Aging and Acquisition of Latin 1 Running Head: AGING AND ACQUISITION OF LATIN Aging and Second Language Acquisition: Differential Success in Learning Latin Grammar Via Implicit and Explicit Feedback

The ability to learn a second language becomes difficult for adults, but not impossible. While implicit processes that can contribute to acquisition of syntax are relatively spared in aging, there are marked deficits in explicit learning, suggesting that language instructions calling on implicit learning might be more effective than that based on explicit learning for older adults. In this experiment, 20 adults ages 66-81 were taught aspects of Latin grammar with either implicit or explicit feedback and were compared to a group of college-age students who previously completed the same task. Overall, the results suggest that the implicit method of teaching the grammar worked better for the older adults but this was not the case for the students. Furthermore, limited exposure to high school Latin markedly increased the ability of older adults to learn the grammar. There were no significant age deficits in learning when only those without previous Latin experience were compared, which bodes well for older adults with strong motivations to learn a second language. Aging and Acquisition of Latin 3 Introduction The critical period hypothesis, first proposed by Penfield and Roberts (1959), suggests that language acquisition becomes significantly impaired if not impossible after a certain age. Since the time the idea was first introduced to neurolinguistics, several researchers have argued over the age at which second language acquisition becomes virtually impossible. Hyltenstam and Abrahamsson (1992) claimed that after the age of 7, people would not be able to attain a nativelike proficiency if their conversations were carefully analyzed. Ellis (1996) and Birdsong (2006) set the age higher, at just before puberty and before the age of 16, respectively. Johnson and Newport (1989) proposed that there are two critical periods with different rates of decline in ability to pick up elements of language. Scovel (1988) proposed that pronunciation is the only aspect of language that is subject to the critical period. Whichever aspect of language is affected, several of these researchers agree that the critical period applies not only to the acquisition of a first language, but also to second language acquisition. While it is evident that adults have more difficulty in acquiring a second language, the critical period hypothesis has been nearly dismissed by researchers. Instead, most current research adopts the idea that the ability to learn a second language decreases steadily with increasing age. Hakuta and Bialystok (2003) used 1990 census data to disprove the idea of the critical period hypothesis in second language acquisition. In their study, they examined native Chinese and Spanish speakers who moved to the United States. By including factors such as age of immigration and level of education, they were able to create a regression that represented level of language acquisition versus age. In their study, Hakuta and Bialystok looked at data from people ages 10-70 who had been in the United States for at least 10 years. If a critical period in second language acquisition does exist, the slope of the regression would be markedly different Aging and Acquisition of Latin 4 before and after the proposed critical age range. Instead of finding a plateau characteristic of a critical period, they found a negative linear relationship between age and the ability to learn a second language. Several researchers have agreed that there is a linear decline in the ability to learn a second language with age instead of a critical period (Singleton 2001, Midford and Kirsner 2005, Hyltenstam and Abrahamsson 2000, Birdsong 2006). Several studies have examined the effect of age of arrival in a new country (and subsequent age of acquisition) on a person’s ability to pick up the language of that country. Birdsong (2006) noted that there is a negative correlation between age of arrival and ultimate attainment of a second language. He also found that later ages of arrival often led to more errors in grammatical judgment and a higher degree of a non-native accent. Flege and Yeni-Komshiam (1999) examined native Korean speakers trying to learn English and also found that the strength of their foreign accent was positively correlated with the age of arrival in a new country. In addition, they used a grammatical judgment task to assess the morphosyntax of the Korean speakers and found that the degree of native-like morphosyntax was negatively correlated with the age of arrival in America. Adults can attain a native-like proficiency in a foreign language Several studies suggest that with significant exposure, adults can approach native-like proficiency in a second language. Marinova-Todd (2003) tested 30 adults who arrived in an English-speaking country after the age of 16 and who had been living in an English-speaking country for at least 5 years. Participants were tested on nine measures including pronunciation, spontaneous speech, read alouds, morphosyntactic accuracy, lexicon, and narrative language. In the study, 3 subjects were indistinguishable from native speakers on all 9 measures and an additional 6 subjects were indistinguishable from native speakers on 7 measures. Birdsong Aging and Acquisition of Latin 5 (1992) examined 20 English speakers with an average age of 40 who were first exposed to French after puberty and found that 15 of them reached native-like abilities in grammar judgment with significant exposure to the language. Bongaerts et al. (2000) studied advanced late learners of Dutch and found that adults (with a mean age of 40) could demonstrate a native-like accent with enough opportunity to practice the language. These studies prove that while learning a second language may be more difficult at an older age, it is not necessarily impossible. Rossi et al. (2006) examined event related potentials in late learners (19-40 years of age) of a second language to compare their neural experience of a language to native speakers. Nonnative subjects were exposed to tasks in German and Italian that had word category violations and morphosyntactic agreement violations and their responses were compared to native speakers. High proficiency late learners of German and Italian had the same ERPs as native speakers during the task, whereas low proficiency late learners had ERPs that were significantly different from native speakers. The results suggest that it might be the level of proficiency in a second language, not age of acquisition, that accounts for the differences between late learners and native speakers. This study also showed that late learners of a second language could be proficient to the level where their neural processing of the second language is the same as native speakers. Different aspects of language are processed in different brain regions Ullman (2001) examined the declarative-procedural model as it relates to languages. Through brain imaging and looking at the differences between Alzheimer’s and Parkinson’s disease along with the differences between anterior and posterior aphasias, he was able to show that syntax is processed mainly in the frontal cortex and basal ganglia whereas lexicon is a function of the medial-temporal lobe, particularly the hippocampus. Birdsong (2006) noted that Aging and Acquisition of Latin 6 while episodic memory, associative memory, and semantic memory show marked age deficits, there are fewer declines in procedural memory throughout the lifetime. Therefore, the ability to process grammar while learning a second language may be relatively spared in aging. The learning of vocabulary, however, may prove more difficult due to the declines in associative and semantic memory. Individual differences and their effects on distinct language processes While general factors related to aging may contribute the most to the difficulties in learning a second language throughout one’s lifetime, individual differences also play a role. McGeorge and Crawford (1997) looked at psychometric intelligence and its effects on series completion, grammar completion, and other tasks that are related to learning a second language. Overall, scores on the WAIS were less related to implicit tasks than to explicit tasks. This suggests that the greatest variability among late language learners may arise in the area of lexicon rather than grammar. Gebauer and Mackintosh (2007) supported the notion that abilities in implicit learning are not dependent upon the level of psychometric intelligence. Using Reber’s (1967) artificial grammar task, a serial learning task, and a process control task, the researchers showed that when subjects received implicit instructions on the tasks, their performance was not correlated to measures of psychometric intelligence. Bley-Vroman (1990) proposed a hypothesis for why success in second language acquisition has more variability than in first language acquisition. While people learning a first language have access to universal grammar—the innate ability to pick up on common regularities and principles of grammar that is equal among individuals, late learners of a second Aging and Acquisition of Latin 7 language are without this tool and success is more closely tied to cognitive differences between individuals. This theory became known as the Fundamental Differences Hypothesis. Age-related cognitive factors and second language acquisition There are several factors that could contribute to the difficulty in learning a second language with increasing age. Park (2000) suggests that the three main deficits that occur with cognitive aging are decreases in working memory, processing speed, and inhibitory control. Decreased suppression makes it more difficult to think in the second language instead of relying on one’s first language. If these factors do influence second language acquisition, there will be a lot of variability in older peoples’ ability to learn a second language because there is a wide range in the severity of deficits related to cognitive aging. Working memory is invaluable in learning a second language because languages involve the learning of sequences, and working memory is responsible for the short-term rehearsal of sequences and the consolidation of languages (Ellis and Sinclair 1996). By studying English learners of Welsh, Ellis and Sinclair found that subjects who were forbidden to rehearse the words did not achieve the same level of grammatical judgment as learners who were allowed to rehearse. Daneman and Case (1981) examined the role of short-term memory on language learning in children. Using an artificial language, they were able to show that children with higher scores on short-term memory tasks were more successful at picking up the second language. More importantly, the researchers discussed the possibility of applying their results to aging, working memory, and language acquisition. Cognitive processing speed and attention are two other characteristics affected by age that may play a role in the difficulty of adult second language acquisition (Kemper 1992). Attention is important in learning a second language because learning a language means learning Aging and Acquisition of Latin 8 complex sequences that are hidden in grammar (Cohen et al. 1990). While examining sequence learning in adults, Cohen et al. found that only simple sequences could be learned in the presence of distraction, whereas complex sequences require the full attention of the learner. With the decreases in attention that accompany aging, learning the complex sequences in grammar makes learning a second language far more difficult. Birdsong (2006) noted the importance of dopamine in learning a second language. The decrease in dopamine levels throughout a person’s lifetime can increase the difficulty in learning a second language. The decrease in dopamine throughout one’s lifetime is related to deficits in higher cognitive functions such as working memory and episodic memory. In addition, Dopamine is important in language acquisition, language processing, and motivation. The decrease in dopamine would be most notable in the functioning of the procedural system and would make it increasingly difficult for someone to learn the grammatical structure of a second language. Several studies pinpoint the onset of the decrease in dopamine receptors around the age of 20 (Li et al. 2001; Volkow et al 1998). It is possible that the decrease in dopamine receptors after this age is related to the linear decline of success in language acquisition with increasing age. Ullman (2005) considered factors that may make grammar learning more difficult among certain age groups. For example, an increase in estrogen during adolescence enhances declarative memory while inhibiting procedural memory, thus making it more difficult for second language learners to pick up the grammar of the second language. Decreased rule abstracting abilities in older adults due to decreases in working memory may also make grammar acquisition more problematic. Age-related motivational and strategic factors and second language acquisition Aging and Acquisition of Latin 9 There are several behavioral reasons for the increased difficulty in learning a second language with aging. Singleton (2001) proposed a reason for the decreased success in second language acquisition for adult learners. Adults who immigrate to a new country can make the conscious choice to associate themselves with other adults who speak the same language as them, thus decreasing their exposure to the new, native language. This makes it difficult to interpret the results of studies focusing on age of immigration and ultimate attainment of a second language. Similarly, Marinova-Todd et al. (2000) found that motivation plays a significant role in adult second language acquisition. Behavioral factors may account for the variability in ultimate attainment among adult second language learners. While the ultimate level of attainment has been of interest to researchers, they are also interested in learning the strategies through which adults learn a second language. Liu, Bates and Li (1992) examined the strategies native Chinese speakers used to learn English. While children under the age of 13 used the same strategies as monolingual English speakers to learn the language, adults over the age of 20 applied Chinese language strategies to their studies of English and were less successful in their acquisition. The strategy that was witnessed by adults over the age of 20 was forward transfer: English learners of Chinese transferred English-like word order into Chinese sentences and Chinese learners of English transferred Chinese animacybased strategies to the new language. Younger learners were able to apply the correct strategy to the correct language. This suggests that after a certain age, adults use a fundamentally different strategy in learning a second language. Aging and strategies for learning a second language Midford and Kirsner (2005) used Reber’s (1967) artificial grammar paradigm to examine explicit and implicit methods of language learning in adults with a mean age of 65.9 years and Aging and Acquisition of Latin 10 young adults with a mean age of 20.6 years. In his study, he varied age (young or old), type of instruction (explicit or none), and complexity in grammar. Explicit instruction consisted of telling the participants to look for the underlying pattern of the artificial grammar, which they were later asked to describe. He assumed that the complex grammar with no instruction condition would lead to the most implicit type of learning, whereas the simple grammar with explicit instruction would lead to the most explicit type of learning. Midford found that the older group was least disadvantaged in the most implicit condition and most disadvantaged in the simple grammar, particularly when explicit instruction was given. The older group showed deficits in all conditions, but they were the least disadvantaged in the most implicit condition. Both groups were disadvantaged when explicit instruction (as compared to implicit) was given with the complex grammar, suggesting the complex grammar of languages is best acquired implicitly. The current experiment will attempt to replicate the findings of Midford using a real language and adults over the age of 65. Participants will be taught Latin grammar with either implicit or explicit feedback and will be tested on their ability to comprehend and create correct grammar. Their retention will be observed by including a follow-up session one week after the grammar is taught. It is predicted that older adults will learn more from Implicit than Explicit feedback. Methods Participants Participants were 20 older adults (11 Female, 9 Male) who had previously participated in an experiment in the Cognitive Aging Lab and indicated that they would be interested in participating in another experiment. They ranged in age from 66 to 81, with a mean age of 72.3 Aging and Acquisition of Latin 11 years. All were native English speakers and were not fluent in any other language. Participants were assigned to one of two treatment groups: the implicit feedback group or the explicit feedback group. A summary of the participants in each group is available in Table 1. To ensure the assignment was random, the treatment to which each participant was assigned alternated for every person who came in for his or her first session. The two groups were matched for gender and previous background in Latin and any other foreign languages. This older adult group was compared to a group of 20 Georgetown students (7 male, 13 female) who had previously completed the Latin learning task. Participants in this group were 18 to 21 years old, with a mean age of 18.7 years. All were native English speakers and were not fluent in any other language. These participants were recruited from introductory Spanish courses and received extra credit in their class for their participation. Apparatus The three sessions of the experiment were completed on a computer. During the learning and testing phases for each session, the participants wore headphones to listen to the pronunciations of Latin words as they were spoken. A paper questionnaire was included to assess the strategies the participants used while learning Latin. The program included vocabulary, grammar training, and testing phases. The computer program used was created for The Latin Project©. Procedure Each participant completed three test days of the experiment. For the older adults, the first two test days occurred one to two days apart and the third test day occurred seven days after the second. Aging and Acquisition of Latin 12 On their first test day, participants completed an online questionnaire that assessed their background in Latin and other languages. This included questions about coursework in and exposure to foreign languages in different settings. Vocabulary training. After the questionnaire, each subject received a Latin vocabulary lesson, until they reached a criterion, as described below. The lesson consisted of 35 nouns and 11 verbs, which are listed in the appendix. During the lesson, a picture appeared on the screen depicting one of the nouns or verbs. Then, a Latin stem with four different, gender-appropriate endings (singular subject, singular object, plural subject, plural object) appeared on the screen as a narrator read each one aloud. Finally, the English translation of the picture appeared on the screen. For the verbs, only two endings (singular and plural) were presented. Each participant spent the same amount of time on the words because the timing of the presentation of each item on the screen was set in the computer program. After the subjects encountered the 35 nouns and 11 verbs once, they were quizzed on each item. During the quiz, an English word was presented and participants were asked to choose between three Latin word stems (without any endings). If they did not reach a 60% accuracy level, they were automatically prompted to go through the vocabulary lesson again until they reached that threshold level. If they reached threshold, they were automatically quizzed again on vocabulary that they had gotten wrong during the first quiz. Participants were permitted to proceed to the grammar training only after they achieved a 100% accuracy level. Grammar Pretest. Once they reached the 100% accuracy level on the vocabulary, the computer led the participants to a pretest that was administered to determine their baseline understanding of Latin grammar. This test was divided into four components to assess visual and auditory learning. During the testing phase, no feedback was given. Prior to each of the four Aging and Acquisition of Latin 13 tests, there was a brief vocabulary review in which a Latin word appeared and participants chose between two English equivalents. Participants were only allowed to move on to each of the four tests once they reached 100% accuracy level on the brief vocabulary review In the first section of the pretest, the Written Interpretation test, a picture appeared on the screen with two people in costume, one of whom was performing an action on the other. An example of this display is included in the appendix. Two Latin sentences appeared on the screen with reversed subject and object assignment and the participants pressed a button to indicate which sentence was the correct representation of the picture. There were 12 questions in this format and eight distracter questions in which one of the sentences contained nouns that were not in the picture. These questions were easy to answer if the participants understood the vocabulary, even if they did not know the grammar. For this reason, these items were not scored. Participants were allowed to choose an option indicating they did not know the answer if they chose to do so. The directions from this section of the test are included below: Read the Latin sentence and choose the correct picture by pressing the A and L keys or the space bar. Place your index fingers on those keys and your thumbs on the space bar. Press A if you think the picture on the left corresponds to the sentence, press L if you think it’s the picture on the right, and press the space bar if you think neither of the two is a correct response. Try to answer as quickly as you can. In the second section of the pretest, the Aural Interpretation test, two pictures appeared on the screen with one person in costume performing an action on another. The subject and object assignment in the two pictures was reversed. A Latin sentence was read aloud and the participants pressed a button to indicate which picture was correctly expressed by the sentence. Aging and Acquisition of Latin 14 As in the Written Interpretation test, there were 12 scored questions (half new, half that are presented in the grammar training sessions) and 8 distracter questions that were not scored. Once again, participants could indicate if they did not know the answer. The instructions from this segment of the test are included below: Listen to the Latin sentence and choose the correct picture by pressing the A key, the L key, or the space bar. Place your index fingers on A and L keys and your thumbs on the space bar. Press A if you think the picture on the left corresponds to the sentence, press L if you think it’s the picture on the right, and press the space bar if you think neither picture represents the sentence you hear. Click the ‘sound’ button when you are ready. You will hear each sentence only once. In the third pre-test, the Grammar Judgment test, a Latin sentence of correct or incorrect grammar appeared on the screen and participants used the keyboard to indicate whether the sentence was grammatical or ungrammatical. Grammatical sentences always included a subject and a verb and sometimes included an object. Ungrammatical sentences had two objects, two subjects, no subject, or an incorrect form of the verb. There were 12 scored questions and 8 distracter questions. Participants could choose an "I don't know" option. The instructions of this section are included below: Decide whether or not each sentence is correct in Latin by pressing the A or L key or the space bar. Place your index fingers on those keys and your thumbs on the space bar. If you think the sentence is RIGHT press A. If you think the sentence is WRONG, press L. Press the space bar if you cannot make up your mind. Aging and Acquisition of Latin 15 In the fourth pre-test, the Sentence Production test, the participants were to create Latin sentences that represented a picture on the screen. The roots of the subject, verb, and object (if there was an object) were provided and there were two word banks of Latin endings (one for nouns, one for verbs). The participants were instructed to drag the correct endings to the roots into their proper place to create a sentence that represented the picture. Ten items were scored (half new, half that are presented in the grammar training sessions) and there were only 5 distracters in this test that were not scored. The distracters were the items that contained only a subject and a verb, but no object. There was no "I don't know" option on this test. The instructions for this test appear below: Make a sentence in Latin that describes what is happening in the picture. Use the vocabulary cues provided by dragging them to the box below the picture. You can move around all items as long as you do not press ‘enter’. Press ‘enter’ when you are done. Try to work through each item as fast as you can. Grammar training. On their second test day, participants were randomly assigned into one of the two grammar training groups, the explicit group or the implicit group. During this test day, participants received training in Latin grammar. There were six different phases of the grammar training that provided either written or aural input. In the first phase, participants were asked to choose between two English translations when a Latin sentence was presented on the computer screen. In the second phase, participants chose between two pictures to indicate the meaning of the Latin sentence. In the third phase, participants saw a picture and chose one of two Latin sentences that best represented the picture. During the fourth phase, participants listened to a Latin sentence and chose one of two English translations for that sentence. In the fifth phase, Aging and Acquisition of Latin 16 participants listened to a Latin sentence and chose one of two pictures that best represented the information conveyed in the sentence. In the sixth and final phase, participants had to indicate whether a picture on the screen represented a sentence they heard. For all of the grammar training phases, after they entered their response, they were told if they were correct or incorrect. They were given unlimited time to enter their response but were instructed to respond as quickly as possible. The Implicit group received only statements indicating whether they were correct or incorrect, whereas the explicit group received this feedback as well as a sentence that described the reason they were wrong. The Explicit group also received explanations when their answers were correct. An example of the difference between the Explicit and Implicit feedback for an incorrect response is: Explicit Feedback: Oops! -us is a subject ending, and -um is an object ending. Implicit Feedback: Oops! That’s incorrect! Every once in a while, a vocabulary quiz appeared on the screen. During this quiz, a Latin word was presented and participants had to choose between two English options. They were only allowed to continue when they reached a 100% accuracy level on the vocabulary. Subjects were allowed to continue the grammar training only after they answered each vocabulary item correctly. The subjects completed two grammar training sessions of the six phases, each lasting approximately 30 minutes. Grammar testing. Following the grammar training sessions, the participants completed a post-test that was identical to the pretest they had completed on the first test day. While the format was identical, the sentences were changed from the previous session. In these tests, half of the sentences had been presented during the grammar training phase, whereas half of the items were new. The distracters were still present in the tests. Aging and Acquisition of Latin 17 One-week-delayed grammar post-test. On their third test day, conducted one week after the second test day, subjects repeated the four tests that they received on the following two test days. As before, the sentences were changed to ensure they had the opportunity to apply their knowledge of Latin grammar to new stimuli. The three versions of the grammar-tests were counterbalanced between participants automatically so there was no effect of order on learning. In addition, the order of the items within each of the four subtests was randomized. After each of the three test days, participants were asked to elaborate on strategies they used while learning the vocabulary and grammar. This provided qualitative information about how they approached the learning of a second language. Results Reported Strategies for Learning the Vocabulary in the Older Group The majority of the older participants reported that they relied on the resemblance to English words to learn the Latin vocabulary. For example, participants easily learned that “Medica” corresponded to “Doctor.” Sometimes, the relationships between the English and Latin words were not as clear and participants formed their own connections. For example, one participant explained she was able to remember the verb “Auscutare,” which means “to listen,” by relating it to another word related to sound: “Acoustics.” Almost every subject relied on similarities to English words and a few indicated this allowed them to focus mainly on memorizing the words that did not look similar to their English equivalents. Three participants spoke out loud while learning the vocabulary. These participants all learned the vocabulary in one try and indicated that speaking out loud helped them remember the words. Aging and Acquisition of Latin 18 About half the participants indicated they paid attention to the endings of the words. Some focused on the genders of the endings, while others focused on whether the endings were plural or singular. No participants mentioned that the endings of the words signified whether the word would be the subject or object in a sentence. One participant mentioned he recognized the endings from attending Mass in Latin, while a few mentioned they vaguely remembered the endings from previous coursework in Latin. Reported Strategies for Learning the Grammar in the Older Group Most of the participants indicated they had no trouble remembering the vocabulary they learned during the first test day. During this test day, most participants said they focused on the ending of the words. Participants in the implicit group were just as likely as the explicit group to mention the importance of the endings of the words, even though they were never told to focus on the word endings. No participant reported having an adequate grasp of the noun endings after training, while almost everybody felt they had an adequate grasp of the two verb endings. Two participants indicated that they began to remember certain endings from previous exposure to Latin in a classroom setting. A handful of participants mentioned that they did not think they performed any better on the postand delayed-tests than they did on the pre-test. Some said the training was confusing and others thought the feedback was presented too quickly. Written Interpretation in the Older Group The number correct out of 12 for the treatment groups (Implicit and Explicit) on the Written Interpretation test during the three test days is shown in Figure 1a. The group that received implicit feedback appears to have performed better than the group that received explicit feedback, but also appeared to be performing better at the baseline. However, a mixed design 2 x Aging and Acquisition of Latin 19 3 (treatment x test day) ANOVA on the number correct yielded only a main effect of test day F(2, 16) = 4.045, p =.0271, which indicated that there was overall learning over the three testing days. Aural Interpretation in the Older Group The number correct out of 12 for the two treatment groups on the Aural Interpretation test is shown in Figure 1b. Again, the group that received the implicit feedback appears to have performed better than the group that received explicit feedback, though the groups were equal on the pre-test and seemed to diverge over the next two testing days. Nonetheless, the ANOVA yielded no significant main effects or interactions Grammar Judgment in the older group The number correct out of 12 for the two groups on the Grammar Judgment test is shown in Figure 1c. While the group that received implicit feedback appears to have performed better than the explicit feedback group on the post-test, this difference seems to disappear on the delayed test. There was a main effect of test day F(2, 16)=4.644, p= .0170, which suggested there was overall learning on the task. There was also an interaction between treatment and test day F(1,2) = 3.323, p= .0488. To examine the nature of the interaction, a t-test was performed for each test day to assess the differences between the two treatments. The implicit treatment yielded significantly higher scores during the post-test, t(18) = 2.304, p=.0334 but not during the preor delayed tests. This suggests the implicit group showed more initial learning on being able to judge the grammaticality of Latin sentences, though the difference did not persist in the delayed-test. Sentence Production in the older group Aging and Acquisition of Latin 20 The number correct out of 30 for the two groups on the Sentence Production test is shown in Figure 1d. The group that received implicit feedback appears to have performed better on the post-test and delayed test than the group that received explicit feedback. There was a main effect of test day, F(2, 14) = 3.660, p=.0387, showing there was overall learning throughout the three testing sessions. The main effect of treatment type (Implicit or Explicit) was not significant, but a t-test indicated that the implicit feedback group performed significantly better than the explicit group on the delayed test, t(16)=2.188, p=.0439, but not on the preor post-tests. Combining the Measures The total number correct for each participant in each test was converted into a percentage and the percentages of the four tests were averaged to create a combined measure that represented overall performance on the four tests. To ensure the validity of this, correlations were performed between each of the tests and the combined measure on each of the test days (Table 2). Overall performances of the older participants are shown in Figure 1e. This graph follows the pattern of the four individual graphs, with the implicit feedback appearing to yield more learning than the explicit feedback. There was a significant main effect of test day, F(1,2)=7.822, p=.020, showing overall learning across the four tests. Although the main effect of treatment was not significant, a t-test indicated that the participants in the implicit feedback group performed significantly better overall than the participants in the explicit feedback group, t(56)=2.458, p=.0171. When broken down by treatment, only the implicit feedback group showed a main effect of test day F(2,8)=6.969, p=.0067 and was therefore the only group to show significant learning. Previous Latin Experience in the Older Group Aging and Acquisition of Latin 21 Because some of the participants (n=5 in the Implicit group, n=4 in the Explicit group) indicated that they had previous coursework in Latin (n=11 did not have previous coursework in Latin), the effects of their Latin experience on their overall performance was assessed. It was clear that previous Latin experience was advantageous in both treatment groups, as seen in Figure 2. There was a main effect of Latin experience, F(1,16)= 12.337, p=.0034, suggesting previous Latin experience enhanced performance on this task. There was also a significant interaction between Latin experience and test day, F(1, 2) = 4.902, p=.0149, and it appeared that overall, those with previous Latin coursework improved more than those who did not. Even though no main effects or interactions with treatment were significant, Figure 2 suggests that treatment mattered more for those without Latin experience than for those with previous Latin coursework. Using a t-test, it became evident that while the Implicit treatment was better overall than the explicit treatment in adults without Latin background, t(30) =2.528, p=.017, the treatment type was not significant for adults who had previous Latin experience. When the results were further split by the test day, the differences between the two treatments were no longer significant. Learning in the Young Group Figures 3a-d show the performances of the younger group on the four tests over the course of the experiment. Neither treatment was significantly better than the other on the four tests during the three test days. Performances on the four tests were converted to percentages and an average was taken across the four tests to give a measure of overall performance. Correlations were performed to ensure each of the four tests at each test day was related to the combined measure and these results can be seen in Table 3. Performances were analyzed using the same methods described Aging and Acquisition of Latin 22 earlier with the older participants. Young participants who received the explicit treatment showed a main effect of test day on the Written Interpretation test (F(2, 18)= 5.004, p=.0187), and on the Grammar Judgment test (F(2,18)= 6.695, p= .0077) and this effect was marginally significant for the Sentence Production test (F(2,18)= 3.143, p=.0675). Participants who received the implicit treatment showed a significant test day effect only on the Grammar Judgment test, F(2,18) = 5.812, p=.0113. Overall performances by the younger group on the four tests over the course of the experiment are seen in Figure 3e. The figure suggests that the explicit feedback provided a greater amount of learning in the younger adults, which is in contrast to the pattern seen in older adults. An ANOVA revealed there was a main effect of test day in the explicit feedback group F(2,9)=11.583, p=.00066 while there was only marginally significant learning in the implicit feedback group F(2,9)=3.268, p=.0616. Age and Learning Given that for the older people previous Latin exposure influenced performance on the four tests and the fact that none of the younger participants had exposure to Latin prior to this experiment, only older adults without previous Latin coursework were used in the analysis of age differences. Figure 4 shows the overall learning of older and younger participants without previous Latin experience in both of the treatment groups. There was no main effect of age, showing that there were no age deficits in this experiment. There was a main effect of test day, F(2,26)= 8.468, p=.0007, showing that there was overall learning among participants without previous exposure to Latin. A t-test was performed to assess the overall effect of treatment on each age group. While neither treatment yielded significantly better results in the younger group, older adults who had the implicit training performed significantly better overall than those with Aging and Acquisition of Latin 23 explicit training, t(30) =2.528, p= .017. However, when broken down by the test day, this treatment effect in the older group is not significant. With the small amount of participants without Latin experience, the main effect of test day in the Implicit group is only approaching significance, but this is not the case in the Explicit group. Discussion Summary of Results The hypothesis that the implicit feedback would be better for teaching grammar to the older participants was supported. While only one measure (Grammar Judgment) resulted in the predicted treatment by test day interaction in the older group, it is evident from the individual and combined graphs that the implicit feedback was more helpful to the older adults than the explicit feedback. This holds true especially when the adults with previous Latin experience (n=9) are removed from the analysis, as shown in Figure 4. With more participants in the group of older adults without Latin experience, it is possible that the differences between the Implicit and Explicit treatments could have become significant during the postand delayed-tests. In addition, the small sample size for adults without previous Latin background reduced power. However, the graphs show a general trend of learning throughout the experiment. In the younger group, the opposite pattern is occurring. In the individual and combined graphs, it would appear that the explicit feedback is more helpful to the younger adults than the implicit feedback. While there were no significant effects of treatment in the young group, it is possible this is due to a lack of power. Overall, the graphs support Midford and Kirsner’s (2005) findings that the older group would fare better in the most implicit form of the task, whereas the young would learn the most in the more explicit form of the task. However, the age deficits seen by Midford and Kirsner are not as pronounced in the current study. Aging and Acquisition of Latin 24 The fact that there was significant overall learning in the older group, and that it was not significantly less than in the young, supports the notion that older adults are still capable of learning certain elements of language (Birdsong 1992, Marinova-Todd et al. 2003). While this experiment certainly did not provide prolonged exposure to a second language, the ability to learn a small amount of Latin in a short period of time is still noteworthy. Possible Explanations for Results The fact that previous Latin experience affected the older adults’ pre-test performance (Figure 2) and ability to learn the Latin vocabulary and grammar is in agreement with Bahrick’s (1984) study of people who received high-school and/or college instruction in a Spanish course. Through a regression analysis, he examined the recall and recognition of adults who were 1-50 years beyond the end of their Spanish instruction. Reading comprehension, recall, and vocabulary and grammar recognition were related to the level of initial instruction and the grades received in the Spanish course. Therefore, Bahrick argued there existed a 50-year “permastore” for adults who studied the Spanish language in school with little to no rehearsal after the end of school instruction. This was particularly interesting in light of the present experiment because the participants who took Latin claimed they remembered little to nothing of the language they studied approximately 50 years before. While they may not have been able to remember anything before this experiment, it became clear that they were being reminded of their previous Latin training during the experiment. For this reason, it would have been unfair to do an age comparison including the older adults with previous background in Latin, since their baseline performances were higher than the younger participants’. The fact that there were no age deficits even when those with previous Latin experience were removed was particularly interesting (Figure 4). This contrasts with a lot of the literature in Aging and Acquisition of Latin 25 second language acquisition and aging (Singleton 2001, Birdsong 2006, Hyltenstam and Abrahamsson 2000). There are several possible reasons for this result. First, the older adults may have been more motivated to do well on the tests. Younger participants were recruited from Introductory Spanish courses and received extra credit for their participation. On the other hand, the older participants were recruited from past experiments in the lab and had indicated that they were interested in coming back for more studies. Throughout the three test days, most of the older participants were very interested in learning about their progress and asked when the final results would be available. This is interesting in light of Marinova-Todd et al.’s (2000) research on motivation and success in learning a second language. It is possible that the motivation of the older adults to perform well made up for any age-related deficits in learning processes. Another reason for the lack of age deficits may be that older adults have had more exposure to foreign languages over their lifetime than the younger adults. This could have particularly helped them to learn the vocabulary because they were able to form associations between words in different languages. There are a couple of experimental variables that could have prevented age deficits from appearing. First, the timing of the pre-, post-, and delayed-tests was different in the older and younger group. Whereas the older group only had one to two days between the preand posttests, the younger group had one week. There is no vocabulary review before the grammar training, and so it is possible that the younger group had simply forgotten the vocabulary by the second day. Another difference between the two groups in terms of testing is that the younger participants completed the experiment in a group setting, whereas the older participants were in a room by themselves. The group atmosphere may have prevented the younger group from engaging in advantageous strategies for learning a language, such as speaking aloud. Aging and Acquisition of Latin 26 Nonetheless, the current study bodes well for older adults who want to learn a language and suggests that implicit methods of learning a language may yield the best results. The fact that older and younger adults appear to be learning preferentially to different types of feedback is particularly interesting in light of a discussion by Craik and Bialystok (2006). They stated that while older adults are still able to represent information as well as young adults, their ability to access certain representations decreases with aging. In language, older adults do not necessarily show deficits in their ability to speak, but they have difficulty accessing words that represent specific ideas. Therefore, while older adults may have benefited from the explicit feedback in the current experiment, their ability to access what they had just learned may have been impaired by the time they were tested on the information. For this reason, the implicit feedback may have been more beneficial because the older adults did not have to consciously access any new information they were taught. Differences in cognitive control may have accounted for the different types of learning in the older and younger adults. Experimental Limitations The main limitation of this experiment was the small sample size of adults who have not had Latin coursework in the past (n=11). Future research involving aging and this Latin task should increase the sample size of adults who have not had previous coursework in Latin. The most common complaint about the task from the group of older participants was that the pictures were too small, dark, and hard to make out. Some said it was hard to determine who was performing the action, while others were confused by the difference between similar-looking characters such as “Woman” and “Maiden” during the vocabulary training. To amend the task in the future, it would be advantageous to pick vocabulary words that are easily distinguishable from each other and make the pictures more clear. Aging and Acquisition of Latin 27 Another common complaint that came from older adults in the explicit training group was that the explicit feedback was presented too quickly to be read completely. While the purpose of this was to match the timing between the implicit and explicit group, future research could include a more self-paced version of grammar training for the older adults to determine whether the implicit feedback condition is still better under self-paced conditions. This way, participants could have time to absorb the feedback before moving on to the next item. In the future, it will be important to see if the advantage of the implicit feedback remains over a longer period of time. Despite these complaints, which were brought forth by the older adults who were therefore more affected, the older adults were still able to demonstrate learning with no age deficit. Conclusion This study has several implications for the field of aging and second language acquisition. It will be important to determine the best method for teaching a second language and to study the success of different teaching methods over a prolonged period of time. Furthermore, it would be interesting to conduct future studies on early, limited exposure to a second language and its effects on the ability of a person to learn that language later in life. The results suggest that older adults are capable of learning some basic grammar rules with limited exposure to a second language, which is encouraging for adults who are motivated to become bilingual. Aging and Acquisition of Latin 28 ReferencesBahrick, H. P. (1984). Semantic memory content in permastore: Fifty years of memory forSpanish learned in school. Journal of Experimental Psychology: General, 113, 1-29.Birdsong, D. (2006). Age and second language acquisition and processing: A selective overview.Language Learning, 56, 9-49.Birdsong, D. (1992). Ultimate attainment in second language acquisition. Language, 68,706-755.Bley-Vroman, R. (1990). The logical problem of foreign language learning. Linguistic Analysis,20, 3-49.Bongaerts, T., Mennen, S., van der Silk, F. (2000). Authenticity of pronunciation in naturalisticsecond language acquisition: The case of very advanced late learners of Dutch as asecond language. Studia Linguistica, 54, 298-308.Cohen, A., Ivry, R., & Keele, S. (1990). Attention and structure in sequence learning. Journalof Experimental Psychology: Learning, Memory, and Cognition, 16, 17-30.Craik, F., & Bialystok, E. (2006). Cognition through the lifespan: Mechanisms of change.Trends in Cognitive Sciences, 10, 131-138.Daneman, M. & Case, R. (1981). Syntactic form, semantic complexity, and short-term memory:Influences on children’s acquisition of new linguistic structures. DevelopmentalPsychology, 17, 367-378.Ellis, N. & Sinclair, S. (1996). Working memory in the acquisition of vocabulary and syntax:Putting language in good order. Quarterly Journal of Experimental Psychology: SectionA, 49, 234-250.Flege, J., & Yeni-Komshiam, G. (1999). Age constraints on second language acquisition. Aging and Acquisition of Latin 29 Journal of Memory & Language, 41, 78-104.Gebauer, G., & Mackintosh, N. (2007). Psychometric intelligence dissociates implicit andexplicit learning. Journal of Experimental Psychology: Learning, Memory &Cognition, 33, 34-54.Hakuta, K., Bialystok, E., & Wiley, E. (2003). Critical evidence: A test of the Critical-PeriodHypothesis for second-language acquisition. Psychological Science, 14, 31-38.Hyltenstam, K. (1992). Non-native features of near-native speakers: On the ultimate attainmentof childhood L2 learners. In R. Harris (Ed.), Cognitive processing in bilinguals (pp351-368). Amsterdam: Elsevier.Hyltenstam, K. & Abrahamsson, N. (2000). Who can become native-like in a second language?All, some, or none? Studia Linguistica, 54, 150-166.Johnson, J. & Newport, E. (1989). Critical period effects in second language learning: TheInfluence of maturational state on the acquisition of ESL. Cognitive Psychology, 21,60-99.Li, S-C., Lindenberger, U., & Sikström, S. (2001). Aging cognition: From neuromodulation torepresentation. Trends in Cognitive Sciences, 5, 479-486. Liu, H., Bates, E., & Li, P. (1992). Sentence interpretation in bilingual speakers of English andChinese. Applied Psycholinguistics, 12, 451-484.Marinova-Todd, S., Marshall, D. & Snow, C. (2000). Three misconceptions about age and L2learning. TESOL Quarterly, 34, 9-34. McGeorge, P., & Crawford, J.R. (1997). The relationships between psychometric intelligenceAnd learning in an explicit and an implicit task. Journal of Experimental Psychology:Learning, Memory & Cognition, 23, 239-245. Aging and Acquisition of Latin 30 Midford, R. & Kirsner, K. (2005). Implicit and explicit learning in aged and young adults.Neuropsychology & Cognition, 12, 359-387.Park, D.C. (2000). The basic mechanisms accounting for age-related decline in cognitivefunction. In D.C. Park & N. Schwarz (Eds.), Cognitive aging: A Primer (pp 3-21).Philadelphia: Psychology Press.Penfield, W. & Roberts, L. (1959). Speech and brain mechanisms. Princeton, NJ: PrincetonUniversity Press.Reber, A. (1976). Implicit learning of synthetic languages: the role of instructional set. Journalof Experimental Psychology: Human Learning and Memory, 2, 88-94.Rossi, S., Gugler, M., Friederici, A., & Hahne, A. (2006). The impact of proficiency on syntacticsecond language processing of German and Italian: Evidence from event-relatedpotentials. Journal of Cognitive Neuroscience, 18, 2030-2048.Scovel, T. (1988). A time to speak: A psycholinguistic inquiry into the critical period for humanspeech. Rowley, MA: Newbury House.Singleton, D. (2001). Age and second language acquisition. Annual Review of AppliedLinguistics, 21, 77-89. Ullman, M. (2001). A neurocognitive perspective on language: The declarative/proceduralmodel. Nature Reviews Neuroscience, 2, 717-726.Ullman, M. (2005). ‘A cognitive neuroscience perspective on second language acquisition:The declarative/procedural model,’ in C. Sanz (Ed.), Mind and Context in Adult SecondLanguage Acquisition: Methods, Theory, and Practice (pp. 141-178), Washington, DC:Georgetown University Press.Volkow, N., Wang, G., Fowler, J., Ding, Y., Gur, R., Gatley, S. et al. (1998). Parallel loss of pre Aging and Acquisition of Latin 31 and postsynaptic dopamine markers in normal aging. Annals of Neurology, 44, 143-147. Aging and Acquisition of Latin 32 AcknowledgmentsThis study is part of The Latin Project©, developed to investigate the relationshipbetween bilingualism, cognition and the acquisition of new linguistic knowledge. All materials,including the treatment and test components, were developed by Cristina Sanz, Harriet Bowdenand Catherine Stafford with support from GU Graduate School grants to Sanz and assistancefrom Bill Garr of Georgetown’s UIS/CNDLS.Beatriz Lado tested the 20 younger participants described in this study. They participatedthrough the Department of Spanish and Portuguese at Georgetown University.I would like to extend a thank you to my thesis advisor Darlene V. Howard, and also toJames H. Howard Jr., Cristina Sanz, and Lauren Treadaway. I would like to thank Jessica Simon,Ilana J. Bennett, and Sunbin Song for their helpful suggestions. Aging and Acquisition of Latin 33 AppendixList of 35 nouns presented in the Vocabulary learning lesson with four endings: singular subject,singular object, plural subject, plural objectagricola agricolam agricolae agricolas FARMERbelua beluam beluae beluas MONSTERdeadeam deaedeasGODDESSdomina dominam dominae dominas MISTRESSfemina feminam feminae feminas WOMANinnupta innuptam innuptae innuptas MAIDENhospita hospitam hospitae hospitas GUESTlarvalarvam larvae larvas GHOSTmagistra magistram magistrae magistras TEACHERmedica medicam medicae medicas DOCTORnupta nuptam nuptae nuptas BRIDEparvula parvulam parvulae parvulas GIRLpoetria poetriam poetriae poetrias POETESSregina reginam reginae reginas QUEENsagasagam sagae sagas FORTUNETELLERturba turbam turbae turbas CROWDvenefica veneficam veneficae veneficas WITCHvetula vetulam vetulae vetulas OLD WOMANamicus amicum amici amicos FRIENDangelus angelum angeli angelos ANGELavusavum aviavosGRANDFATHERcoquus coquum coqui coquos COOKdeusdeum deideosGODdominus dominum domini dominos MASTERfamulus famulum famuli famulos SERVANTfunambulus funambulum funambuli funambulos ROPEWALKERlautus lautum lautilautos GENTLEMANmagus magum magimagos WIZARDmedicus medicum medici medicos DOCTORmusicus musicum musici musicos MUSICIANparvulus parvulum parvuli parvulos BOYpotentissimus potentissimum potentissimi potentissimos KINGprocus procum proci procos SUITORstultus stultum stulti stultos FOOLtignarius tignarium tignarii tignarios CARPENTER Aging and Acquisition of Latin 34 AppendixList of verbs that are presented in the vocabulary learning lesson.amare TO LOVEauscultare TO LISTENbasiare TO KISScogitare TO THINKindagare TO LOOK FORiuuare TO HELPlaudare TO PRAISEsalutare TO GREETspectare TO LOOK ATuocare TO CALLvisitare TO VISIT Format of the Written Interpretation test Format of the Aural Interpretation test Aging and Acquisition of Latin 35 AppendixFormat of the Grammar Judgment test Format of the Sentence Production test Aging and Acquisition of Latin 36 Table 1Description of older participants in the Implicit (a) and Explicit (b) feedback groups. 72.800 4.803 1.519 66.000 80.00029.333 .707 .236 28.000 30.00066.222 9.404 3.135 47.000 79.0009.600 2.914 .921 5.000 15.0006.111 1.900 .633 4.000 10.000Mean Std. Dev. Std. Error Minimum MaximumAgeMini-Mental State ExaminationWAIS-III VocabularyWMS-III Digit Span Forw ardWMS-III Digit Span Backw ards71.700 4.855 1.535 67.000 81.00029.500 .707 .224 28.000 30.00066.556 10.573 3.524 45.000 80.0008.400 1.265 .400 7.000 11.0006.500 1.900 .601 3.000 9.000Mean Std. Dev. Std. Error Minimum MaximumAgeMini-Mental State ExaminationWAIS-III VocabularyWMS-III Digit Span Forw ardWMS-III Digit Span Backw ardsa.