Dodson_Honors_Thesis_2006

THE PENNSYLVANIA STATE UNIVERSITY

SCHREYER HONORS COLLEGE

DEPARTMENT OF SPANISH, ITALIAN AND PORTUGUESE

INVESTIGATING THE SELECTION MECHANISM THAT FACILITATES

LANGUAGE PRODUCTION IN BILINGUALS

STEPHANIE R. DODSON

Spring 2006

A thesis

submitted in partial fulfillment

of the requirements

for a baccalaureate degree

in Spanish

with honors in Spanish

Reviewed and approved* by the following:

Judith F. Kroll

Liberal Arts Research Professor of Psychology and Linguistics

Thesis Supervisor

Priscilla Meléndez

Professor of Spanish

Honors Adviser

* Signatures are on file in the Schreyer Honors College.

2

Abstract

How do bilinguals select the language of the words they speak? While a bilingual

is speaking one language, are words from the other language active and competing for

selection? If so, some mechanism is required to control this unwanted activation in order

to facilitate production in the desired language. This study investigates the cognitive

processes that facilitate lexical selection in highly proficient bilinguals. Using semantic

blocking and language-switching paradigms, we address the question of whether the

selection mechanism entails active inhibition of the non-target lexicon, or whether only

items from the target language behave as candidates for selection.

Previous studies have shown that picture naming is slower when pictures are

blocked in the same semantic category (e.g., all fruits), and also when bilinguals are

required to switch languages from one trial to the next (Costa & Santesteban, 2004;

Damian, Vigliocco, & Levelt, 2001; Kroll & Stewart, 1994; Meuter & Allport, 1999). A

set of 80 pictures was selected on the basis of semantic properties, with ten items

corresponding to each of eight semantic categories. Relatively proficient bilinguals were

instructed to name each picture in Spanish or English, with background color functioning

as a response language cue. Stimuli were presented in alternating semantically blocked

and mixed series, with the cue to name in one language or the other alternating regularly

every two trials.

The results demonstrated robust effects of language switching that were reduced

in the context of semantically blocked lists. We consider the implications of these results

for models of language selection that assume that proficient bilingual speakers have

3

achieved a level of control in using their two languages that no longer requires active

inhibition of the more dominant language.

4

Table of Contents

Chapter I: Introduction…………..5

Semantic Competition in Monolingual Lexical Selection…………..7

Semantic Interference Paradigms…………..9

Lexical Selection in Bilinguals: Specific to One Language?.............13

Language Switching Paradigms…………..18

Combining Language switching and Semantic Blocking Paradigms………21

Chapter II: Experiment 1…………..23

Method………..23

Data Analysis………….27

Predictions………………31

Results…………..32

Discussion…………..37

Chapter III: Experiment 2…………..38

Method………..38

Results…………..39

Discussion…………..40

Chapter IV: General Discussion…………..41

Chapter V: References…………..46

Appendix A: Language History Questionnaire…………49

Appendix B: Stimuli from Experiments 1 & 2…………51

Appendix C: Academic Vita…………53

5

Chapter 1: Introduction

Bilingual speakers possess an extraordinary facility for choosing the right word in

the intended language during speech production. They can produce speech in one

language to the exclusion of the other, or intentionally switch back and forth between

lexicons within a given discourse. The latter act, known as code-switching, is a common

feature of bilingual communities and an oft-misunderstood phenomenon. Code-switching

is a cognitively demanding task that requires a great degree of control over both

languages. In recent decades, scholars have become interested in this behavior as a

vehicle for investigating how the mind processes language.

A common misperception about code-switching is that it denotes inadequate

proficiency in one or both languages (Hammink, 2000). In reality, proficient bilinguals

are capable of sticking to one language when conversing with monolinguals, but may

elect to code-switch in the presence of other bilinguals. Thus, code-switching does not

refer to the accidental blurting of a word or phrase in the wrong language, but rather to

the intentional mixing of languages among speakers who understand both.

Studies such as Poplack’s 1980 investigation of a Puerto Rican community in

New York City, entitled “Sometimes I'll start a sentence in Spanish y termino en

español,” have yielded scientific predictions about the nature of code-switching. Code-

switching constraints are descriptive, not prescriptive, meaning that they don’t instruct

bilinguals on how to code-switch, but rather predict the types of switches speakers are

likely to make. Poplack’s equivalence constraint states that intrasentential code-switching

will only occur at points in the sentence where syntactic structure and word order are the

6

same in both languages, resulting in a sentence that is grammatical in either language

(Poplack, 1978/81; 1980). Consider the following:

According to the equivalence constraint, bilinguals favor switches such as A, which

occurs between clauses and respects the syntactic rules of both languages. B, however, is

ungrammatical because it occurs between the clitic and the verb, a point where Spanish

and English have different word orders (“gave him” vs. “him gave”). A proficient

bilingual is unlikely to make a switch such as B because it would make little sense to

interlocutors, disrupting the process of communication (Poplack, 1978/81; 1980).

Highly proficient bilinguals have a greater capacity for code-switching than do

novice second-language learners because they are better able to control and maneuver

between their two lexicons and produce sentences that are grammatical in both

languages. The ability to shift languages at a moment’s notice indicates a high degree of

control over both lexicons. This high level of control is also evidenced by the fact that

bilinguals are able to confine their speech to one language when conversing with

monolinguals. Clearly, highly proficient bilinguals possess some mechanism that allows

them to select the correct word in the desired language during spoken discourse. The

7

present study investigates the nature of this mechanism and its implications for models of

speech production.

Semantic Competition in Monolingual Lexical Selection

Lexical selection is the process of choosing the word you want to say. Scholars

who study speech production characterize this process as a contest, in which many words

compete for selection and the most appropriate candidate wins. But how does the contest

work, and who are the competitors? To answer these questions, we must examine the

cognitive processes involved in speech production.

There are three distinct levels of representation involved in language processing

(see Caramazza, 1997; Costa, 2005; Francis, 2005; Levelt, 1989). The semantic, or

conceptual, level represents the concepts, or semantic information, that the speaker

intends to convey. The lexical level represents these concepts as words. Phonemes, or the

sound units that comprise words, are represented at the phonological, or sublexical, level.

Before lexical items may be accessed by the speech production system, the corresponding

representations must be activated, or made available for selection. A high level of

activation indicates a high degree of availability for production, while a low level of

activation indicates a small degree of availability (see Caramazza & Costa, 2001; Costa,

2005; Roelofs, 2001; Schriefers et al., 1991).

Currently accepted models of speech production assert that during semantic

processing, representations of semantically related concepts are activated simultaneously

with that of the target concept (see Caramazza, 1997; Costa, 2005; Damian et. al., 2001;

8

Kroll & Stewart, 1994; Levelt, 2001). In other words, when the concept apple is

activated, related concepts (pear, grapes, etc.) become activated simultaneously.

Figure 2 represents how a monolingual speaker produces the word “dog.” When the

speaker sees a picture of a dog, the semantic representation for dog becomes activated

along with representations for related concepts such as cat, horse, leash, and collar.

Activation flows from the semantic to the lexical level. The lexical representation with

the highest level of activation spreads to the phonological level, at which point the

speaker produces the spoken word “dog” (see Costa, 2005).

The higher the activation levels of the various lexical candidates, the more

difficult it becomes for the speaker to select the correct word (see Costa, 2005;

Caramazza & Costa, 2001; Roelofs, 2001; Schriefers et al., 1991). To investigate how the

9

lexical selection “contest” works, many researchers have conducted studies in which the

activation levels of semantic representations are manipulated, with the goal of observing

how semantic context affects the speaker’s ability to say the right word. In the next

section, we will discuss several of these studies and how they relate to the present

research.

Semantic Interference Paradigms

Experimental semantic interference paradigms have been used to support the

claim that semantically related concepts become activated along with the target concept

and behave as competitors during lexical selection. (e.g. Bloem & La Heij, 2003; Bloem,

Van den Boogaard, & La Heij 2004; Kroll & Stewart, 1994; Damian et. al., 2001; Stroop,

1935). Such paradigms, which manipulate the semantic context in which stimuli are

presented, are a useful tool for examining the mechanisms governing language

production in monolingual and bilingual speakers.

10

Stroop Task

Stroop (1935) devised a procedure in which a series of color words was printed in

colors different from those represented by the words. Participants were instructed to

name the color of the print while ignoring the word names. In Figure 3, for example, the

participant is shown the word “green,” which is printed in red text, and is asked to name

the red color while ignoring the written word “green.” She has trouble saying the right

word because she is distracted by the presence of another color word.

11

Stroop (1935) found that participants took longer to name colors printed in the

distracter words than to name the same colors printed as solid squares. This outcome

makes sense when viewed in the context of currently accepted models of language

processing. The participant in Figure 3 is receiving conceptual activation from two

sources: the color word “green” and the visual color red. Selecting the correct word is

more difficult in this context because the heightened activation levels make selection of

the target word “red” more difficult than it would be without the presence of a distracter

word.

Semantic Blocking Paradigms

Semantic blocking experiments such as the one used in the present study require

participants to perform picture naming or translation tasks while semantic context is

manipulated in such a fashion that stimuli are presented as part of a series of related or

unrelated items (see Damian et. al., 2001; Kroll & Stewart, 1994).

12

The goal is to observe how recent activation of semantically related items affects the

speaker’s ability to produce the appropriate response.

Kroll and Stewart (1994) reported higher naming latencies in the blocked

condition, a result which they used to support the claim that semantically related items

behave as competitors during lexical selection. By their logic, the speech production

system endures competition from other experimental items, which enjoy an increased

level of activation because the participant recently named them. When other items in the

group bear a semantic relation to the target item, the degree of competition is greater due

to the already higher activation levels of the related items. This competition impairs

performance, slowing naming latencies. For example, naming the picture “elephant” is

more difficult when the speaker must choose “elephant” over other animals he recently

13

named. When “elephant” is presented in a series of unrelated items such as “pear” and

“bicycle,” it is easier to choose the correct word because the unrelated items pose a lesser

degree of competition.

Lexical Selection in Bilinguals: Specific to One Language?

For someone who knows two or more languages, choosing the right word is even

more complicated than for someone who knows only one. A highly proficient bilingual

has two equivalent words for nearly every concept. For example, when a Spanish-English

bilingual wants to express the concept bicycle, she has two options: English “bicycle”

and Spanish “bicicleta.”

A bilingual must not only select the desired concept, but must also produce the right word

in the appropriate language. To accomplish this, the bilingual must employ some

14

mechanism that allows her to select words from one language to the exclusion of others.

The nature of this mechanism has been widely debated.

A few central questions arise when researchers try to answer the question of how

a bilingual selects the right word in the intended language. In the previous section, we

established that lexical selection is a sort of “contest” governed by the activation levels of

representations for potential candidate words. In the case of bilingual speakers, do

representations from both languages enter the contest? Is selection restricted to items

from the target language only? Does the system actively inhibit representations from the

language that is not in use? In the next section, we will discuss several models that have

been presented in the literature and how the present study will shed further light on the

nature of lexical selection in bilinguals.

Language Specific and Non-specific Models of Selection

Two models that figure prominently in the literature on bilingual language

production are the language specific and language non-specific hypotheses (see Calomé,

2001; Costa, 2005; Finkbeiner, Gollan, & Caramazza, in press). Both models seek to

answer the question of how the lexical selection “contest” works by predicting which

representations behave as competitors. The language specific hypothesis (see Figure 5-A)

states that although representations from both languages become activated during speech

production, only items from the target language behave as competitors. The language

non-specific hypothesis (see Figure 5-B), on the other hand, predicts that representations

from both languages behave as competitors.

15

Figure 5. Language Specific and Non-specific Models of Selection

Representations from both languages compete for selection

Language Non-specific Selection Model

Only representations from the target language compete for selection

Language Specific Selection Model

B.

A.

Both Images from Costa (2005)

16

If lexical selection is language non-specific, then some mechanism must be in

place to keep representations from the non-response language from competing for

selection. One possibility researchers have proposed is that this mechanism entails active

inhibition of the non-response language (e.g., Green, 1998; Meuter & Allport, 1999).

Inhibitory Control Model

Green’s (1998) Inhibitory Control Model (ICM) postulates a mechanism that tags

lexical representations belonging to the non-response language and prevents them from

17

competing for selection. The bilingual speaker in Figure 6 wants to say the word “chair”

in her L1, English. Representations for chair and semantically related candidates, both in

English and in the speaker’s L2, Spanish, receive activation at the lexical level. The

speech production system assigns each word a tag based on the language to which it

belongs. The system then inhibits lexical items from the non-response language—in this

case, L2 Spanish—while leaving words carrying the target language tag free to compete

for selection.

Language Switching Paradigms

Not only do bilinguals possess the ability to select lexical items from the

appropriate lexicon during speech production; they are also able to intentionally switch

back and forth between languages within a given discourse. Language switching

paradigms have previously been used as a tool to investigate the processes underlying

language selection in bilinguals (Costa & Santesteban, 2004; Meuter & Allport, 1999).

A typical language switching experiment is illustrated in Figure 7. The participant

is instructed to name pictures that appear on a computer screen, one at a time, while

alternating language every other trial. Background or picture color function as a response

language cue (in the above example, pictures with red backgrounds are named in English,

while those with black backgrounds are named in Spanish).

18

A language switching paradigm generates four subtypes of critical trials, which

are summarized in Table 1:

Table 1. Trial Types Generated by a Language Switching Paradigm

L1 SWITCH Trials performed in the participant's

dominant language (L1), for which the

previous trial was performed in the non-

dominant language (L2)

L1 NON-SWITCH Trials performed in the participant's L1, for

which the previous trial was also

performed in L1

L2 SWITCH Trials performed in the participant's L2, for

which the previous trial was performed in

the other language

L2 NON-SWITCH Trials performed in the participant's L2, for

which the previous trial was also

performed in L2

Reported naming latencies are generally higher for switch versus non-switch trials, a

result which suggests that there is a cost to production as the system accommodates the

19

language change (Costa & Santesteban, 2004; Meuter & Allport 1999). Meuter and

Allport (1999) reported that for non-balanced bilinguals (those with a clearly dominant

L1), the magnitude of switch cost was greater when switching into dominant L1 than

when switching into weaker L2.

The authors attributed this asymmetry to the relative levels of suppression needed to

switch between languages, asserting that the bilingual speaker must activate and suppress

each language as a whole unit. The stronger a bilingual's ability in a given language, the

greater the effort needed to suppress it; hence, the higher cost to switching into dominant

L1.

If L2 learners exert differential levels of suppression for L1 and L2, this begs the

question of whether switch cost asymmetry would be absent for bilinguals with near-

Image from Meuter & Allport (1999)

20

equal abilities in two languages. Costa and Santesteban (2004) reported switch cost

asymmetry for L2 learners, but not for highly proficient Spanish-Catalan bilinguals.

In Barcelona, where the experiment was conducted, children speak Spanish and Catalan

from early childhood. These bilinguals are therefore more balanced than those who learn

a second language later in life. Surprisingly, these highly proficient bilinguals had

symmetrical switch costs not only when naming in L1 and L2, but also in L1 and a much

weaker L3. The authors used these results to assert that the very nature of the selection

mechanism is different for balanced and non-balanced bilinguals.

Combining language switching and semantic blocking paradigms

The present study seeks to investigate how bilingual language production is

affected by simultaneous manipulation of semantic and linguistic context. For this

purpose, we constructed a combined language switching and semantic blocking

paradigm, in which participants name pictures presented in alternating semantically

21

blocked and mixed lists while switching language every other trial. The new paradigm

generates eight types of critical trials, which are illustrated in Figure 10:

Half the trials are performed in the dominant language (L1) and half in the non-dominant

language (L2). For each language, one subgroup consists of trials where a language

switch occurs, while the other contains non-switch trials. Of the switch and non-switch

trials, half are presented in the semantically blocked condition and half in the mixed

condition.

The reasons for this combined paradigm are multifold. In the first place, we will

have the opportunity to observe whether the semantic interference effect interacts with

switch costs during picture naming in L1 and L2. We will also have the opportunity to

see how the effects differ between our two groups of bilinguals: balanced (no dominant

L1) and L1 dominant (the L1, English, is clearly stronger). Our goal is to use the results

to gain a better understanding of the cognitive processes governing lexical selection in

22

bilinguals--specifically, to address whether the selection mechanism entails active

inhibition of the non-target lexicon and whether selection is specific to just one language.

We will also have the opportunity to ascertain whether the selection mechanism is

different for balanced and non-balanced bilinguals.

Chapter 2: Experiment 1

Method

Participants

Participants were recruited from the Penn State community and paid for their

participation. All participants spoke Spanish and English. Mean age was 21 years, with

18 being the youngest and 28 the oldest. Twenty-seven were female and fifteen were

male. All had normal hearing and normal or corrected vision.

Participants were originally grouped on the basis of their native language (the first

language they learned, or the language they spoke with their parents while growing up).

However, because most participants were currently attending an English-speaking

university, it was necessary to reassess participants' language dominance on the basis of

self-ratings and performance in the respective languages. Furthermore, some participants

had been born to Spanish-speaking parents but were raised in the United States and

attended English-speaking schools, resulting in their so-called “native language” being

the weaker language and English the dominant language.

23

Participants were re-grouped into two categories based on their language

dominance. Participants in the balanced group were those who enjoyed nearly equal

proficiency in both languages, with no clearly dominant L1. The L1 dominant group

consisted of bilinguals with higher proficiency in their first language, English.

Table 2 shows the mean self-ratings of the respective bilingual groups (L1

dominant and balanced) for their proficiency in Spanish and English. Participants were

asked to rate their reading proficiency, writing proficiency, speaking ability, and speech

comprehension ability for each language on a scale from 1 to 10, with 1 being the lowest

possible proficiency and 10 the highest (see Appendix B for the complete language

history questionnaire). The balanced group reported nearly equal proficiency in both

languages, with mean self-ratings of 9.34 for Spanish and 9.46 for English. The L1

dominant group, on the other hand, had self-ratings of 7.03 for Spanish and 9.70 for

English, a disparity that indicated higher proficiency in English.

Apparatus

The experiment was programmed with E-studio and run on a Dell PC using the

program E-prime. Stimuli and instructions were presented on the monitor and participants

24

clicked the space-bar to begin each trial. Participants spoke into a microphone connected

to a button-box with a voice key that recorded their reaction times, or the time it took

them to begin to articulate the name of the picture. Responses were recorded for later

transcription and coding with respect to accuracy.

Procedure

Language History Questionnaire. Participants completed a questionnaire regarding their

language experience, foreign language education, and time spent abroad (see Appendix B

for the complete questionnaire). They self-rated their reading and writing proficiencies,

speaking ability, and speech comprehension ability for each language they spoke or had

studied. Self-ratings were on a scale from 1-10, with 1 being not proficient and 10 being

very proficient. Participants were asked to identify which language they spoke with their

parents; language(s) in which they received their education; and countries where they

were born, attended elementary school, and attended high school. They were also asked

to describe study abroad experience; list number of semesters of foreign language study;

and state if they were an international student or a foreign language graduate student,

major, or minor.

Picture Naming. Participants were asked to name pictures in Spanish and English, with

background color functioning as a response language cue. The first trial consisted of a

fixation point presented for 500 ms., followed by a blank screen for 30 ms., followed by a

picture stimulus. Subsequent trials consisted of a blank screen for 30 ms., followed by a

picture stimulus. Participants were instructed to name pictures as quickly and accurately

25

as possible, and to say "no" if they didn't know the name of a picture. Pictures remained

onscreen indefinitely until a spoken response was produced, at which point pictures

disappeared and a second fixation point appeared. Participants controlled the pace at

which pictures appeared by pressing the space bar to start the next trial.

Stimuli

Ninety-two black and white line drawings served as stimuli, with 80 presented in

critical trials and 12 used for practice. The 80 test stimuli were selected on the basis of

semantic relationship, with ten items corresponding to each of eight semantic categories

(see Appendix A for a complete list of stimuli). Half the categories were classified as

living (farm animals, body parts, fruits and vegetables, and things that fly), and half as

non-living (clothing, kitchen items, furniture, and tools). Items were controlled for

cognate status (non-cognates of Spanish and English).

Design

Stimuli were presented in alternating semantically blocked and mixed series.

Critical trials comprised two blocks of 40 trials each, with a short pause after trial 40.

Semantic condition was alternated every 10 trials, so that each block of 40 consisted of

two blocked series and two mixed series. Background color (yellow or blue) alternated

every two trials.

Twelve practice trials preceded the experimental trials. Practice consisted of six

semantically blocked trials followed by six mixed trials. Practice items and semantic

categories were different from those presented in critical trials.

26

Participants saw each stimulus only once, in either the blocked or mixed

condition, with half the stimuli appearing in each condition. The condition that items

appeared in was counterbalanced across subjects, with items from the respective

categories presented in a blocked context for half of participants, and a mixed context for

the other half. Starting condition was counter-balanced across subjects. Distribution of

living and non-living stimuli was counterbalanced so that each subject saw 2 living and 2

non-living categories in each condition.

Twelve counterbalanced versions of the design were produced, with the order of

presentation of stimuli within the conditions randomized. Twenty-four subversions were

generated by counterbalancing starting language across subjects, with respective

background colors corresponding to Spanish for half of participants and English for the

other half.

Data analysis

Three main variables were considered in the statistical analyses: response

language (L1 or L2), semantic condition (blocked or mixed), and trial type (switch or

non-switch). Analyses of variance (ANOVA’s) were conducted on mean naming

latencies and percent accuracy for each of the 8 trial types (see Figure 10 for a description

of the respective trial types).

Erroneous trials were excluded from the reaction time (RT) analysis. RT’s of less

than 300 ms. or more than 3000 ms. were excluded as outliers, along with RT’s beyond

2.5 standard deviations from the mean for each respective language (L1 and L2). First

trials from each group of 40 were excluded as warm-up trials. Additional trials were

27

excluded due to technical error (microphone failure, participant pressing keys during

trial, background noises triggering microphone).

Due to the wide variability in participants' responses, two separate RT analyses

were conducted: a liberal analysis and a conservative analysis. For the conservative

analysis, responses were counted as correct only when the participant produced the exact

term used by experimenters to identify each picture. The liberal analysis included

additional responses (synonyms, semantically related items, and items that physically

resembled the target item) that could be considered plausible names for the picture. After

determining that there was no significant disparity in results between the liberal and

conservative analyses, we elected to use the liberal criteria for the final analyses because

it allowed more data points to be included.

Calculating Switch Costs and Semantic Interference

To examine how language switching and semantic context affect picture naming

performance, it is necessary to understand the calculations that allow us to quantify these

effects. Figure 11 explains how we calculate switch costs, or the difference in mean

naming latencies between switch and non-switch trials. Switch costs reflect how fast the

speech production system is able to adapt to the change of language. In other words, how

much slower is speech production when the speaker must switch languages from one

utterance to the next?

28

29

Figure 12 summarizes the calculations used to determine magnitude of semantic

interference. By calculating the difference between mean naming latencies in the blocked

and mixed conditions, we can examine whether naming performance is affected by

semantic context. Significantly slower naming times in the blocked condition are an

indication of semantic interference.

30

Predictions

In the first place, we expect to replicate the robust effect of semantic interference

reported by Kroll and Stewart (1994) and Damian et. al (2001). We also expect to

replicate the switch cost asymmetry reported by Meuter and Allport (1999), with greater

switch costs into the L1 than into the L2. Finally, we expect to replicate the differential

pattern of switch cost asymmetry between balanced and L1-dominant bilinguals reported

by Costa and Santesteban (2004). We hypothesize that results will indicate a differential

selection mechanism for balanced and non-balanced bilinguals, specifically that

proficient bilingual speakers have achieved a level of control in using their two languages

that no longer requires active inhibition of the more dominant language.

31

Results

The ANOVA on the mean RTs produced a main effect of response language, with

longer overall naming latencies in the L2 than in the L1, F (1, 40) = 17.89, p < .001.

There was also a significant interaction between response language and type of bilingual,

F (1, 40) = 12.94, p < .001. The L1 dominant group had faster overall naming latencies in

both languages than the balanced group, and while both groups had faster RT’s in the L2,

the L1 dominant group had a greater differential between mean RT’s in the two

languages.

Effect of Semantic Blocking

The effect of semantic blocking for each response language and for each bilingual

group is shown in Figure 13. The ANOVA on the mean RTs failed to produce a main

effect of semantic blocking, F (1, 40) < 1. Semantic condition showed no significant

interaction with response language, F (1, 40) < 1, or with type of bilingual F (1, 40) < 1.

32

The three-way interaction between type of bilingual, response language, and semantic

condition also failed to reach significance, F (1, 40) < 1. Thus, the results failed to

replicate the robust effects of semantic interference reported by Kroll and Stewart (1994)

and Damian et. al (2001). Before we discuss possible explanations for this outcome, let’s

examine the effect of language switching from our combined paradigm.

Effect of Language Switching

The effect of language switching for each response language and for each

bilingual group is shown in Figure 14. The ANOVA on the mean RTs produced a main

effect of switching, with longer naming latencies for switch than for no-switch trials, F

(1, 40) = 8.26, p < .001. Analysis of errors also produced a switching effect, with higher

accuracy for non-switch trials, F (1, 40) = 3.18, p < .001. However, the significant

observation of a language switch cost did not interact with type of bilingual, F (1, 40) <

1, nor with response language, F (1, 40) < 1. The three-way interaction, between type of

33

bilingual, response language, and trial type also failed to reach significance, F (1, 40) < 1.

These results failed to replicate the switch cost asymmetry reported by Meuter

and Allport (1999), with greater switch costs into the L1 than into the L2, and also failed

to replicate the differential pattern of switch cost asymmetry between balanced and L1-

dominant bilinguals reported by Costa and Santesteban (2004). That is, neither group of

bilinguals produced a switch cost asymmetry, although both groups suffered significant

switch costs following a change of language.

Interaction between Switching and Blocking Effects

The ANOVA on the mean RTs produced a significant interaction between

semantic condition and trial type, F (1, 40) = 3.05, p < .001. The three-way interaction,

between, response language, semantic condition and trial type was also significant, F (1,

40) = 1.26, p < .001. Before we take a closer look at the nature of this interaction, we will

reanalyze the effect of semantic interference by including data from just the non-switch

trials, or those in which the speaker did not experience a switch cost. The effect of

semantic blocking for each response language and for each bilingual group, including

data from the non-switch trials only, is shown in Figure 16.

34

In our analysis of non-switch trials, neither bilingual group experienced

semantic interference in L1. The balanced group was slightly faster overall than the L1

dominant group, but this difference was not significant. In L2, the situation was markedly

different. In the first place, the L1 dominant group had slower overall naming latencies

than the balanced group. Secondly, the L1 dominant group suffered semantic interference

in the L2 blocked condition, while the balanced group did not. Paired t-Tests produced a

significant effect of semantic interference due to blocking in L2 non-switch trials for the

L1 dominant group, t22 = -1.56, but no significant interference effect for the balanced

group, t18 < 1.

Next, we will examine how trial type (switch or non-switch) interacts with

semantic condition (blocked or mixed). Figure 15 shows the mean magnitude of switch

costs in the respective conditions (see Figure 11 for a review of how to calculate switch

costs).

35

For picture naming in L1, the L1 dominant group produced larger switch costs

than the balanced group in both the blocked and mixed conditions. In the L2 mixed

condition, the L1 dominant group experienced larger switch costs than the balanced

group. In the L2 blocked condition, however, the L1 dominant group experienced no

switch costs whatsoever. Considering that the L1 dominant group produced a semantic

blocking effect in L2, it is possible that the interaction between semantic blocking and

language switching caused switch costs to be eliminated in the L2 blocked condition.

In both languages, the balanced group had greater switch costs in the mixed than

in the blocked condition. The balanced speakers also produced a switch cost asymmetry

in the mixed condition, with significantly higher switch costs in L2. In the blocked

condition, however, the balanced group did not produce a switch cost asymmetry, with

nearly equal switch costs for L1 and L2. Surprisingly, the L1 dominant group

experienced no switch cost asymmetry in the mixed condition, the only condition in

which it produced switch costs.

36

Discussion

It is noteworthy that both the L1 dominant and balanced groups experienced

greater switch costs in the mixed than in the blocked condition. One possible explanation

is that the speech production system can only be slowed down to a certain point. In other

words, semantic interference was already slowing production in the blocked condition,

which gave the system time to catch up so that the effects of language switching, which

also had the potential to slow down the system, were less pronounced. Another way to

look at it is that since naming latencies were generally faster in the mixed condition, the

efficiency enjoyed by the system in this condition had the potential to suffer more

damage than in the blocked condition, when the system was already bogged down by

semantic interference. Let’s imagine language switching as a lawsuit and switch costs as

the monetary damages we hope to collect. We can do greater damage and expect to

collect more money by suing Bill Gates, who has a lot to lose, than by suing someone

who is broke and already heavily in debt.

Another striking outcome was the fact in the analysis of non-switch trials,

neither bilingual group suffered semantic interference for picture naming in L1, but both

experienced it in L2. This outcome is especially puzzling when viewed in the context of

previous studies that employed semantic blocking paradigms (Damian et. al, 2001; Kroll

& Stewart, 1994). To investigate how the semantic blocking paradigm affects naming

performance in participants who only know one language, we decided to conduct the

experiment with a group of monolingual, English-speaking participants.

37

Chapter 3: Experiment 2

Method

Method, procedure, stimuli, and design were identical to those used in Experiment

1, except that participants named pictures only in English. Background color still

alternated between yellow and blue, but no special instructions were given regarding

these colors.

Participants

Twelve undergraduate students were paid for their participation. All participants

were monolingual in English. Four had studied a foreign language at Penn State for no

more than four semesters, while eight had not studied a language at the university level.

None had lived or studied abroad in a non-English-speaking country or were planning to

major or minor in a foreign language. For those who had studied a foreign language in

high school or college, mean self-rating of L2 proficiency was 2.11 (1= not proficient,

10= very proficient), indicating that all participants were functionally monolingual and

had minimal foreign language proficiency. Mean age was 19.6 years, with 18 being the

youngest and 24 the oldest. Five were female and seven were male. All had normal

hearing and normal or corrected vision.

38

Results

Figure 17 shows mean naming latencies in the semantically blocked and mixed

conditions for the monolingual speakers. The ANOVA on the mean RTs failed to

produce a main effect of semantic blocking, with similar naming latencies in the blocked

and mixed conditions, F (1, 40) < 1. Analysis of accuracy, however, did produce a

significant main effect of semantic blocking, with higher percent accuracy in the

semantically mixed condition than in the blocked condition, F (1, 40) = 8.84, p < .001.

To determine whether language switching, and not the effect of switching background

color, was the true cause of switch costs in the bilingual experiment, we examined

whether color switching had any effect on mean naming latencies. The ANOVA

produced no effect of trial type, with similar naming latencies in color switch and non-

color switch trials F (1, 40) < 1. Figure 18 shows mean naming latencies for each

semantic condition and for each trial type, demonstrating that color switching does not

influence naming times.

39

Discussion

Considering that the monolingual speakers did produce a semantic effect of

accuracy, though not in the RT analysis, it is evident that some degree of interference due

to blocking did take place. The failure to produce a semantic interference in the RT

analysis can possibly be explained in terms of the nature of our experimental paradigm.

While both Kroll and Stewart (1994) and Damian et. al (2001) required participants to

name pictures in a monolingual (L1) context, as in the present experiment, there were

fundamental differences in the nature of the paradigms employed. In the first place, the

semantically blocked and mixed series were longer in the previous studies. This could be

problematic for producing semantic interference, since the magnitude of semantic

interference increases as the speaker moves through a progression of related items (see

Belke, Meyer, and Damian, 2005).

40

Figure 19 illustrates how compounding activation levels of semantically related

items make picture naming more and more difficult as the speaker progresses through a

blocked series. The higher the activation levels of different animal words, the more

difficult it is for the speaker to produce the correct response. The semantically blocked

series used in the present study consisted of only 10 trials each; it is possible that this was

simply not long enough for semantic interference from blocking to take effect.

Another factor that may have influenced results is the repetition of the same

picture within a given series. Repetition of stimuli within the lists further increases the

magnitude of semantic interference, as it progressively compounds the activation levels

of the different stimuli. Belke et. al (2005) reported that semantic interference only arose

after all the items in a blocked series had already been viewed and named once. The

paradigm used by Damian et. al (2001) repeated the same item five times within a given

block, though never consecutively, while in the present study, each picture appears just

once in each condition (once in a blocked series and once in a mixed series).

41

In summary, the fact that the monolingual speakers had lower accuracy in the

blocked condition indicates that some degree of semantic interference due to blocking did

take place. Although the RT analysis produced no significant interference effect, it is

likely that semantic blocking would have affected the mean RT's had our experimental

paradigm involved longer series and repetition of items within a given series.

Chapter IV: General Discussion

In Experiment 1, semantic interference in the blocked condition was observed

only for bilinguals performing non-switch trials in the L2. The effect was more

pronounced in the L1 dominant group and did not reach significance in the balanced

group, although the balanced speakers did have slower naming latencies in the blocked

condition. Both groups showed robust effects of language switching that were reduced in

the context of semantically blocked lists. In Experiment 2, monolingual speakers failed to

produce semantic interference due to blocking in the RT analysis, but did experience

semantic interference in the analysis of accuracy, with higher accuracy in the mixed than

in the blocked condition.

The most surprising result was the lack of switch costs experienced by the L1

dominant group in the L2 blocked condition. It is also noteworthy that the balanced

speakers produced a switch cost asymmetry in the mixed condition, with significantly

higher switch costs in L2, yet experienced no switch cost asymmetry in the blocked

condition, with nearly equal switch costs for L1 and L2. Since all other variables were

42

equal except for type of bilingual, it is reasonable to assume that these disparities denote

some fundamental difference in the way balanced bilinguals and non-balanced L2

learners process language.

Let’s return to the central issues currently being debated in the literature. First and

foremost, there is the question of whether lexical selection in bilinguals is language

specific or language non-specific, and if it is non-specific, whether the selection

mechanism entails active inhibition of the non-response language. Results from the

present study may be interpreted in several ways. Costa & Santesteban (2004) used the

differential patterns of switch cost asymmetry observed in balanced and L1 dominant

bilinguals to argue that the nature of lexical selection is different for balanced and non-

balanced bilinguals. According to their hypothesis, lexical selection is language specific

for non-balanced L2 learners, who must exercise differential levels of inhibition in their

L1 and L2. Balanced bilinguals, however, reach a “threshold” at which selection becomes

language non-specific, which could explain the absence of switch cost asymmetries for

the balanced bilinguals.

In the present study, it is clear that the nature of selection is somehow different for

balanced and for L1-dominant bilinguals. However, results from Experiment 1 failed to

replicate the differential pattern of switch cost asymmetry between balanced and L1-

dominant bilinguals reported by Costa and Santesteban (2004). That is, neither group of

bilinguals produced a switch cost asymmetry, although both groups suffered significant

switch costs. Let’s review Figures 14 and nine, which show reported switch cost

asymmetries (or lack thereof) for the two types of bilinguals from the present study and

from Costa and Santesteban (2004), respectively.

43

The failure to replicate the pattern of asymmetry may be explained in terms of the nature

of the experimental paradigms employed. First and foremost, the present study entailed

semantic blocking as well as language switching, while the paradigm employed by Costa

and Santesteban did not include a semantic interference component. The significant

interaction between trial type and semantic condition observed in the present study,

which showed that switch costs were reduced in the context of semantically blocked lists,

may have altered the pattern of switch cost asymmetry that would have been observed for

switching only. Evidence to support this hypothesis includes the fact that the balanced

speakers showed asymmetry in the mixed condition, but not in the blocked condition.

Another aspect of our experimental paradigm that may have affected results is the

absence of repetition of the same picture. The paradigm employed by Costa and

44

Santesteban (2004) entailed multiple repetitions of the same item within each list. Figure

19 illustrates how the magnitude of semantic interference increases as a participant

progresses through a semantically blocked series, as activation levels of semantically

related items become higher and increase the degree of competition experienced.

Repetition also serves to progressively increase the activation levels of the different

stimuli. It follows that the magnitude of switch costs would progressively increase due to

multiple repetitions of the same item within each series. Thus, this fundamental

difference between the two paradigms may have prevented the present study from

replicating the pattern of asymmetry reported by Costa and Santesteban.

Results from Experiment 2, in which the monolingual speakers failed to produce

semantic interference due to blocking in the RT analysis, but did experience semantic

interference in the analysis of accuracy, with higher accuracy in the mixed than in the

blocked condition, may shed further light on results from Experiment 1. Taking into

consideration this outcome and results from Belke et. al (2005), which reported that

semantic interference only arose after all the items in a blocked series had already been

viewed and named once, this indicates that semantic interference did occur to some

degree, but did not have time to fully take effect during the course of each series.

Considering the fundamental differences in the paradigms employed in the

present study and in Costa and Santesteban (2004), it is difficult to compare results from

the two experiments. However, results from both experiments indicate that there is a

fundamental difference in the way that balanced bilinguals and non-balanced L2 learners

process language. The fact that the balanced speakers produced a switch cost asymmetry

in the mixed condition, with significantly higher switch costs in L2, yet experienced no

45

switch cost asymmetry in the blocked condition, with nearly equal switch costs for L1

and L2, is consistent with the hypothesis that balanced bilinguals have developed a

selection mechanism that no longer requires active inhibition of the non-dominant

language. The switch cost asymmetry was eliminated in the blocked condition due to the

interaction with semantic blocking, but in the mixed condition, no differential levels of

inhibition were required to suppress the non-response language. Thus, we can tentatively

conclude that our results support Costa and Santesteban’s hypothesis that lexical

selection is language non-specific and involves active inhibition until bilinguals reach a

certain threshold of ability, after which it becomes language specific and requires no

inhibition. Future research is required that uses a paradigm with multiple repetitions and

longer series in order to allow for semantic interference to fully take effect.

46

Chapter 5: References

Belke, E., Meyer, A. S., & Damian, M.F. (2005). Refractory effects in picture naming as

assessed in a semantic blocking paradigm. The Quarterly Journal of Experimental

Psychology, 58A, 667-692.

Bloem, I., & La Heij, W. (2003). Semantic facilitation and semantic interference in word

translation: Implications for models of lexical access in language production.

Journal of Memory and Language, 48, 468-488.

Bloem, I., Van den Boogaard, S., & La Heij, W. (2004). Semantic facilitation and

semantic interference in language production: Further evidence for the conceptual

selection model of lexical access. Journal of Memory and Language, 51, 307-323.

Calomé, A (2001). Lexical activation in bilinguals' speech production: Language-specific

or language-independent? Journal of Memory and language, 45, 721-736.

Caramazza, A. (1997). How many levels of processing are there in lexical access?

Cognitive Neuropsychology, 14, 177-208.

Costa, A. (2005). Lexical access in bilingual production. In J. F. Kroll & A. M. B. De

Groot (Eds.), The Handbook of Bilingualism: Psycholinguistic Approaches (pp.

308-325). New York: Oxford University Press.

Costa, A. & Santesteban, M. (2004). Lexical access in bilingual speech production:

Evidence from language switching in highly proficient bilinguals and L2 learners.

Journal of Memory and Language, 50, 491-511.

Damian, M. F., Vigliocco, G., & Levelt, W. J. M. (2001). Effects of semantic context in

the naming of pictures and words. Cognition, 81, B77-B86.

47

Finkbeiner, M., Gollan, T. & Caramazza, A. (in press). Bilingual lexical access: What’s

the (hard) problem? Bilingualism: Language and Cognition.

Francis, W. S. (2005). Bilingual semantic and conceptual representation. In J. F. Kroll

& A. M. B. De Groot (Eds.), The Handbook of Bilingualism: Psycholinguistic

Approaches (pp. 251-267). New York: Oxford University Press.

Green, D. W. (1998). Mental control of the bilingual lexico-semantic system.

Bilingualism: Language and Cognition, 1, 67-81.

Hammink J. E. (2000). A Comparison of the Code Switching Behavior and

Knowledge of Adults and Children. Unpublished doctoral dissertation,

University of Texas at El Paso. Retrieved March 31, 2006, from

http://hamminkj.cafeprogressive.com/CS_paper.htm.

Hermans, D., Bongaerts, T., De Bot, K., & Schreuder, R. (1998). Producing words in a

foreign language: Can speakers prevent interference from their first language?

Bilingualism: Language and Cognition, 1, 213-229.

Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture

naming: Evidence for asymmetric connections between bilingual memory

representations. Journal of Memory and Language, 33, 149-174.

La Heij, W. (2005). Selection processes in monolingual and bilingual lexical access. In

J. F. Kroll & A. M. B. De Groot (Eds.), The Handbook of Bilingualism:

Psycholinguistic Approaches (pp. 289-307). New York: Oxford University Press.

Meuter, R. F. I. (2005). Language selection in bilinguals: Mechanisms and

processes. In J. F. Kroll & A. M. B. De Groot (Eds.), The Handbook of

48

Bilingualism: Psycholinguistic Approaches (pp. 349-370). New York:

Oxford University Press.

Meuter, R. F. I., & Allport, A. (1999). Bilingual language switching in naming:

Asymmetrical costs of language selection. Journal of Memory and Language, 40,

25-40.

Poplack, S. 1980. “Sometimes I'll start a sentence in Spanish y termino en español:

Toward a typology of code-switching.” Linguistics, 18. 581-618.

Schriefers, H., Meyer, A. S., & Levelt, W. J. M. (1990). Exploring the time-course of

lexical access in production: Picture-word interference studies. Journal of

Memory and Language, 29, 86-102.

Stroop, J. Ridley. (1935). Studies of interference in serial verbal reactions. Journal of

Experimental Psychology, 18, 643-662.

49

Appendix A: Stimuli from Experiments 1 & 2

Picture/ English

Name

Spanish Name Semantic Category Animacy

hair pelo body parts LIVING

hand mano body parts LIVING

arm brazo body parts LIVING

leg pierna body parts LIVING

hair pelo body parts LIVING

foot pie body parts LIVING

ear oreja body parts LIVING

finger dedo body parts LIVING

eye ojo body parts LIVING

lips labios body parts LIVING

bee abeja things that fly LIVING

fly mosca things that fly LIVING

bird pájaro things that fly LIVING

eagle águila things that fly LIVING

peacock pavo real things that fly LIVING

duck pato things that fly LIVING

penguin pingüino things that fly LIVING

ostrich avestruz things that fly LIVING

butterfly mariposa things that fly LIVING

swan cisne things that fly LIVING

rooster gallo farm animals LIVING

pig cerdo farm animals LIVING

donkey burro farm animals LIVING

rabbit conejo farm animals LIVING

horse caballo farm animals LIVING

chicken gallina farm animals LIVING

cow vaca farm animals LIVING

goat cabra farm animals LIVING

cat gato farm animals LIVING

sheep oveja farm animals LIVING

grapes uvas Fruits and Vegetables LIVING

corn maíz Fruits and Vegetables LIVING

carrot zanahoria Fruits and Vegetables LIVING

watermelon sandía Fruits and Vegetables LIVING

apple manzana Fruits and Vegetables LIVING

Celery apio Fruits and Vegetables LIVING

pumpkin calabaza Fruits and Vegetables LIVING

cherry cereza Fruits and Vegetables LIVING

mushroom seta Fruits and Vegetables LIVING

strawberry fresa Fruits and Vegetables LIVING

tie corbata Clothing NON-LIVING

hat sombrero Clothing NON-LIVING

glove guante Clothing NON-LIVING

shoe zapato Clothing NON-LIVING

coat abrigo Clothing NON-LIVING

sock calcetín Clothing NON-LIVING

skirt falda Clothing NON-LIVING

pants pantalones Clothing NON-LIVING

belt cinturón Clothing NON-LIVING

shirt camisa Clothing NON-LIVING

50

iron plancha Tools NON-LIVING

pliers alicates Tools NON-LIVING

axe hacha Tools NON-LIVING

screwdriver destornillador Tools NON-LIVING

chain cadena Tools NON-LIVING

scissors tijeras Tools NON-LIVING

saw sierra Tools NON-LIVING

nail clavo Tools NON-LIVING

ruler regla Tools NON-LIVING

hammer martillo Tools NON-LIVING

pot olla Kitchen NON-LIVING

bowl taza Kitchen NON-LIVING

pitcher jarra Kitchen NON-LIVING

spoon cuchara Kitchen NON-LIVING

bottle botella Kitchen NON-LIVING

glass vaso Kitchen NON-LIVING

fork tenedor Kitchen NON-LIVING

knife cuchillo Kitchen NON-LIVING

rolling pin rodillo Kitchen NON-LIVING

clothespin pinza Kitchen NON-LIVING

stool taburete Furniture NON-LIVING

rocking chair mecedora Furniture NON-LIVING

chair silla Furniture NON-LIVING

couch sofá Furniture NON-LIVING

bed cama Furniture NON-LIVING

desk escritorio Furniture NON-LIVING

lamp lámpara Furniture NON-LIVING

dresser cómoda Furniture NON-LIVING

table mesa Furniture NON-LIVING

stove horno Furniture NON-LIVING

51

Appendix B: Language History Questionnaire

Subject #:________________________________ Date:_________________

Language History Questionnaire

This questionnaire is designed to give us a better understanding of your experience with

languages. Please answer these questions as thoroughly and accurately as possible.

1. Gender: Female Male 2. Age: ________ years

3. Right-handed Left-handed

4. Do you have any known visual or hearing problems (corrected or uncorrected)?

No Yes [Please explain] ______________________________

5. Please name the countries…

Where you were born: ____________________________

Where you attended elementary/middle school:_____________________

Where you attended high school:_______________________

6. Which language(s) do you speak at home with your parents? ___________________

Please self-rate your proficiency in the following areas for your first language (the one

you use at home with your parents):

Reading proficiency (1= not literate, 10= very literate): __________________

Writing proficiency (1= not literate, 10= very literate): __________________

Speaking ability (1= not fluent, 10= very fluent): __________________

Speech comprehension ability (1= unable to understand conversation, 10= perfectly able

to understand): __________________

7. Now, please rate your proficiency in these areas for any second language(s) that you

speak/have studied.

Language:______________________________________

Reading proficiency (1= not literate, 10= very literate): __________________

Writing proficiency (1= not literate, 10= very literate): __________________

Speaking ability (1= not fluent, 10= very fluent): __________________

Speech comprehension ability (1= unable to understand conversation, 10= perfectly able

to understand): __________________

Language:______________________________________

Reading proficiency (1= not literate, 10= very literate): __________________

Writing proficiency (1= not literate, 10= very literate): __________________

Speaking ability (1= not fluent, 10= very fluent): __________________

52

Speech comprehension ability (1= unable to understand conversation, 10= perfectly able

to understand): __________________

8. Besides foreign language classes, have you attended school or university in a language

other than the one you use at home? Please explain:

________________________________________________________________________

________________________________________________________________________

________________________________________________________________________

________________________________________________________________________

9. Have you taken foreign language classes? If so, please list language(s) studied and # of

semesters.

Elementary/middle/high school: ____________________________________________

College: _______________________________________________________________

10. What grades do you normally earn in foreign language classes?

Mostly A’s

Mostly A’s and B’s

Mostly B’s

Mostly B’s and C’s

Mostly C’s

11. Please check all of the following that apply to you:

Taking a second language for a requirement, and interested in being a major or minor.

Taking a second language, but not interested in being a major or minor.

A second language minor.

A second language major.

A second language graduate student.

An international student.

12. Have you studied or lived abroad? If so, please list:

Country Approx. dates Length of Stay Language

13. Please list any other information regarding your language experience.

53

Appendix C: Academic Vita of Stephanie R. Dodson

[email protected]

48 Hampton Place

Walkersville, MD 21793 USA

(301) 514-7187

EDUCATION

The Pennsylvania State University, University Park, PA Fall 2002 – Spring 2006

• B.A. in Spanish

• Minor in Linguistics

• Schreyer Honors College

• Cumulative GPA 3.92 / 4.00

• Expected Graduation Date: May 2006

GRANTS AND AWARDS

• Fulbright English Teaching

Assistantship to South Korea July 2006 – July 2007

• Phi Beta Kappa Society Spring 2006 - Present

• Spanish Department Certificate

of Excellence Award Spring 2006

• John W. White Scholarship

for Excellence in Spanish Spring 2005

• Schreyer Honors College International Thesis

Research Grant Summer 2004

• Student Enrichment Grant,

College of Liberal Arts Summer 2004

• Academic Excellence Scholarship Fall 2002 - Spring 2006

• Dean’s List Fall 2002 - Present

• National Honors Society Fall 2001 - Present

FOREIGN LANGUAGE PROFICIENCY

• Spanish - Advanced conversational, grammatical, and written proficiency

• German, French - Elementary written proficiency

STUDY ABROAD EXPERIENCE

Study Abroad at El Colegio San Pedro Nolasco, Concepción, Chile July 2001 - January 2002

• Developed Spanish skills

High School trip to Yokohama, Japan June 2000

• Visited a Japanese high school for one week

INTERNATIONAL RESEARCH EXPERIENCE

Universitat de Barcelona, Spain May - August 2004

• Designed an experiment in collaboration with Spanish researchers for project

investigating language processing in bilinguals

54

• Recruited research participants from the Barcelona area and conducted an

experiment

• Analyzed data using Excel and discussed results in the context of background

literature

• Attended seminars and lectures

UNDERGRADUATE LABORATORY RESEARCH

Language and Cognition Research Lab, Penn State University January 2003 - May 2006

• Conducted honors thesis research under the supervision of Dr. Judith Kroll

• Designed an experiment on language processing in bilinguals

• Performed data collection and analysis, recruited participants and trained new

research assistants

• Utilized software including Excel, StatView, E-Prime, DmDx, Word and

Powerpoint

MULTILINGUAL VOLUNTEER EXPERIENCE

Mission of Mercy Summer 2002

• Served as bilingual receptionist and community interpreter for Spanish-speaking

patients

Frederick Memorial Hospital Summer 2002

• Served as intermediary between Spanish-speaking patients and medical

personnel

YOUTH MENTORING EXPERIENCE

Big Brother/ Big Sister Spring 2003 - Fall 2004

• Interacted with a local youth during weekly mentoring sessions

Frederick County Public Schools June 2002

• Helped teach a three-week ESL course for middle school students

WORK EXPERIENCE

Language and Cognition Research Lab

• Conducted research and trained research assistants June – August 2005

• Assisted graduate students with research projects May - August 2003

MBNA America May – August 2003

• Interacted with the public as a telesales associate

Poffenbarger Veterinary Clinic June 1998 - June 2002

• As a veterinary assistant, assisted doctors, performed secretarial work, answered

phones, filled prescriptions, cleaned facilities and cared for animals

CLUBS AND ACTIVITIES

• Penn State Spanish Club Fall 2003 - Spring 2006

• Penn State Linguistics Club Spring 2005 - Spring 2006

• Penn State German Club Fall 2002 - Spring 2003

• Private voice lessons January 1996 - July 2001

STUDENT LEADERSHIP POSITIONS

• Spanish Club Reporter Fall 2004 - Spring 2005