disabilities.
What I want to present here is a review of conclusions from some papers I looked through when I was trying to settle the issue of whether studies on language development in twins are "generalizable" to the non-twin population. That is, whether or not twins are significantly different from the non-twin population with respect to language.
To preview the conclusion, the consensus in the literature seems to be that being a twin does in fact make a child more prone to language delays and disorders due to several biological and social factors. As we shall see, though, any of these factors can also affect singletons; twins are often just more prone to them. Furthermore, most studies show twins catching up to their singleton peers on standardized language tests during early childhood. The conclusion, then, is that twins should not be considered a special population that is differentially at risk for the mere fact of being a twin.
The pervasive assumption of the inferiority of twins in language originates with two papers from the 1930's which remain two of the largest studies to date of the language of twins as a group. These are Day (1932) and Davis (1937).
Day studied 80 pairs of twins and 140 singletons. The children were aged 1.5 to 5.5. There were 20 pairs of twins at each of 2, 3, 4, and 5 years of age. 50 spontaneous utterances were recorded from each child while playing with experimenter's toys. On several gross measures of language complexity (such as sentence length, number of different grammatical categories in a sentence) twins were found to be as much as two years behind singletons by the age of five.
Davis carried the same methodology to twins aged 5-9 and found that on structural measures the twins caught up with singetons on average, but were still more likely to have articulation problems.
However, there are several flaws in these studies (I am not the first to point them out). First, no effort was made to exclude twins who had language, speech, or hearing pathology. Since these are more common in twins (probably for reasons I'll mention below), it is possible some of these subjects were included in the sample and lowered the overall results correspondingly. Second, no information was reported on birth weight or time of gestation of the twins, both of which may be factors in language delay. Third, birth order was not considered as a possible factor, and later birth order is also known to be associated with language disturbances (Matheny & Bruggeman 1972). Fourth, the twins in this study were not always observed independently (i.e., separately), and the fact that two children of the same age were competing for the attention of the experimenter may have led to shorter sentences. Fifth, the twins were averaged as a group and then compared to singletons. However, since each twin's data are not independent of his/her co-twin's, it is not acceptable to average the results in this way and then compare them to singletons as a group.
B. Case studies
In addition to these large, (somewhat) controlled studies, there were a few case studies of twins' secret language, with the assumption that this inter-twin language was related to the delays experienced with the culturally transmitted languge (Luria & Yudovich 1959; Zazzo 1960). These are not terribly helpful to the general issue.
Working under the assumption that twins are prone to language delays and impairments, researchers in the 70's and 80's wanted to know what factors of being a twin were responsible for the delays.
A. These are some often-cited studies which set out to examine the myriad of biological variables that might contribute to language problems in twins.
Lytton (1980) and Conway, Lytton, & Pysh (1980) report a study comparing twins and singletons on language measures. Like the earlier studies, they also found that twins tended to have shorter sentences than singletons, less speech overall, less speech directed towards the mother, and a slight difference in vocabulary as asseseed by an IQ test. Although there were several biological variables that separated the two groups (including birth weight, Apgar score, and time of gestation), environmental variables such as maternal speech to child were found to account for more of the variance in the language measures. This suggests that both biological and environmental variables contribute to language delays in twins, and the authors conclude that the environmental variables are more important.
Another attempt to sort out these biological variables was presented by Mittler (1970). He studied 200 twins and 100 singletons at 4 years of age using the Illinois Test of Psycholinguistic Abilities. Overall, twins were found to be about 6 months behind singletons at this age. However, factors other than twin-ness turned out to be significant predictors of performance. These were birth weight and age at first word, with those producing their first word later than 18 months performing significantly worse than those producing their first word before that age.
Another study is reported by Record, McKeown, and Edwards, 1970. They obtained scores for thousands of twins and singletons on an exam given in Great Brittain after the 11th year of school. They found that twins performed about 5 points lower than singletons on this exam on average. Interestingly, they found that single survivors of twin pregnancies performed in between the two populations.
Finally Akermann & Thomassen (1991) compared a group of twins and singletons and found that low birth weight accounted for differences on standardized language and locomotor tests.
Another relevant study is Hay, Prior, Collett, and Williams 1987. Although these authors don't examine potential biological factors, they also document delay in the language development of twins. They gave 30 month old twins and singletons a test called the Reynell, which includes a comprehension as well as an expressive component. On the expressive test, twin boys were found to be 2 to 6 months behind norms at this age while twin girls were found to be about normal. For comprehension, twin boys were 0 to 4 months behind, while twin girls were again at or slightly above normal.
C. Examination of the biological factors alone
An attempt to sort out premature birth and/or low birth weight was reported by Mohay, Burns, and Luke 1986. They compared low birth weight twins with low birth weight singletons (less than 1500g). They used the Griffith Scale of Mental Development which includes a hearing/speech component. They tested children from one month to four years of age, and though there were sporadic differences between groups at some ages and not others, there were no significant language differences after the age of 2. Furthermore, the overall score remained within the average range at all ages. And, furthermore, by 4 years, twins were marginally above singletons on several measures. The implication is that there is some kind of interaction between low birth weight and twin-ness, such that twins are not affected by it as harshly as singletons.
In another study, Cescato & Mertin (1986) studied only birth weight (in singletons) and found that very low birth weight children continued to lag behind normal birth weight children in the language subtest of the McCarthy even at the age of 4. So low birth weight may have detrimental effects, though perhaps more so in singletons than in twins.
D. Non-biological variables
Many researchers have come to the conclusion that it is not biological but social factors that are responsible for language delays. Several studies have now found that young twins receive less directed speech from their caretaker and participate in fewer situations where their attention is jointly engaged with the caretaker. Both of these situations are thought to be necessary (to some extent) for language learning.
A recent summary of twins' language learning situation and how it differs from that of singletons is quoted below. This is from "Intelligence, Language, Nature, and Nurture in Young Twins" by J. Steven Reznick, a professor of psychology at Yale. It appears in R. J. Sternberg and E. L. Grigoreko (eds), _Intelligence, Heredity, and Environment_. Cambridge University Press. 1996(?) (Actually I'm not sure if it's out yet.) Any typos are due to me, and remarks in  are mine.
Reznick's conclusion is that the twin situation may not be different from the general sibling situation. The situation he is referring to is the fact that the lower a child's position in birth order (i.e., the later s/he is born) the more disadvantaged s/he is in terms of IQ scores. In other words, the more older siblings you have, the lower your IQ. Note that the effect is slight but significant because it supposedly happens quite consistently, on average. So, there seems to be a potential disadvantage (in terms of access to environmental resources) associated with having any siblings, and being a twin is the most extreme case possible of having a sibling.
I hope this review does not make the situation for twins' language sound doomed and hopeless. First of all, all the effects mentioned in this summary are found _on average_, not in every case. Secondly, environmental and social variables can be influenced by parents in obvious ways. Furthermore, it is clear that it is not the mere fact of being a twin that fosters language delay, but rather a combined effect of perinatal and environmental factors, any of which could theoretically affect any child, twin or not.
I also recommend chapter 5 (by Kay Mogford) in
for a more thorough review.